196906 196906

User Manual: 196906

Open the PDF directly: View PDF PDF.
Page Count: 56

Open PDF In BrowserView PDF
~O 10




IS 13~~VW S 011

~,,()I? F



F7 ?NY 3


June 1,1969
Vol. 18, No.6


Computers and Medicine

and automation

COBOL spoken here. Just one of the PDP-10 languages
for on-line program development. With COBOL,
time-sharing users can develop, edit, and debug data
management programs on-line. Run them too. Equally
well under time-sharing and batch processing.
PDP-10 speaks more than COBOL. FORTRAN-IV,
MACRO-10, BASIC, and AID. And all these programming
languages are re-entrant. That is, one copy is shared
by many users to make efficient use of core. Protected,
too, by hardware to prevent accidental destruction.
PDP-10 COBOL has another special feature. The
programs that a user develops for his particular
application are also re-entrant. Like the compiler,
they can be shared by other users.
But that isn't all. The PDP-10 serves up to 63 time-sharing
users with all these languages. Simultaneously. And
handles program development, batch processing,
and real-time operations. All at the same time.
More voice in software. More choice in hardware.
Low cost. The PDP-10 speaks your language.




Digital Equipment
Maynard, Mass.
Designate No. 20 on Reader Service Card





WHAT IS c~~rs~ ?

COMPSO will be the first Regional Computer Software &
Peripheral Show to be presented in the three major business
and industry centers in the United States. COMPSO East,
Midwest and West will provide computer users with the
latest information on software, peripherals, supplies and

c~~rs~ BE HELD?

SHOWS & CONFERENCES will be held in:



.'. ; I;



i·, ,) .:' ~~: ;; J\

• .'"

~ ~.1

• New York: New York Hilton, January 19-21,1970
• Chicago: Pick-Congress, February 17-19, 1970
• Los Angeles: Anaheim Convention Center, April 7-9, 1970

J\ 1~ rJ

COMPSO provides you with an opportunity to exhibit your
products and services in one or more of the three largest
and most active markets for computers and related services
in the country. These cities are world centers for communication, manufacturing, marketing, transportation, international
trade and finance; headquarters for almost evety one of the
500 largest corporations in the country.

~'i" ~. '~ : z

SHOW & CONFERENCE will be attended by thousands from
business, industry, finance, education, communications and
government. These prime budgetary and buying executives,
department heads, educators, and others have a growing and
vital interest in software and peripheral use.

Nare, peripherals, services or supplies we urge you
~ time by pre-registering. Write today.

San Jose

Public Library

for further information, contact:

t, New York, N.Y., 10018

Careful usage of books is expected and any soiling,
damage or loss is to be paid for by the borrower.

on Reader Service Card



Letters To The Editor
Vol. 18, No. 6 -

Computers and Education
Your March, 1969 issue which dealt
with "Computers in Education" was interesting and well done. However, I find
it incredible that you could devote almost an entire issue to this topic and not
include any of the work being directed
by public school systems. Our own
project in Michigan (INDICOM \Vaterford Township School District)
and the Philadelphia Project (GROW)
have literally tens of thousands of hours
of pupil experience with Computer Assisted Instruction. In addition, many
hours of curriculum materials have
been planned and prepared.
It would seem that there is far too
great a tendency to think of the public
school environment as it was and not as
it is. I do not see any possibility of
Computer Assisted Instruction being
imposed from the outside. I also believe
that the experience base from which to
devise the strategy and to make the instructional decisions necessary is most
prevalent within public school systems.
This is not intended to minimize the
value or the necessity of university and
other involvement in improving public
school curriculum. However, I believe
that it will be done much more effectively and efficiently if it is done within
a public school setting.
I do not believe the analogy used by
one of your authors in comparing the
possible growth of CAl with that of
automobiles and color television is appropriate. In the case of CAl it will not
be the user (student) that promotes it
but rather the professional. For this
reason it is imperative that classroom
teachers be totally involved in the development of Computer Assisted Instruction.
As a personal aside, it often becomes
difficult as Director of' an operating
project to have to reconcile much of
what is being written about CAl with
reality. One of my co-workers once
said, "I wish the people who write about
CAl would find out more about it".
This does not apply to the authors in
the above mentioned issue, but it might
well apply to the field generally.
Thank you and keep up the good
Director, INDICOM Project
Waterford Township School District
1325 Crescent Lake Rd.
Pontiac, Mich. 48054
Ed. Note - Thank you very much for
your comments. We would be most interested in having you prepare an arti4

cle for us on computers in public school
systems. Would you be able to write
such an article and submit it to us for
possible publication?
We were very favorably impressed
with the article on Computers and Education on page 16 of the March issue of
Computers and Automation. We would
like to make copies of this fine article
available to several other members within our organization who have an interest
in the use of the computer in educational institutions.
I, therefore, would like to request permission to reprint a limited number of
copies of this article. Naturally, if we
were granted this permission, we would
credit Computers and Automation as
being the copyrighted source of the article.
Thank you for your consideration.
Industry Marketing - Education
Honeywell EDP
60 Walnut St.
Wellesley Hills, Alass. 02181
Ed. Note - Permission was granted to
reprint a limited number of copies.

June 1, 1969


Edmund C. Berkeley

Associate Editor

Sharry Langdale

Assistant Editors

Moses M. Berlin
Linda Ladd Lovett
Neil D. Macdonald

Software Editor

Stewart B. Nelson

Market Research Director

I. Prakash

Advertising Director
Art Directors

Bernard Lane
Ray W. Hass
Daniel T. Langdale

Contributing Editors

John Bennett
Andrew D. Booth
John W. Carr III
Ned Chapin
Alston S. Householder
Peter Kugel
Leslie Mezei
Rod E. Packer
Ted Schoeters

AdtJisory Committee

T. E. Cheatham, Jr.
James J. Cryan
Richard W. Hamming
Alston S. Householder
Victor Paschkis

Fulfillment Manager

William J. McMillan

Advertising Representatives

Teaching Machine Language
I read with interest your editorial,
"Machine Language, and Learning It",
in your February issue.
Because my company is concerned
with on-line process control, I would be
most interested in further information
regarding the book you mentioned in
your editorial, The Elements of Digital
Computer Programming. Could you
please tell me the name and address
and price of this book?
We feel that a lack of good instructional material for programming in Australia will result in a number of inquiries
about the book in this country, especially since the material concentrates on
machine language which is of particular
use in I/O process routines.
We find your magazine most informative and interesting, and look forward to
future issues.
Applications Engineer
Hawker De Havilland Australia Pt)'. Ltd.
P.O. Box 78
Lidcombe, N.S.W., Australia 2141
Ed. Note - The book The Elements of
Digital Computer Programming was
written by Edwin D. Reilly, !r., and
Francis D. F ederighi, and is published
(Please turn to jJage 7)

NEW YORK 10018, Bernard Lane
37 West 39 St., 212-279-7281

CHICAGO 60611, Cole, Mason, and Deming
221 N. LaSalle St., Room 856, 312-641-1253

PASADENA, CALIF. 91105, Douglas C. Lance
562 Bellefontaine St., 213-682-1464

ELSEWHERE, The Publisher
Berkeley Enterprises, Inc.
815 Washington St., 617-332-5453
Newtonville, Mass. 02160

Editorial Offices

Computers and Automation is published 13 times a
year (12 monthly issues plus a midyear directory issue
published June 30) at 815 Washington St., Newtonville
Mass. 02160, by Berkeley Enterprises, Inc. Printed i~
U.S.A. Subscription rates: United States, $18.50 for
1 year, $36.00 for 2 years, including annual di·
rectory issue - $9.50 for 1 year, $18.00 for two
years without annual directory; Canada add 50¢
a year for postage; Foreign, add $3.50 'a year for
postage. Address all U.S. subscription mail to' Berke·
ley Enterprises, Inc., 815 Washington St., New'tonville,
Mass. 02160; addres~ all Eur~pean subscription mail
to: Box 52, 6354 Vltznau, SWitzerland. Second Class
Postage paid at Boston, Mass.
Postmaster: Please send all forms 3579 to Berkeley
Enterprises, Inc.,' 815 Washington St.,. Newtonville,
M~ss. 02160. ([) Copyright, 1969, by Berkeley Enter·
prlses, Inc.
Change of address: If your address changes, please
send, us both your new address and your old address
(as It appears on the magazine address imprint), and
allow three weeks for the change to be made. .

and automation
Vol. 18, No.6, June I, 1969

The magazine of the design, applications, and implications of information processing systems.

Special Feature:
Computers and Medicine

by Robert T. Stelloh
The design, implementation, and operation of a very general and highly machinetransferable programming system for the solution of a specific problem.


by Dr. James A. Boyle
An up-to-date review of the theoretical and practical using a computer as a diagnostic aid to clinicians.


work being done in

by Dr. Herbert A. Haessler
How a tailored medical history, responsive to the patient's problem, can be obtained
through a dialogue between the patient and a computer terminal.


The Large-Scale Education and Involvement of Employees
by Howard M. Runck
How the non-technical education of nursing personnel has contributed to the effectiveness of the computer system - and the willingness of employees to work with it at the Los Angeles County Dept. of Hospitals' Computer Center.


The front cover picture
shows a computer-produced profile of an emotionally disturbed patient
at the Institute of Living,
a psychiatric h 0 s pit a l
in Hartford, Connecticut.
The computerized information system used by
the hospital is described
on page 46 in this issue.

by Edward O. Joslin
How to obtain a realistic description of V/orkload by: (1) obtaining an accurate mix
of representative benchmark programs; and (2) specifying a computer system in terms
of a series of expected workload levels.

Re gular Features



Computers and Medicine, by Edmund C. Berkeley

C&A Worldwide

Report from Great Britain, by Ted Schoeters


Across the Editors Desk - Computing and Data Processing


Advertising Index

Jobs and Careers in Data Processing

Leadership in a Changing Society, by Joseph C. Wilson

Ideas : Spotlight

Computing Must Acquire an Excellence of Professionalism and Responsibility, by Edward E.
David, Jr.

The Relevance of Computer "Thinking" -

As We Go To Press
Calendar of Coming Events


Classified Advertisements


Multi-Access Forum




The Special Interest Committee on Social Implications of Computers of the Association for
Computing Machinery - Discussion, Part 2


"The Misdirection of Defense and the Social Responsibilities of Computer People" Comments, by Mrs. Phyllis Hyde


The Growing Semantics Problem, by Grace J. Kelleher


LeHers to the Editor


New Contracts


New Installations


New Patents
by Raymond R. Skolnick



Will the Plight of Moe the Elk Be the Plight of the American Citizen?, by Congressman
Cornelius E. Gallagher


Problem Corner


Formation of New Health Record Association to be Discussed at June Meeting, by Fred



Who's Who in the Computer Field, 1968-69 -



by Neil Macdonald
by Walter Penney, COP

Proof Goofs
by Neil Macdonald



Computers and Medicine
This is the first issue af Computers and Automation which
has a special cancentratian an the field af "Camputers and
Medicine". This field af applicatian far camputers will
daubtless became ane af the mast impartant. The 1968
Camputer Directary of Computers and Automation reparted
abaut 135 applicatians af camputers in medicine and in
haspitals. This figure is likely to' rise steadily fram year to'
year. Yet as shart a time as ten years agO', few camputer
pea pie wauld have expected this develapment.
Why has the change happened? There are several reasans:
• Medical Knowledge. A very large quantity af medic"l
knawledge naw exists, and much af it is very new.
It is difficult to' apply large quantities af knawledge
thraugh the small memaries of human beings. Sa this
canditian calls far applicatian af camputers to' retrieve infarmatian.

In this issue, Dr. James Bayle in his article an "Autamated
Diagnasis" paints aut same af the prablems af setting up an
infarmatian retrieval system in the specific area af medical
diagnasis. A lack af knawledge abaut the "art" af diagnasis;
the incansistency in the abservatians af twa dactars viewing
the same patient under the same circumstances; the appasitian af patients to' "rabat" diagnasis; and the unclearness and
at times inaccuracy af a dactar's nates abaut his patients,
particularly when they are read by anather persan, - ~re
same af the barriers that need to' be avercame to' develap
practical, useful, infarmatian retrieval systems in this area.
But these prablems are nat cansidered to' be insuperable.
As Dr. Bayle camments: "The day may nat be far aff when
the diagnastic camputer with its wide range of diagnastic
pragrams is merely anather familiar machine in the dactar's
affice ar autpatient clinic."
• Speed. The time to' apply medical knawledge is shart.
When a human being is suffering fram an illness, the
illness will aften nat wait beyand a few days, sametimes a few haurs, sametimes a few minutes, far the
appropriate actian. This canditian calls far the applicatian af camputers in real time.

It was this call that inspired the autO' mati an af the Milwaukee Blaad Bank described in Rabert T. Stellah's article,
"An Autamated Blaad Bank System for the Milwaukee Blaad
Center". This applicatian illustrates the significant rale camputers can play in preserving human life, in this case, because af the increase in speed with which the right kind af
blaad can get to' the right patient at the right time.
• Greater Efficiency from· Larger Systems. Jaining
clever and trained human beings - dactars, nurses,
haspital administratars, etc. - with camputers intO'
gaad systems can lead to' far greater efficiency in the
care of patients and a substantial saving in the cast af
haspital administratian.


These larger systems are mare camplicated, harder to' aperate well, and they call far a wide range af applicatians af
camputers. But success is being achieved in their develapment, as shawn by the system described by Dr. Herbert
Haessler in his cantributian to' this issue, "Recent Developments in Autamating Medical Histary". Here medical histaries af patients are abtained thraugh the patient's use af a
camputer terminal. The results have been very satisfactary,
and patient respanse has been very favarable.
• Fruitful Careers. In the field af camputers and medicine, there is an excellent chance to' apply camputers
in navel ways to' achieve navel and impart ant results
in the field af increasing health and saving life.

Here is a gaad new field nat anly far camputer peaple,
but far medical peaple as well, to' explare, experiment, innavate, and create, in building their awn careers and reputatian with successes. Haward Runck's article, "Camputer
Planning far Haspitals" reparts an the enthusiasm af nursing
persannel participating in the planning af a haspital camputer system in Las Angeles Caunty, and indicates the exciting passibilities in this area.
• Social Importance. Of all the fields af applicatian af
camputers, surely here is ane with a very large degree
af sacial implicatians, sacial value, and sacial impartance.

Recently a camputer prafessianal (a friend af mine) resigned fram a large electranics firm in the Bastan area with
much military business, in arder to' gO' to' wark in the medical field in Michigan, where he cauld devate his knawledge
af camputers to' medical applicatians - to' advancing life
instead af shartening it.
Here is "vating with the feet" to' apply camputers to'
sacially useful ends.

Of caurse, humanity is faced with a papulatian explasian.
But haw much better, mare humane, and less wasteful it is
to' use methads af birth cantral and birth avaidance to' adjust
the number af human beings to' the resaurces af the Earth
far supparting them - instead af applying famine, disease,
and war. Haw much better it is to' use camputers and medicine sa that human life which has been barn can be preserved
and made healthy.
The staff af Computers and Automation is pleased to' be
able to' facus this issue an "Camputers and Medicine".




LETTERS (Continued trom page 4)

by Holden Day Inc., 500 Sansome St.,
San Francisco, Calif. The address of
the MOHAC Users' Group address
(discussed in the editorial) is MUG,
Box 2675, Schenectady, N.Y. 12309.
Social Responsibility

Although my original intention was to
insure that I receive no more than one
issue of your magazine per month, I
would now request that you cancel and
forward a refund in full.
This action has been prompted by
C&A's continued emphasis upon what
Mr. Berkeley considers to be the "Social
Responsibilities of Computer People".
I and all other computer professionals
with whom I am acquainted are perfectly capable of recognizing my social
responsibilities. Fortunately, I subscribe
to far better publications than C&A to
assist me in recognizing and evaluating
these responsibilities, since I consider
C&A's editorial policy for the most part
to be heavily weighted to the left of
center. From now on, I intend to turn
to other publications for my technical
New Floyd Rd.
AIounted Route
Rome, N.Y. 13440
Ed. Note - A refund
sent as you request.


full is being

Computer Training

The article by Swen Larsen, "Computer Training in Private Schools", in
your March, 1968 issue has been utilized
by our school as part of our philosophy.
The copy we have has become somewhat moth-eaten over the months. I
wonder if we could obtain another copy
of this article.
Your publication is a welcome addition each month, and is read by several
people here at the institute. Keep up
the good work!
Director of Faculty
Computer Programming Institute
of Delaware-Kansas City, Inc.
4949 Johnson Drive
Shawnee Mission, Kansas 66205
Ed. Note - Thank you for your kind
letter. We are pleased to send you an
additional copy of Mr. Larsen's article.
Definition of Terms

In their article describing NCAR's
case study of time-sharing vs. instant
batch processing (March, 1969), Adams
and Cohen say that they "tried to avoid
the so-called 'Hawthorne Effect', where

people who are being studied change
their behavior as a result of being studied".
In various studies of problems related
to Computer Assisted Instruction I have
encountered reference to this "Hawthorne Effect" without sufficient explanation. Isn't the effect to which Adams
and Cohen refer more aptly related to
Heisenberg's uncertainty principle? If
so, what is the Hawthorne effect? I
have been led to believe that it is used
to describe situations where students
using CAl exhibit initial improvement
which then wanes as the novelty of CAl
wears off ( also called the "novelty
effect") .
Can you give me a more precise explanation of the differences between
these three terms as related to CAl:
The Hawthorne effect, Heisenberg's uncertainty principle, and the novelty effect?
Thank you.
Univac Educ. Systems Programming
P.O. Box 3525
St. Paul, Minn. 55101
Ed. Note-The Hawthorne effect refers to a study conducted at the H awthorne plant of Western Electric Co. in

the 1930's. This study showed that personal factors (in this case personal attention given to the group of production
workers) had as much effect on quantity
of production as any other factor, such
as lighting, comfortable work spaces, etc.
The results were reported in a classic
book Management and the Worker by
F. f. Roethlisberger and W. f. Dickson,
Harvard University Press, 1939.
Heisenberg's uncertainty principle, also
called Heisenberg's indeterminancy principle, belongs to quantum physics. It is
the postulate of quantum mechanics that
asserts that "in the simultaneous determination of the values of two canonically conjugated variables (such as the
velocity and position of an electron),
the product of the smallest possible uncertainties in their value is of the order
of Planck's constant h" (which is a unit
of action) . To understand such a statement as this it is necessary to know
quantum physics, and since I do not
know that subject, all I can do is quote
you the definition of the principle and
suggest that you look up the subject.
I am sorry I do not have a definition
of the "novelty effect". Maybe you can
find a definition of it in the papers and
articles which use the term, or from the
authors who wrote them.

Payroll Systems go on-line
faster with ALLTAxnr
the software package
available in basic COBOL
for all compilers.
ALL T AX calculates payroll
withholding taxes with one
standard formula and a table of
variables for each state and city.
It elimina tes programming of
individual formulas and substantially reduces program maintenance and memory requirements.
ALLT AX is approved by all
states. It's easy to install, completely tested and documented.
ALLTAX is always up-todate. Automatic program maintenance for existing withholding
taxes and new taxes is available
at minimal cost.
Find ou t why more than 100
companies from coast-to-coast
are using this low-cost package.

W ri te today for full informa tion:

Management Infonnation Service
P.O. Box 252, Stony Point, N.Y. 10980
Please send full details on your
ALL TAX software package.
Name _____________________
Title ____ _
Company _____________________
City _ _ _ _ _ _ _ _ _ _ __
State _______________ Zip _ _ __
L ______________


Management Information Service
Stony Point, N.Y. 10980 • (914) 942-1880
ALLTAX is available only from Management Information Service
and Pro-Data Computer Services.
Designate No. 27 on Reader Service Card



You have to be a little careful when you tell our new film recorder,
FR-80, what to do. Because there are so many things it might do.
If you have an engineering drawing on mag tape, for example,
FR-80 can reproduce it on film to government specifications.
It will summon up eight different line widths, 128 different
kinds of characters including your special symbols, make
them larger or smaller, rotate them, make them italic,
or place them as sub- or super-scripts.
If you have a digitized parts list, FR-80 generates
your own preprogrammed form simultaneously
with the data. Lists can be merged or selectively
sorted as the film recording is made.


3S mm film, which is superior
to any CRT recorder system in
the world. Such sharpness
makes possible extreme enlargements, like E-size drawings from microfilm.

If you have proposal to produce, FR-80 will
set type in book quality, like this ad.
FR-80 will accept tape formatted for any
peripheral and produce alphanumerics
and graphics with equal quality, because
it is a general purpose computer at
heart. What quality? 80 line pairs
per millimeter resolved on the

story. It will give you a better view from the top.
Information International, 12435 W. Olympic Blvd.,
Los Angeles, Calif. 90064, (213) 478-2571; 89 Brighton
Ave., Boston, Mass.02134,(617)787-4700; 7880Coolridge
Drive, Camp Springs, Maryland 20031 (30 1) 449-4248.


Designate No. 18 on Reader Service Card

PFR-3 •.•
... we have further improved our
Programmable Film Reader/Recorder
to make it the most sophisticated image
analyzer available. New software has
been developed for these applications:
Image Restoration - Using digital filtering, the PFR-3 reconstructs images that
have been degraded by object motion,
out of focus conditions, or atmospheric
turbulence. Ideal for such applications
as biomedical X-rays and recognition
of targets.
Iso-Density Mapping - Continuous-tone
photographs are converted to images
with a contour-like effect having discrete
bands shaded within themselves. Complex, amorphous shapes are more easily
identified and studied.
Spatial Filtering - Ill-defined shapes in a
photograph are given clean separation
(high-frequency filtering). Unwanted
separations, such as the raster lines in a
television image, are filled in (lowfrequency filtering).
And PFR-3 is still interpreting and
extracting information from oil well logs,
oceano logical charts, bacterial cultures,
cine-theodolite film, displayed wave
forms, and many other kinds of image all automatically. Write or call us if
you'd like help on your application.
Information International,
12435 West Olympic Blvd, Los Angeles,
California 90064, (213) 478-2571;
89 Brighton Avenue, Boston,
Mass. 02134, (617) 787-4700;
7880 Coolridge Drive, Camp Springs,
Maryland 20031, (301) 449-4248.


Raymond R. Skolnick
Patent Manager
Ford Instrument Co.
Div. of Sperry Rand Corp.
Long Island City, N.Y. 11101

3,428,955 / Shintaro Oshima, Musashino-shi, and Tetsusaburo Kamibayashi,
Kitaadachi-gun, Saitama-ken, Japan /
Kokusai Denshin Denwa Kabushiki
Kaisha (also known as Kokusai Denshin Denwa Co., Ltd.) Chiyodaku,
Tokyo-to, Japan, a joint-stock company of Japan / Woven wire memory

The following is a compilation of
patents pertaining to computers and associated equipment from the "Official
Gazette of the U. S. Patent Office,"
dates of issue as indicated. Each entry
consists of: patent number / inventor(s)
/ assignee / invention. Printed copies
of patents may be obtained from the
u.s. Commissioner of Patents, Washington, D.C. 20231, at a cost of 50 cents

3,428,956 / Andrew H. Bobeck, Chatham, N. J. / Bell Telephone Laboratories, Incorporated, New York, N. Y.,
a corporation of New York / Rotational mode memory circuit having
flux closure paths.

February 18, 1969

3,428,958 / Harry Putterman, Elizabeth,
N. J. / General Precision Systems Inc.,
a corporation of Delaware / Non-destructive read-out memory and constant current driver.

3,428,951 / Edward Lindell, Woodland
Hills, Calif. / Ampex Corporation,
Culver City, Calif., a corporation of
California / Memory addressing apparatus.
3,428,954 / Charles Antoine Marius David, Charenton, France / Societe Industrielle Bull-General Electric (Societe Anonyme) , Paris, France /
Element for resistive permanent memory.

3,428,957 / David Rodney Hadden, Jr.,
Cold Spring Harbor, N. Y. / The
United States of America as represented by the Secretary of the Army /
Data storage device using sonic wave




3,430,213 / Kenneth R. Shoulders,
Woodside, Calif. / Stanford Research
Institute, Menlo Park, Calif., a corporation of California / Data· storage
and logic devic.e.
(Please turn to page 58)

~@~AJ ~M[J® [Q)@J17[J@~~ [Q)[J@©~®[[lIl)~
~"Jur~~U U:U@~AJ ~@(ifr~~"J[)[J® u81~[Q)8~~~M
u®~r~ ur~ @[[U 17@M[J ~@[[Lrl)[Q)Mr~®[J


All you can lose
is an ulcer.
There's one sure way of finding out
whether a software package will do
the job you want it to do. Test it. On
your computer. We're so confident
of our new TAXPAK that we'll send
it on a free trial basis. In COBOL or
BAL. This software package includes
the latest federal, state and city
taxes, plus FICA and other deductions. T AXPAK eliminates timeconsuming, individual programming,
reduces maintenance and increases
payroll efficiency.
And don't worry about upcoming
changes in tax structure. The low
price tag on T AXPAK includes program maintenance for one full year.
It's tough to stay ahead of shifting
t~xes,. but you can with TAXPAK.

Find out for yourself ... send this
coupon TODAY.
p_ ••••••••••••••••••••••••••••••••••••• q

Wagner Automation Center, Inc.
7520 Red Road, South Miami, Fla. 33143
Yes. I want TAXPAK on a
Free Trial basis.
I'm getting a computer soon.
Send TAXPAK documentation.


Name _ _ _ _ _ _ _ _ _ _ _ __
Title _ _ _ _ _ _ _ _ _ _ _ _ __
Company _ _ _ _ _ _ _ _ _ _ __
Address _ _ _ _ _ _ _ _ _ _ __
City _ _ _ _ _ _ _ _ _ _ _ _ __

State/Zip _ _ _ _ _ Phone _ _ _ __

t"J8[Jfiil~r L\~lI'~orn~r~~OU'il CGIl'il'~~rr,



7520 Red Road, South Miami, Fla. 33143 • Phone (305) 665-4733
Designate No. 17 on Reader Service Card



Talk to the IBM-360-the NCRG. E. - RCA - Etc.

Secure a higher future for yourself in Data
Processing with a complete Home Study
Course in COBOL - Set your own time;
pace and place. In a short time you can
master COBOL. Write for information about
our NEW accredited:' home study course.



JOINT COMPUTER CONFERENCE in Boston May 14-16. Only 20,000 were
expected. This, combined with a
lack of hotel space, exhibit
space, and parking space, will
probably insure that SJCC will
not be held in Boston again -- at
least not until improved facilities are available.
Only 4500 of the required 8000
hotel rooms were available, and
some of those were so far away
that Conference attendees became
Boston commuters. This, in turn,
created traffic and parking problems. Exhibit space available
was reportedly about half that
neededj 127 companies that wanted
exhibit space were turned down,
and the amount of space assigned
to exhibitors left many dissatisfied. Workmen, carpenters, and
decorators were also in short supply. Even the lease caused problems •. The computer convention was
scheduled to begin moving in five
hours before a prior show ended.

Write Automation Training, Inc. Dept. 14
5701 Waterman' St. Louis, Mo. 63112
Division of Technical Education Corp.
;.An accredited membf'r National Horne Study CrJlJllcii

Designate No. 10 on Reader Service Card


Manufacturer's Maintenance
Also Extensive Software
For Petroleum Exploration
Data Processing
Box 231, Computers & Automation,
815 Washington St., Newtonville,
Mass. 02160

ADD/SUBTRACT 6-DIGIT HEXADECIMALS in seconds with 100% accuracy.
The pocket HEXADDER $15. Free
brochure from Hexeo, Dept. C, Box
55588, Houston, Texas 77055

on May 15 by stockholders of both
firms. SDS will function as a
largely independent, wholly-owned
SUbsidiary of Xerox. Max Palevsky,
SDS president and chief executive
officer, will become a member of
the Xerox board and chairman of
its executive committee. Arthur
Rock, SDS chairman; Dan L. McGurk,
executive vice president; and Sanford Kaplan, senior vice president
of administration, will also be on
the Xerox board.
Xerox stockholders also approved
a 3-for-1 stock split made possible by an increase in authorized
common shares from 30,000,000 to
90,000,000 with no change in par
value. The merger agreement provides that SDS stockholders receive one share of the old Xerox
stock for every 'two shares of SDS.
With the stock split, SDS shareholders get three Xerox for every
two SDS shares.
At the same time that the merger was announced, Computer Access,
a division of Computer Sharing Inc.
(which is a subsidiary of Scientific Resources Corp.) announced that
it has leased the SDS 940 computer
equipment operated by Scientific
Data Systems in E1 Segundo, Calif.
The take-over was to "take place
immediately". Computer Access,
with headquarters at Bala-Cynwyd,

Pennsylvania, is planning to operate at the SDS facility until they
acquire a permanent location of
their own.

H.R. 10791, was sponsored by Rep.
Jack Brooks, D-Tex. "The executive
branch can no longer prepare the
budget submitted to Congress each
year without the use of computers,"
Brooks said. "The complexity of
the budget and the vital importance
of Congress' maintaining control
over federal expenditures make the
use of computers by the Congress
an absolute necessity."
Under the bill, the Comptroller
General and Director of the Budget
Bureau would develop a computer
system to support the budget and
appropriations cycle for use in
the Federal Government.

AN AGREEMENT FOR A "COMMON AUTOMATED RESERVATION SYSTEM" for airlines and travel agents has been
filed by Telemax Corp. of Fairfield,
N.J. The proposed agreement is between the Air Traffic Conference of
America, an association of scheduled domestic air carriers and Atar
Computer Systems, Inc., a new company whose reservation system is
reportedly not yet operational.
Telemax alleges that the agreement between the two companies
would grant Atar an absolute monoply on automated domestic reservation services for travel agents
and commercial firms. The petition
further alleges that the agreement
would sanction a concerted boycott
by the airlines against Telemax and
any other Atar competitor that provides services and equipment similar to those provided by Atar.
Telemax was the first company to
develop a fully computerized instantaneous reservation system for the
travel industry. Its system currently has over 1000 subscribers in
all 50 states.
The petition requests that the
Board disapprove the agreement between the Air Traffic Conference and
Atar or, in the alternative, order
a formal hearing in which Telemax
be given the opportunity to present
evidence and cross-examine witnesses.

I. To the Editor from Dr. B. L. Schwartz
The Mitre Corp.
Westgate Research Park
McLean, Va. 22101
Continuing the discussion of the question of computer
"thinking", treated (again!) in the Multi-Access Forum for
March, I feel obliged to observe that the three questions you
raise in your rebuttal to Mr. Shaw and Mr. Blessing are
excellent; but your three answers are inadequate.
1. What is thinking?
You observe that the usual meaning derives from the precomputer period, when it clearly and unambiguously referred
to an animate activity. Thus, before the computer age, thinking was something that could be done only by a living being.
Yet just three sentences later, you declare that "it is not
scientifically honest to change this definition". But you then
immediately do proceed to change it by a purely arbitrary
extension of the term to try to make it apply to something
it did not previously apply to. Regardless of whether the
reader agrees with or disagrees with your answer, he cannot
be persuaded by such an argument.

2. Does a computer think?
You treat this point by setting up a straw man: programmed computer ("thinking") vs. nonprogrammed computer ("nonthinking", and by your own admission, nonexistent!). But the question is not whether the computer is
programmed or not, but whether the program is the product
of "thinking". What people mean when they use the word
"computer" in question 2 is not: "programmed computer",
but rather "programmable computer", a hardware entity.
Thus again, your argument misses the point of the question.
The comparison with programmed humans is entirely spurious. Humans are indeed conditioned (although I would
hardly call it "programmed") by their years of interactions
with the outside world. But this conditioning does not lead
to each situation having a single well-defined (if unanticipated) response, as it does with a programmed computer. It
is precisely because humans are not fully predictable, because
when they do have choices they can think out which they
prefer in any situation, that they differ from machines.
3. Is it worthwhile to decide whether computers think or
'do not think?
What is or is not worthwhile is a matter of individual
judgement. I fully acknowledge your right to weigh the factors, reach your own conclusion, and present arguments in its
favor for others to consider. Having read your conclusion
and its supporting argument, I remain un convinced. On the
contrary, I feel it is important to terminate the nonproductive effort spent in arguing such unanswerable questions, so
that the energy can be directed into more fruitful channels.
If you feel that such arguments have been fruitful, I would
be happy to learn of any specific examples of their fruits.

II. To the Editor from Sam Bisignano
1115-H University Village
Michigan State Univ.
East Lansing, Mich. 48823
In the March 1969 issue of Computers and Automation
you cunningly answered the three questions:
1) What is thinking?
2) Does a computer think?
3) Is it worthwhile to decide whether computers think
or do not think?
I sincerely believe that the manner in which you interpret
"think" is a suitable restriction of the word, cleverly done, to
allow the performances of a computer to fall into this category. I agree with you that computers, just as human beings
and many animals, "can take in problems, mentally work out
sensible solutions, and apply these solutions in situations", but
is this all there is to thinking? If so, then for any brilliant
human mind, one could find a computer capable of superseding the thoughts and decisions this human might have. However, is there a computer capable of composing music, for
example, more beautiful than Beethoven's Ninth Symphony?
Not yet; I know!
I do appreciate the achievements of computers today and
in the future, but we should not conceive of a computer as
possessing a superhuman mind. If we do, and if it be true
and we allow this superhuman mind to succeed our mind's
creativity, then possibly some day an IBM or an SDS computer will be editor of Computers and Automation (a tentative
title), and will publish an article such as: "Is it worthwhile
to decide whether humans think or do not think?", in the
"eyes" of a computer.

III. From a Pastoral Letter from
The Rt. Rev. Robert L. DeWitt
Bishop, Diocese of Pennsylvania
202 West Rittenhouse Square
Philadelphia, Pa. 19103
As yet not widely known or understood is the fact of
cybernation - the control of systems involving the interaction
of automation machines and computers. Highly sophisticated
production and processing systems are now being designed
which can outperform the most exceptional human capacity.
They can perform the most complicated tasks faster and more
precisely than can humans; and they do not receive wages,
vacations or fringe benefits, nor do they tire or retire. They
can detect their own errors and those of others, make judgments, remember, search their memories for programmed data
and for the new implications of data acquired. The potentialities of cybernation systems are virtually unlimited and
their application is not limited simply to mechanical tasks
but reaches up into the level of at least middle mapagement.

What this adds up to is simply that the time is near at
hand when machines will do as creditable a job of original
thinking as do most middle level people holding responsible
positions today. Work, as we have traditionally known and
understood it, may exist only for an elite minority of highly
trained and highly specialized people. For the majority of us,
there will be no work as work. Obviously our attitudes towards work, leisure, play and social responsibility must
change radically between now and the time soon coming.

IV. From Henry Jean, Vice Pres. and Gen'I Mgr.
Christopher Douglas Associates
801 Second Ave.
New York, N.Y. 10017
In regard to your discussion on computer thinking, we
would like to report on a system we have developed. This
system of programs can make the computer associate ideas.
The system brings about the association of concepts regardless of the actual words the operator uses.
The person using the system doesn't have to know anything
about computers or programming to operate this system; all
he must be able to do is read and write English.
The user simply types out his question, and the computer
answers. If the question is too inexact, or too improperly
phrased, the computer will tell the user he is a little off the
A thesaurus of acceptable terminology will, however, give
the user exactly the right words if he wants to use the thesaurus. It can be either a separate printed document, or the
user can ask the computer to show him the right sections of
the thesaurus.
If the user elects not to use the thesaurus, the computer
will continue to respond until the subject is narrowed sufficiently. Then the desired kind of answer will come out automatically.
We expect the system to be especially useful to those who
need frequent access to the computer but who do not like
being forced to read many feet of printout paper to locate
the answers they want.
If, for example, we apply this system to a large personnel
operation, the expense of the system can be credited against
the cost of the time the personnel executive now spends sweating out his latest statements, analyses, and search skills. Also,
the system will save two to three times the cost of specialized
programming and gives him the reports and programs he
needs within hours or days instead of weeks.
The manager or technician, whoever uses the system, is free
to do very nearly as he likes regardless of his computer programmers and operators.
The system can be used to automate with one computer, in
a time-sharing mode, all the normal business operations that
ordinarily require separate computers or specialized programming.
This system translates into logical or mathematical form

the concepts that the user types into the machine in English.
The relative values expressed by the mathematical statements
form the limits within which the computer relates or matches
the data sought.
We have applied well-developed military command-andcontrol computer methods to business and commercial problems in developing the system.
We also took advantage of .multi-dimensional concepts'to
help the computer locate different groups of data that are
analogous though not necessarily identical.
The user can get anything out of the system he wants, provided it's in there to begin with.
This system will tell you everything it knows or as little as
you like, right down to a one-figure or two-figure answer.
We would like to invite your readers to investigate the
extent to which this system in their opinion really demonstrates functions ordinarily considered to be thinking functions.

V. From the Editor
In reply to Mr. Bisignano, personally I look forward to the
day when a computer will be editor of Computers and Automation, and will produce much better editorials and discussion than the present editor is capable of. I do believe, however, that such a day is distant, probably on the order of 20
to 40 years away.
In reply to Dr. Schwartz, of course, as soon as thinking is
defined as an activity which can only be done by living
beings, then by logic a computer does not think, since (1) a
computer is a machine and (2) a machine is not living. But
the day may come when living machines are made either by
human beings or by other living machines. Until then, however, the argument is settled by the definition.
In order to make thinking done by machines to be not
fully predictable, all that is needed is to include in a computer a source of random events or numbers, such as a cosmic
ray counter. Then the "thinking" done by a computer is no
longer predictable.
Of course there are large areas of thinking done by human
beings in which so far computers have not been programmed
to perform successfully. Examples include: driving a school
bus; understanding spoken language; composing a "good"
symphony; interviewing people about a news event and writing a report on it; etc. For a number of years such areas of
thinking will remain the province of human beings. One of
the big barriers, for example, that I see to programming
such work on a computer is: enabling a machine to "observe"
an environment and interpret what it perceives in that environment as objects or events that can be named as language
names them. I am not even sure how to program a computer to recognize a red traffic light at any of a thousand
street intersections - something which most human beings
who drive a car can do "without thinking", i.e., without using
more than a small part of their minds.

I. A letter to Robert M. Shapiro from
Bernard A. Galler, Pres.
Association for Computing Machinery
211 East 43 St.
New York, N.Y.
In response to your letter of March 3, 1969 published in
Multi-Access Forum in the April issue of Computers and
Automation, it came as a complete surprise to Jean Sammet

(the chairman of our Committee on Special Interest Committees and Groups) and the Headquarters staff, as well as
myself, to hear that you are the secretary of SICSIC with a
mailing list of 100 names. Jean had no knowledge of this,
nor could your statement be confirmed from the files in her
possession. Unfortunately, SICSIC has never seen fit to comply with the provision of ACM Bylaw 7 which requires every
SIG and SIC to file a roster of members at Headquarters
yearly, and to submit a report to the Council yearly. Thus

there is no official record of any of the activities of SICSIC,
and we cannot be expected to ferret out a secretary, mailing
list, and round table discussions. Because this Special Interest Committee was dissolved, I must ask you not to issue
statements in the name of either ACM, or ACM SICSIC.
I am delighted to find out that there is activity under way
in N ew York, but I wonder whether you are perhaps confusing local chapter activity with a national SIC. We are investigating the status of this.
With regard to your formal request that SIC2 be reinstated immediately, my answer must unfortunately be negative. Each SIC commits ACM to spending a fair amount of
money, and we do not currently have evidence that this is a
viable committee, let alone a thriving one. However, Mr.
Robert Bigelow is attempting to start from scratch to comply
with our requirements for a new Special Interest Committee,
and you should certainly contact him at 39 Grove Street,
\,yinchester, Massachusetts 01890. You may also be unaware
that there is an AFIPS committee in this area, and we are
represented on it by Dr. Anatol Holt.
I have been delighted to see the number of people who are
concerned about this subject, because I think there should be
a great deal of involvement in this area. What we wish to do
now is make sure that if a new Special Interest Committee is
formed, it will be strong enough to become a permanent
financially self-supporting Special Interest Group within a
year, as is our policy.

II. From Jean E. Sammet, Chairman
ACM Committee on Special Interest Committees
and Groups
IBM Corp.
545 Technology Square
Cambridge, Mass. 02139
In the "Multi-Access Forum" in your April, 1969, issue you
published a letter to ACM President B. A. Galler from Mr.
Robert Shapiro who stated that he was the. Secretary of
SICSIC. An answer from Professor Galler was sent to Mr.
Shapiro essentially indicating that the claim was a surprise to
Professor Galler, the Headquarters staff, and myself, and
could not be confirmed by any files in my possession. Further
details and comments are included in the letter from Professor Galler [published below]'
Your statement that "It seems ... unlikely that the President of the Association for Computing Machinery by his sole
action has the power under the Constitution to dissolve a
Special Interest Committee." would be correct if we were
dealing with a Special Interest Groul}, but is not correct as
applied to a Special Interest Committee. The fundamental

difference between a SIC and a SIG is that the former is a
temporary organization formed to ascertain whether there is
enough interest and activity to sustain a permanent organization (namely a SIG). Action was taken because of the provision in ACM Bylaw 7, Section 6, which states that "A special interest committee shall be established for a period of
one year. The Council may, at its discretion, continue a committee for an additional period of time if circumstances warrant. A special interest group shall exist until eliminated by
Council action." Please note carefully the difference in treatment between Special Interest Committees and Special Interest Groups. Professor Galler announced at the December
1968 Council meeting that he had taken the action of dissolving SICSIC and nobody protested this or raised any questions
about it. According to the Bylaw, it was in some legalistic
sense already dead, and the action was not "null and void".
However, you and your readers might appreciate a little more
background on how this came about.
When I was appointed Chairman of the ACM Committee
on Special Interest Committees and Groups, one of the first
things I did was to determine which Special Interest Committees had been in existence for more than one year. I then
contacted each chairman (including the chairman of SICSIC)
requesting that he inform me of any special reasons that his
SIC should not comply with the provision that it convert to
a SIG or else dissolve. In all cases except SICSIC, there
was indication of interest and activity from the chairman
and/or the members, and I therefore requested the Council
to approve (and they did) their continuance as Special Interest Committees until specified dates. However, no answer
was received from the SICSIC chairman, and no mailing list
(indicating current members) was available. No ACM members had requested information from, nor expressed complaint about, the inactivity of SICSIC. I felt, under the circumstances, that I had no choice but to recommend dissolution, and this recommendation was supported by several
other responsible officials. The recommendation was reluctantly accepted by Professor Galler.
I must regretfully disagree strongly with your statement
(on page 14) that "There is no doubt at all that it is a
vigorous and functioning Special Interest Committee". If it
had been, then it never would have been dissolved. It is
essential to understand that Special Interest Committees and
Groups are formed by the wish of the members of the society,
and not by direct action of the leaders. When the chairman is non-responsive to correspondence, no activities are
undertaken, and no members can even be identified, this is
not a committee of any kind, let alone a group of people who
voluntarily banded together to fulfill some mutual interests
and goals. In essence, we removed some words from some
pieces of paper.

Mrs. Phyllis Hyde
Box 4068
Santa Barbara, Calif. 93103

(2) The clear-sighted diagnosis of technology-gone-wrong
as illustrated by the acquisitive determination of
ABM policy;

I like a number of things about your April, 1969 editorial
("The Misdirection of Defense - and the Social Responsibilities of Computer People") :

(3) The fact that 80% of computer application is
civilian, not military. But does this mean really not
military-serving? Or might this look entirely different were military-supportive civilian applications subtracted from that figure?

(1) The title, which sees social responsibility as belonging to people - one at a time; not the "industry" or
some other magic personification of an "it" which
exists only in the head of a (primitive-overlaidintellectual) person;

I read eagerly through your suggestions for how "Computer
people ... should now seek to fulfill their social responsibilities". What you outlined is not to be gainsaid. But the

frankness of your article about the money facts of life for the
military-industrial complex in a demilitarized economy
seemed noticeably to shy at the logical carry-through of
people responsibility: some one person has to opt for his own
larger self-interest vs. his own narrower dollar-interest, like
refusing to serve the military-industrial complex dollar-

proliferation at the cost of humane value-proliferation. That
might mean a hunk of jJeople paychecks - not just the computer industry - which is precisely the way the Defense
Department is going to get "reoriented". People will do it,
one at a time, if it gets done at all; and it will be peojJle, one
at a time, who "make less money".

Grace J. Kelleher
400 Army Navy Drive
Arlington, Va. 22202
As a member of the Long Range Planning Committee of
the Operations Research Society of America (ORSA), I have
undertaken the task of investigating various facets of the
increasing interchange between the terms "operations research" and "systems analysis". I am sure you have noted
how these terms are more or less used interchangeably by
many people today. The semantics become even more complicated when we consider that the terms "cost-effectiveness"
and "cost-benefit analysis" also are being used interchangeably with "operations research" (OR). And the problem is
further complicated by the fact that many employers and
professional recruiters have shortened the term "computer
systems analyst" to "systems analyst", thereby precluding the
general use of an occupational term which was intended to
apply (or could apply) to the analysis of any type of system
(space systems, logistic systems, physiological systems, etc.).
This semantics problem is more apparent to some than to
others, and also disturbs some more than others. However, the
long-term problem, as I see it, is that communication between
prospective employer and prospective employee is becoming
more and more complicated because of the subjectivity with
which each may choose the term that best describes his needs
on the one hand, or his capabilities on the other.
This topic will be discussed at the Session of our Committee at the ORSA National Meeting June 17-20 in Denver,
Colorado. To aid in your consideration of this semantics
problem, a list of selected excerpts from current literature is
shown below. Viewed collectively, I think they provide a fair
reflection of the semantics confusion that exists and tends to
be growing.

The Semantic Jungle of Operations Research, Systems
Analysis, et al. (as reflected in a series of excerpts
drawn from current literature)
Although certain mathematical techniques have become
associated with and identified with operations research, these
t~chniques do not constitute OR. Rather OR is the application of scientific methods, techniques and tools to problems
involving the operation of systems so as to provide those in
control with optimum solutions to problems.
OR is problem oriented rather than technique oriented; the
scientific methods are a means to an end and not an end in
Operations research: a scientific methodology for examining, defining, analyzing and solving complex problems.
Systems analysis is a body of techniques and theories for
analyzing complex problems.
The comprehensiveness of OR's aim is an example of a
"systems" ajJProach, since system implies an interconnected
complex of functionally related components.
The term management science refers to applications of the
operations research approach in the general area of management.
The establishment of the position of Assistant Secretary of
Defense for Systems Analysis confirmed the role of systems
analysis in DoD and marked the "coming of age" of military
cost-bene fit analysis.
Essentially, cost-bene fit anlysis is operations analysis at one

higher level of optimization, with the various physical inputs
to the operation being variables rather than givens.
The methodology of operations analysis as developed
during 'VorId War II concentrated on the best use of currently available equipment without regard to cost or worth.
Consequently, operations analysis at its inception stopped just
short of ajJplied economics.

Congressman Cornelius E. Gallagher
Chairman, Special Subcommittee on Invasion of Privacy
U.S. House of Representatives
Washington, D.C.
An elk named Moe, which resides in Yellowstone National
Park, was recently electronically connected to a recently
launched U.S. Satellite, Nimbus 3. The comings and goings of
Moe are carefully and accurately recorded by the spy satellite.
The animal's graceful movements are now just a jagged curve
on a graph; his investigation of the beautiful plains is but
another statistic.
No'matter how fleet footed his flight from the night, Moe
can never outrun the super sleuth which hovers over head and
antlers. In other words, Moe may be nimble, but he cannot
beat Nimbus.
Unfortunately, it is not absurd to draw a comparison bet\veen the plight of Moe the elk and the growing plight of
John Doe, the American citizen. Everywhere our technology
is producing new means of invading, indeed destroying, individual privacy in America. The very notion that a satellite flying thousands of miles from the earth can record even

the most insignificant movements of an elk in Yellowstone
National Park demonstrates how awesome this technology has
Indeed, with snooping devices on the rampage, the Creator
may soon not be alone in observing the fall of a sparrow.
Unless we Americans fully comprehend that the destruction
of privacy means the undoing of all our basic rights, then we
may witness a slow and subtle repeal of our National Constitution.
The example of Moe provides an eery lesson for us human
beings. Who can say that the bugging of Moe is not a mere
prelude to the bugging of men? Who can say that the insatiable thirst for statistics may not someday be quenched by the
creation of a total surveillance society in which the movements of man, like the movements of poor Moe the elk, are
open to general scrutiny.
Yellowstone National Park has become an open book to the
Nimbus satellite that records the every movement of one of
its inhabitants. Obviously, it is time to take steps to insure
that our homes are not similarly violated by some inquiring
snooper who, while perhaps acting with the best of intentions, has lost sight of humanity.

Number Puzzles for Nimble Minds
- and Computers
Neil Macdonald
Assistant Editor
A "numble" is an arithmetical problem in '.vhich: digits
have been replaced by capital letters; and there are two
messages, one which can be read right away and a second
one in the digit cipher. The problem is to solve for the digits.
Each capital letter in the arithmetical problem stands for
just one digit 0 to 9. A digit may be represented by more
than one letter. The second message, which is expressed in
numerical digits, is to be translated (using the same key) into
letters so that it may be read; but the spelling uses puns or is
otherwise irregular, to discourage cryptanalytic methods of
We invite our readers to send us solutions, together with
human programs or computer programs which will produce
the solutions.

We have a computer program for the Digital Equipment
Corporation PDP-9 computer which will solve addition
Numbles such as Numble 695. The program was created
by Stewart B. Nelson, Software Editor of Computers and
Automation. A copy of the punched paper tapes for the
symbolic program and the working binary program, and a
copy of the operating instructions will be sent to any reader
on request for the nominal price of $8.00. Please enclose
this amount with your order, and send it to:
Computers and Automation
N umbles, Att'n Neil Macdonald
815 Washington St.
Newtonville, Mass. 02160

Solution to Numble 695
In Numble 695 in our May issue, the digits 0 through 9
are represented by letters as follows:







0= 3



H E = L A







K,L= 7

The full message is: \\Then sorrow is asleep, awake it not.
Our thanks to the following individuals for submitting their
solutions to N umble 694: A. Sanford Brown, Dallas, Tex.;
T. P. Finn, Indianapolis, Ind.; Ross F. Garbig, Toronto,
Ontario, Canada; Ron Geist, Allentown, Pa.; Joe King, Anchorage, Alaska; George Seminara, New Carrollton, Md.; and
Elizabeth C. \Volfe, Baltimore, Md. 21203.

Fred Moncrieff
Ad Hoc Committee
P.O. Box 432
Ann Arbor, Mich. 48108
Formation of a new health record association will be discussed at a meeting in Ann Arbor, Michigan, June 23-24.
Some 100 persons from throughout the United States are expected to attend.
The purpose of the new association would be to fill the
needs for a multi-disciplinary forum for all persons concerned
with medical data systems and to increase the number of personnel in the medical and health record field.
The spread of computerization, telecommunication, comprehensive planning, integrated health care, regional medical

programs, health data banks, and other modem developments
demonstrate the need for such a broad-based forum.
In addition to medical record librarians, many other persons today are making major contributions to the field and,
through their efforts in research and development, are shaping the future of health record practice. Among these are
physicians, biostatisticians, systems analysts, and computer experts concerned with the recording, storage and retrieval of
health data.
It is anticipated that the new health record association
would supplement rather than duplicate the functions and
goals of existing groups.
All persons interested in the meeting and / or the aSSOCIation are invited to write to the address above.


Robert T. Stelloh, Program Manager
International Computing
3235 Kifer Rd.
Santa Clara, Calif. 95051

Mr. Stelloh received his education at the Univ. of
Missouri School of Mines, Rolla, Missouri, graduating
with an M.S. in Mining Engineering in 1961. He has
worked for Lockheed Missiles and Space Co. as a systems
programmer, with particular interests in machine-independent programming and medical data processing; and
as a medical systems consultant. He is currently a senior
partner of ICC and manager of their office in Santa
Clara, California, with specific responsibility for the
ABIIS, and for a medical business system under development by ICC.
In his letter submitting _this article, Mr. Stelloh said:
"Please excuse the very rough draft state of the manuscript. As senior author, I had the responsibility for
completion of the article, including preparation of the
final manuscript. Unfortunately (from many aspects), I
became aware that I had tuberculosis ten days ago, and
was immediately and unexpectedly incarcerated in an
isolation ward of the county hospital. This had the
initial effect of distracting me from the article, and the
final effect of forcing an inexpert typist (me) to type it
on a very tired typewriter. I beg your understanding.
If I may be of further assistance, I can be reached at
[name of hospital]."
The editors of C&A would like to publicly express
their admiration of Mr. Stelloh for submitting his article
under these conditions - and to wish him a, full and
rapid recovery.

The Automated Blood Inventory Information System
(ABIIS) is a general system for data file creation, maintenance, and data retrieval. It was conceived in response to
the question "Can automation help in blood resource management?". Feeling that the answer is a strong "Yes", the
Milwaukee Blood Center (MBC) and International Computing Company (ICC) are developing the ABIIS as a prototype system to demonstrate the feasibility of inventory
control and a management information system for a blood
bank, and to serve as a test bed for the design of a production system. This article outlines some problems unique to
blood resource management, the design approach used with
ABIIS, and some implementation and operational aspects

Goals of a Blood Bank
Human blood is in many ways as priceless as a human heart
or a human kidney - it fills a unique need in the preservation of human life, and it can only be obtained from human
beings. Accordingly, the primary goal of a blood bank is to
meet all requests made upon it for blood, with the strong
secondary goal of minimizing waste.
In view of the 21-day shelf life of blood and the wide fluctuations in demand for blood, these goals are in strong
conflict, and make determination of optimum inventory
levels a very difficult problem.
Associated problems include control of costs, control of
incoming blood from volunteer donors (i.e.,' trying to match
unpredictable input with equally unpredictable output demands), control of average inventory age, control of disease
transmission via blood transfusions, and determination of
proper inventory distributions, both between bank and hospital, and between ~hole blood and blood components.
The Milwaukee Blood Center is a regional blood center
serving the blood needs of 34 member hospitals in a fourcounty area of southeastern Wisconsin centered around Milwaukee. Approximately 50,000 units of blood are processed
each year, providing for the needs of about one-third of the
population of Wisconsin.
Blood is collected at an average rate of 175 units per day
from an active donor population numbering over 70,000. A
daily inventory of whole blood and blood components is

Note: This work is being performed under contract PH-4368-1425 of the National Blood Resource Program, administered
by the National Heart Institute of the National Institutes of


"Human blood is in many ways as priceless as a human heart or a human kidney - it fills a unique need in the preservation of human life)
and it can only be obtained from human beings. The automation of
the Milwaukee Blood Center was conceived as a positive response to the
question: 'Can automation help in blood resource managementl))

maintained at an average level of 1500 units, with about
40% of the inventory at the blood center. Approximately
1000 transactions are performed against this inventory each
day, to record the following operations:
• Draw, process, and add a unit of blood to the inventory
• Convert whole blood units into blood components
• Ship blood units to and from member hospitals
• "Cross-match" blood units to ensure donor-recipient
• Transfuse blood into a recipient, or otherwise dispose
of it
• Maintain a file of blood donors
• Maintain a file of blood recipients
Functional Requirements
Review of the overall blood center operations and needs led
to the development of the following key functional requirements for a blood center information system:
• Operation by personnel unskilled in data processing.
Blood center personnel must be able to use the system
with a minimum of special training, which dictates
very simple and even helpful interface procedures and
the use of familiar interface equipment such as typewriters.
• Production capability.
The system will be used to process large numbers of
transactions on a day-to-day basis, which implies userlevel efficiency or economy, reliability both in terms
of the machine and the user, and the need for fixed
interface procedures in spite of future system additions and modifications.
• Experimental capability.
The system will be used to support yet-undefined experiments in blood resource management, and thus
must be very flexible and versatile, in contrast to the
production requirements. In addition, these experiments must be supported without disturbance to the
production system.
• Provide data for research.
The system should provide a long term historical record of the blood center operation, to facilitate any
future studies of the blood center. Data conversion to
machine readable form can be a very costly item.
Since this system must capture the data for its ongoing operation, it can readily save the data for later
use at very little additional cost.
• Prototype design.
The major thrust of the contract is a feasibility study
and design of a production system. Thus, the protoCOMPUTERS and AUTOMATION for June, 1969

type ABIIS must have the flexibility to allow testing
of concepts and development of management policies,
and must contain program "instrumentation" to aid
in the design process.

At this point, it was decided the Milwaukee Blood Center
and National Heart Institute needs would not be met by the
normal approach of careful development of specifications
with a "cast-in-concrete" program written to those specifications, but that they would only be met by a very general and
flexible software system. Additionally, it was decided to acquire a dedicated hardware system to allow maximum freedom in development of the overall system, provide unlimited
online operation, and to allow the project to be under the
complete control of the Milwaukee Blood Center. The Scientific Data Systems Sigma 2 computer system was selected,
with the following hardware configuration:
• Sigma 2 CPU - 16K 16 bit words, 1.1 microsec. cycle time
• RAD disc storage unit - 3 megabytes, 17 millisec.
access time, 188 KC transfer rate
• Magnetic tape units (2) - 20KC transfer rate
• Communications controller
• KSR 33/35 teletype terminals (5)
Program Design Goals
With the functional requirements in mind, a number of program design goals were established to serve as guides for the
detailed program design. These goals are:
• Machine transferability.
To be of maximum benefit to the NHI, programs
developed for the Milwaukee Blood Center computer
should be easily transferred to other computers. This
is accomplished primarily through the use of a "machine-independent" language such as FORTRAN or
COBOL, since compilers are provided by most computer manufacturers for their specific computers. A
high degree of transferability is also achieved by actively ignoring specific machine characteristics such as
word length, number of words, and input/output devices; and by segregating any required machine-dependent instructions into a few, easily identified program segments.
~ Dependability.
The day-to-day operational requirements impose
the need for back-up and recovery procedures to allow continued operation in the face of equipment

failures, and the need for extensive user error detection and correction procedures to insure the input of
reliable data. Dependability also implies the need for
careful modification of the system to avoid the introduction of errors and attendant degradation of a once
reliable system.

• Flexibility.
A great deal of flexibility is required to allow the
system to accommodate blood resource management
experiments and to test systems design concepts as
they evolve through operational experience. This
flexibility begins with the user interface, in terms of
what the system accepts from and provides to the
user; continues with the data base design, which
must be capable of being easily expanded to include
additional items of interest, and modified to allow
examination of the inter-relationships of these data
items; and concludes with the overall program structure, which must allow the inexpensive and reliable
introduction of major program changes and additions.

• Simplicity.
The system/user interface must be simple, reasonable,
easy to learn, and must provide help when requested.
Internally, simplicity implies modular program structure and avoidance of sophisticated programming
techniques, to minimize the effort required for program modifications.
The "machine-independent" language chosen for the
ABIIS was FORTRAN, since the ABIIS is apt to be installed on other small computers, and, while FORTRAN is
implemented to some degree on most small computers,
COBOL is not. To further enhance machine transferability,
_the FORTRAN used corresponds very closely to the ASA
Basic FORTRAN standard!, further restricted to those language features felt most likely to be implemented by other
computer manufacturers. Character and bit manipulation
was facilitated by developing machine language functions for
AND, OR, LSHIFT (left shift n bits), and RSHIFT (right
shift n bits).
To achieve the flexibility required for the ABIIS, a set of
general purpose file handling routines were developed in
FORTRAN for both random access and sequential files.
These routines are called in lieu of the standard FORTRAN
input/output statements. Use of calls on these routines
ERASE, etc.) allowed the ABIIS files to be designed to best
meet ABIIS objectives, without regard to the restrictions inherent in standard FORTRAN input/output statements.
Additionally, the input/output is uncoupled from the formatting for formatted input/output (teletype, etc.), which
provides more flexibility and simplifies overlapping of computation with input/output.
File Maintenance and Retrieval Language
A file maintenance and information retrieval language
called QUERY was designed to allow file processing operations in a blood center environment to be expressed simply
and succinctly. It was felt that use of this language would
allow the majority of the software to be developed in parallel
with the definition of specific user requests and report formats, and would allow the rapid implementation of these
l"FORTRAN vs. Basic FORTRAN," Communications of
the ACM, Vol. 7, No. 10, Oct. 1964, pp. 591-625.

specific user requirelucntS, both initially and as they changed.
QUERY has now been implemented as an interpretive processor, written in FORTRAN. This proved to be a very
successful approach, allowing the programming and specification to be performed in parallel as planned. It has also
had the very desirable effect of allowing significant ABIIS
modifications to be made by programmers knowledgeable
only in the QUERY language.
Language for Terminal User
A simple language was designed for the terminal user,
consisting of a number of verbs, such as ADD (add a unit
of blood to the inventory), with modifiers for each verb to
either request further input, or restrict the output in the
case of output-directed verbs. This language is implemented
as a number of QUERY language programs, one for each
verb. Use of these verbs includes access to an extensive system of cues or prompts, which are text messages designed to
elicit the correct response. The terminal user has full control over the amount of information presented for his assistance, to the extent of suppression of all cues when he becomes sufficiently proficient. However, the cues are always
available, and are automatically given in the case of an error,
or when the user forgets the form or content of a data field.
It can be seen from the preceding that ABIIS consists of a
hierarchy of program modules. This complete hierarchy
consists of:
• The terminal interface routines, written in FORTRAN, which initiate and maintain a dialogue with
the terminal user. These routines provide the user
prompts associated with the terminal language verbs,
and call up the appropriate QUERY program for
each verb. With respect to the overall ABIIS structure, these routines know only the logical structure of
data files and user dialogues, and the calls on the file
• QUERY programs, written in QUERY, which provide a means for the terminal user to interrogate or
modify the data files. These programs must know the
logical structure of data files, and the format of file
handler calls and terminal interface routine calls.
• The QUERY processor, written in FORTRAN,
which executes QUERY programs. This program
knows the structure of QUERY programs, file handler calls, and terminal interface routine calls.
• The file handler, written in FORTRAN, which carries out all data file manipulation in response to requests from higher level routines. The file handler
translates the logical structure of ABIIS files into the
physical structure understood by the manufacturersupplied software and hardware. To do this, the file
handler knows both the logical and physical structure
of the data files, and the format of calls on the input/
output support routines.
• The input/output support routines, written partially
in FORTRAN and partially in machine language,
which interface between ABIIS and the hardware.
These routines are machine-dependent, and must know
the physical structure and characteristics of the input/
output devices (disc, tape, communications equipment) and the format of calls on the manufacturersupplied input/output handlers.
In addition, some overall ABIIS control routines exist
which initiate ABIIS execution and otherwise control execution under the framework provided by the manufacturersupplied operating system. These routines are by their nature dependent on the operating system. Finally, a number

of ABIIS maintenance routines are available to modify or
become aware of the current state of ABIIS software and
files. These routines make use of the file handler and terminal interface routines, and are thus machine independent.

Minimizing Program Development Time'
Although a high degree of machine transferability implies
little dependence on manufacturer supplied software, it was
decided to minimize program development time and cost by
making as much use of this software as possible. The dependence is automatically minimized by the program hierarchy,
which has the effect of channelling all potential references
to manufacturer software through ABIIS routines, primarily
the input/output support routines. For example, a user may
ask for information regarding a specific blood unit. This request is received by a QUERY program, converted by the
QUERY processor into calls on the file handler, which further converts it into calls on the input/output support routines, which finally convert it into a machine-dependent call
on the manufacturer-supplied disc read routine to procure
the appropriate disc record.
After consideration of factors such as machine availability,
time available for program development, and overall development costs, the program development was begun on a commercial IBM 360/50 time-sharing system. This approach
also deferred the difficulty of implementing a large software
system on a small computer, and enforced adherence to some
of the subtler aspects of machine transferability. When the
file handler and QUERY interpreter became operational on
the IBM 360/50, the system was transferred to the Sigma 2
The ABIIS is now operational, on a test basis, on the
Sigma 2 as a non-resident foreground task under their Realtime Batch Monitor (RBM) operating system. During
ABIIS operation, RBM functions primarily as an input/
output processor, by initiating all requests to the hardware
and processing all hardware interrupts. Since RBM provides approximately 10K words for ABIIS, and ABIIS currently includes over 20K words of program code and constants, the overlay loader capabilities of RBM are used extensively. RBM is also extremely useful during program preparation, serving all the normal functions of a general operating system, and allowing use of the disc file for source
and object program storage.

TSI's new, all solid state, dual audio tone keyboard
brings the exclusive, ultra-reliable TSI Proximity Key
to designers and users of telephone-type communications, credit checking, stock quotation and related
terminal equipment.
Inclusion of TSI's Proximity Key eliminates the maze
of mechanical parts found in conventional equipment.
It insures totally reliable, non-contacting, wear free
operation. Key bounce is eliminated. There is no need
for a high current relay closure type of circuitry.
The TSI dual audio tone keyboard is directly compatible, electronically and mechanically, with current
telephone type push button equipment. It is available
in either 3" x 4" or 4" x 4" matrix operations using
12 and 16 keys respectively.
Full details including price and delivery schedules are
available upon request.

Constructed to the same high
standards and ultrareliability of
larger TSI keyboards, the TSI Mini
is designed for limited space applications requiring thin line con·
TSI PROXIMITY KEYS Available on an individual
basis in either single or multiple output configurations.

This article presents some of the considerations which led
to the development of a very general and highly machinetransferable programming system for the solution of a specific
problem on a very specific small computer. The approach
taken entails considerably more effort than coding to a set
of rigid program specifications. However, since even "production" business systems are changed almost as much as
they are executed, it is felt that the increased ease of modification soon outweighs the additional development effort,
and this is particularly true for a prototype system.
Finally, with respect to efficiency, the authors are firmly
convinced that an excellent design coded in a higher level
language (therefore, "inefficient") is superior to an adequate
design coded in machine language (theref9re "efficient"),
and that this is true even for systems programs such as the file
handler and QUERY interpreter portions of the ABIIS.
Since there is seldom enough time to do everything perfectly,
it is better to spend the time on the design.

If you produce key punch, card reader,
paper tape or disc file equipment, you
should be using TSI Proximity Transducers ... the most reliable and eco·
nomical method for parity checking,
rack peak detection, displacement sensing and hole detection. Sizes from J{/'
0.0. to X" 0.0.

For keyboards write for Bulletin K-9000-A;
for Proximity Transducers Bulletin PT-4000-A;
or telephone.

Easton and Wyandotte Roads Willow Grove, Pa. 19090
(215) 657-0655

Designate No. 16 on Reader Service Card


James A. Boyle, M.D.
Department of Medicine
Basic Sciences Bldg.
Univ. of Calif. at San Diego
La Jolla, Calif. 92037

((At present no one has produced a practical diagnostic system which
takes into account the risk of misdiagnosis. However such systems have
been developed from a theoretical standpoint - and the way is now
open for their incorporation into practical diagnostic computer models."

What have statistical methods of diagnosis by computer to
recommend them? The data (patients' complaints, clinical
features and laboratory tests) on which doctors make decisions as to what is wrong with a patient are usually highly
variable in most diseases and there exist specially designed
statistical methods for decision making in such circumstances.
Moreover it seems reasonable to suppose that the computer
with its huge capacity for information storage and data processing would be of value to the doctor faced with a difficult
diagnostic problem. We still have a long way to go before the
computer becomes a routine diagnostic aid and there are
those who believe, with some justification, that this day will
never come.

Diagnostic Situations in Clinical Medicine
Medical diagnosis may be divided into three categories. In
one the doctor is faced with the interpretation of clinical
diagnostic tests such as the analysis of an electrocardiogram
or electroencephalogram. In this area the use of the computer
is well established as a diagnostic aid. For example in some
hospitals the on-line analysis by machine of electrocardiographic patterns is offered to the clinician as a routine service;
many hours of his time are thereby saved.

James A. Boyle graduated in medicine at the University
of Glasgow, Scotland in 1960. He first became interested
in the possibilities of automated diagnosis in 1964. At
present he is working in the Department of Medicine at
the University of California at San Diego.


In the second category of diagnosis is the situation where
the doctor is starting from scratch with a new patient. Here
he has initially at any rate no idea of what the final diagnosis
is going to be. It may be that the patient will turn out to
have a ruptured appendix or he could be suffering from an
overactive thyroid gland. It seems unlikely that computerassisted diagnosis will ever be feasible in this situation because
there are far too many possible diagnoses that a new patient
could have and the amount of information required to prime
the machine would be astronomically great. Nonetheless in
one study where the results of the Cornell Medical Index
Questionnaire (CMI) \vere analysed by machine, promising
results were obtained. The CMI takes a fairly comprehensive
medical history from a patient. It lists 195 questions of the
"Yes" or "No" variety pertaining to alterations of body
functions known to be associated with disease states. The
patient is also asked about his past medical history and his
family and psychiatric history. A correct machine diagnosis
was reached on the basis of these data (out of a total of 60
possible diagnoses) in 48 per cent of 350 patients. A clinician,
experienced in the use of the CMI was correct in 43 per cent
of these cases.
In the third category of diagnosis the doctor makes the
correct diagnosis from a relatively small number of alternatives, for he knows roughly the area in which the patient's
trouble lies. For example if the patient presents a goiter then
the field of diagnostic possibilities is narrowed down to
some form of thyroid disease. The number of diagnoses the
doctor has to consider falls from almost infinite to perhaps
less than ten. In this area of diagnosis the computer may be
the greatest value as an aid to the clinician.

Forming the Computer Memory of Disease
Samples of patients with soundly established diagnoses are
needed to provide the computer with a memory or "experience" of the diseases in question. Once these have been
obtained, the incidence of symptoms and signs which these
patients exhibit and the results of special investigativ.e procedures are recorded in the form of a data or probability matrix.
Part of such a matrix is shown in Table I where a few of the
Table 1

having the largest probability. This procedure, modified
slightly to take account of the frequency of occurrence of the
diseases themselves in the general population (which obviously
has some bearing on the probability that the patient has or
does not have a certain condition), is sometimes called the
Bayesian approach and it was suggested as being applicable
to medical diagnosis as far back as 1959. It assumes that the
symptoms and signs which the patient exhibits are conditionally independent, that is to say that the likelihood of
finding a certain symptom in a patient is not influenced by
the presence or absence of another symptom.

Probability of disease:
Simple Thyroid
Outcome Disease
Goiter Cancer

Practical Example of the Theory


Clinical Data
Precipitin test:





An example may serve to illustrate the diagnostic strategy
which many programs employ and may explain the foregoing
more clearly. Part of a probability matrix is shown in Table
II. This Table depicts the likelihood of finding a certain






Table 2

Roentgenologic evidence of
deviation of trachea:





Vocal cord paralysis:





Fixation of goiter to





0-3.0 0.6316
3.1-5.0 0.3421
5.0 up 0.0263



0-1.0 0.3478
1.1-10.0 0.5870
10.0 up 0.0652



Protein bound iodine
(lJ.g per cent):
Duration of disease
(years) :

These figures are derived from the percentage frequencies
of occurrence of symptoms and signs in the three diseases.
For example, two per cent of patients with Hashimoto's disease were found to have a goiter which was fixed to the
surrounding tissues.

symptoms (discomfort, duration of disease), signs (vocal
cord paralysis, fixation of the enlarged goiter to the surrounding tissues), and results of investigative procedures (precipitin
test, roentgenologic evidence of tracheal compression or deviation, plasma protein-bound iodine) are shown for a sample
of patients with three different forms of thyroid disease:
Hashimoto's disease, simple goiter and thyroid cancer.
The data accruing from examination of a new patient are
fed into the machine and compared with the data in the
probability matrix. The most probable or likely diagnosis is
then computed.

Theoretical Considerations
Computer programs have been used by different workers to
calculate the most probable diagnosis. Most of these techniques are based on the general theory of allocation. Suppose
that we have a complete set of data on one patient and we
wish to allocate him to one of several diagnostic possibilities
or diseases. Suppose too, that we know the likelihood of finding each individual piece of diagnostic information that this
patient exhibits, given that he has first one of the diseases and
then another. Then, provided that the patient has been randomly drawn from a population of patients with these
diseases, we can use conditional probability theory to calculate
the probability that he has each of the illnesses we are concerned about. Vve then allocate the patient to the condition

Clinical Data

Outcome ~







Patient has f 3 • Likelihood of disease A = 0.05. Likelihood
of disease B = 0.05. Likelihood of disease C = 0.80. Thus
disease C is 16 times more likely than either disease A or
disease C.

clinical feature F in each of three diseases A, Band C. The
feature has three outcomes: fv f2' and f 3 • As an example of
such a situation one might be talking about a feature such
as the color of hair and the three outcomes might be red,
blond or dark. Suppose that a patient presents with feature
outcome f 3 • Then the likelihood that this is disease A, is
0.05. Diseases Band C have likelihoods of 0.05 and 0.80
respectively. Thus on the basis of these data we can say that
this patient is 16 times more likely to have disease C than
either disease A or B.
Table 3
Clinical Data
















Patient has f3 and s3' Likelihood of disease A 0.05
x 0.02
0.001. Likelihood of disease B 0.05 x 0.10
= 0.005. Likelihood of disease C = 0.80 x 0.60 = 0.48.
Thus disease C is now 480 times more likely than disease A
and 96 times more likely than disease B.



Now consider Table III. This Table shows the likelihood
of a particular symptom S in the three diseases. The possible
outcomes that this symptom may have are Sv S2' and S3' Such
a symptom might be shortness of breath and the outcomes
might be, "Not present," "Present but mild" and "Present and
severe." Suppose that our patient not only has feature f3 but
that he also has symptom S3' Then we multiply the likelihoods for each of these events occurring separately and arrive

at an estimate of the likelihood of their joint occurrence in
each of the three diseases. This has been done in Table III
where it can be seen that he now has a greater than ever
chance of having disease C rather than the other conditions.
Further information about his complaints and the features
recorded by the doctor during his physical examination will
alter the likelihoods of diagnosis in the way which has been
described. Finally after allowance has been made for the
relative frequency with which the diseases occur the patient is
allocated to the disease with the highest likelihood (in this
case, on the basis of somewhat limited information), disease C.

Diagnosis of Congenital Heart Disease
The potential value of the computer as a diagnostic aid to
the clinician has been studied in a wide variety of situations
in medicine. The first practical study to appear was that of
Dr. Homer Warner and his colleagues in 1961. They applied
a machine to the differential diagnosis of congenital heart
disease. This is a field which often poses particularly difficult
problems even to the expert and to the average doctor it may
well be a closed book. Data on 53 symptoms and signs pertaining to children with 32 different forms of congenital heart
disease were fed into the machine. A series of new patients
was studied. It was found that the diagnosis calculated by the
computer agreed with the actual diagnosis (which was found
at operation or by cardiac catheterization studies at least as
often as did the diagnosis determined by three experienced
cardiologists. Presumably therefore the machine performs
better than would a doctor inexperienced in the management
of children with congenital heart disease, although there is
little or no proof of this. By following a simple checklist of
questions on a proforma sheet, the doctor is able to generate
the data which the machine needs to make the diagnosis.
The computer program thus gives all doctors the facility of
arriving at correct diagnoses; it allows the accumulated
experience of a few experts in the field to be available to all.

Diagnosis of Thyroid Disease
In continuing studies conducted in the Nuclear Medicine
Laboratory at the University of Florida College of Medicine,
a computer program has been developed which will make a
correct assessment of the thyroid function of an individual
after his symptoms, signs, and laboratory test results have
been supplied to the machine. The original probability matrix
for this program was constructed from data from 879 patients.
There are many interesting points about this program. First,
the probability matrix is an accumulating one in that it allows
the patient's symptoms and signs and laboratory tests to be
entered into the matrix after the correct diagnosis has been
confirmed. Thus the program is to some extent self-learning
for it becomes modified by its past experience.
Second, a great deal of knowledge of a practical sort has
been gathered over the last few years with use of the program.
Although the basic diagnostic decision model is founded on
Bayesian conditional probability theory, the program has been
modified by subroutines which search routinely for some
10-15 specific combinations of laboratory tests and probability values that are produced by some rare thyroid diseases,
that otherwise might be missed by the machine. Over 1300
cases of thyroid diseases or of patients suspected of having
thyroid disease have been examined by the program. The
most recent published experience of the workers at the University of Florida showed that only 2.8 per cent of the last
500 patients screened had been misdiagnosed by the program.
This is an impressively small number and most doctors would
probably feel quite happy if they could approach this degree
of accuracy in diagnosis.

The third intriguing feature of this computer program for
diagnosis is that it has been used in another country (Germany) and has given very good results. Thus it would appear
that once a program has been developed by one set of
doctors to deal with the differential diagnosis of one set of
diseases, it can be used with profit by another set of doctors.
This facility is important for it is precisely because we wish
to place the accumulated experience of one doctor at the
service of another that the idea of computer-assisted diagnosis
is so attractive.
Many other studies can be cited to show that at the experimental level, computer-assisted differential diagnosis, where
the doctor is concerned to choose between a small number of
diagnostic possibilities, is feasible. Thus programs have been
written to help in the diagnosis of abdominal pain, benign
versus malignant gastric ulcer, bone tumors, adrenal disease
and even headache.

Limitations of Computer Diagnosis
There are a number of limitations to the use of computers
for diagnostic purposes. It may be said at the outset that
none of these are insuperable; but a great deal of work
remains to be done if the technic is ever to win wide acceptance in the practice of clinical medicine. The concept of
"robot diagnosis" may cause antipathy to many patients, for
it appears to strike at the roots of the doctor-patient relationship, a relationship which is still very important in most
Western countries. Machines are now programmed in some
medical centers to take a history from patients and as manmachine communication becomes faster and therefore easier,
increasing exposure to procedures of this kind may bring first
familiarity and then confidence. Many patients today rely
heavily on machines of one sort or another for their medical
well being. The day may not be far off when the diagnostic
computer with its wide range of diagnostic programs is
merely another familiar machine in the doctor's office or
outpatient clinic.
The value of the computer yet remains to be proven however in a real life situation rather than in an artificial experimental setting. We need to know more about the incidence
of various diseases in the population and of the frequency of
symptoms and signs associated with them. Some work has
already been undertaken along these lines but much more
needs to be done: collecting data of this sort is laborious
and time-consuming. Moreover the data must be accurate;
but doctors are not especially noted for the accuracy of the
notes which they make about their patients. More too needs
to be known about the diagnostic process; in common with
the process of creative thinking, the art of diagnosis is poorly
understood. Consequently attempts to mimic it on a machine
are, from an esthetic point of view, crude.
There is a relative lack of information on the importance
of the observer error that exists between doctors when two
or more of them attempt to record the same set of clinical
features on the same patient. The relevance of this error to
the usefulness of the information generated by the doctor
during his examination of the patient, and to the validity of
machine diagnoses based upon it, is only too obvious.
Finally some thought must be given to the costs of misdiagnosis. If the doctor or machine misses a case of the
common cold, not too much damage is done. If, however, he
misses an early case of potentially remediable carcinoma, the
outcome is disastrous. At present no one has produced a
practical diagnostic system which takes account of the risk of
misdiagnosis of one condition compared with another. Such
systems have been developed from a theoretical standpoint
however; the way is now open for their incorporation into
practical diagnostic computer models.

FI ELD, 1968-1969 - ENTRI ES

Walter Penney, CDP
Problem Editor
Computers and Automation
"You go without me", Pete said as Joe came by to pick
him up for lunch: "I'll get a sandwich at the snack bar."
Joe looked at the scope which showed a circle of dots.
"That must be pretty interesting to make you pass up lunch.
What is it?"
"This is going to be the big attraction at the Summer
Joint Computer Conference. Our booth is going to feature
the new Data-Visi-Scope and visitors will be invited to play
this game I'm working on. The computer will be the opponent and the progress of the game will be displayed on the
DV Scope."
"Game? Not another Nimbonacci, I hope."
"N 0, this game is foolproof. A win or loss is completely
under the control of the machine."
"How are you going to arrange that?" Joe was beginning
to forget about lunch too.
"Well, the game consists of 21 points equally spaced as
though outlining a circle - the vertices of a regular icosaengon, you might say." Pete couldn't resist parading the newfound knowledge he had just acquired from Lambros, the
classical scholar.
"Icosa-shmicosa, how does it work?"
"The player and the machine play alternately, connecting any two points with a straight line. No lines may cross
and each point may be used only once. Whoever can't make
a move loses."
"And you're saying that no matter what the player does the
machine can always win?"
"Yes, or lose if the program calls for it as it will every so
often in order to keep up the interest. We were thinking of
allowing about one win per hour with some small prize for
the player who won."
"I don't think it will work."
Will it?

Solution to Problem 695: Search Research
The minimum average number of steps is 5.72, but a 50-50
split (or as near this as possible) is not the only way this could
be obtained. For example if the 50 had been divided 21-29
and these divided 9-12 and 13-16 respectively, each of these
being split as nearly evenly as possible, this would have
yielded the same average number of steps.
Readers are invited to submit problems (and their solutions)
for publication in this column to: Problem Editor, Computers
and Automation, 815 Washington St., Newtonville, Mass. 02160.


Who's Who in the Computer Field 1968-1969 (the
Fifth Edition of our Who's Who) will be published by
Computers and Automation during 1969. In this edition, we hope to include upwards of 10, 000 capsule
biographies, including as many persons as possible
who have distinguished themselves in the field of
computers and data processing.
If you wish to be considered for inclusion in the
Who's Who, please complete the. following form or
provide us with the equivalent information. The
deadline fqr receipt of entries in our office is
Monday, June 30, 1969. (If you have already sent
us a form some time during the past eighteen months,
it is not necessary to send us another form unless
there is a change of information. )

(may be copied on any piece of paper)
1. Name? (Please print) _ _ _ _ _ _ _ _ _ _ _ _ __
2. Home Address (with Zip) ? _ _ _ _ _ _ _ _ _ _ __
3. Organization ?_:--_ _ _ _ _ _ _ _ _ _ _ _ _ __
4. Its Address (with Zip) ? _ _ _ _ _ _ _ _ _ _ _ __
5. Your Title ? _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __
6. Your Main Interests?
( )
(Please specify)


Year of Birth?
Education and Degrees ? _ _ _ _ _ _ _ _ _ _ _ __
9. Year Entered Computer Field ? _ _ _ _ _ _ _ _ __
10.0ccupation? _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __
11. Publications, Honors, Memberships, and other
Distinctions ? _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __


(attach paper if needed)
12. Do you have access to a computer? ( )Yes ( )No
a. If yes, what kind of computer?
Manufacturer? _ _ _ _ _ _ _ _ _ _ _ _ _ __
Model _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ___
b. Where is it installed:
Manufacturer? _ _ _ _ _ _ _ _ _ _ _ _ _ _ __
Address? _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __

Is your access: Batch? ( ) Time-shared?
Other? ( ) Please explain: _ _ _ _ _ _ _ __
Any remarks ? _ _ _ _ _ _ _ _ _ _ _ _ _ _ __

13. Associates or friends who should be sent Who's Who
entry forms?
Name and Address

(attach paper if needed)
When completed, please send to:
Who's Who Editor, Computers and Automation,
815 Washington St., Newtonville, Mass. 02160


Herbert A. Haessler, M.D.
Medidata Sciences Inc.
140 Fourth Ave.
Waltham, Mass. 02154

((Patient response to automated medical history taking has been quite
favorable. When patients are asked routinely whether they would prefer
to give their history to the machine as they have done; to a physician;
to a nurse; or whether they have no preference, over half the patients
expressed no preference. And of those who did express a p'reference, the
machine was favored over the physician by a margin of approximately
three to one."

Dr. Haessler is the Medical Director of Medidata
Sciences. He is a graduate of Marquette University
School of Medicine and served as a Resident in Pediatrics
and a Post Doctoral Fellow in Internal Medicine at The
Massachusetts General Hospital. For tho past nve years
he has been working in the area of computer-based medical data handling. He participated in the design of· the
Medidata Sciences' Profile 320 Medical History Taking
Terminal and wrote the General Medical History discussed in this article.


Until relatively recently, there have been few serious attempts at either investigating the procedure of medical history
taking or at automating the process even though it is among
the most time-consuming parts of the physician-patient interaction. Not until 1949 did the Cornell Medical Index,l the
common ancestor of most of today's automated medical history systems appear. In this fixed format paper and pencil
questionnaire the patient is asked to answer everyone of
approximately 150 questions. It covers a conventional medical systemic review with added emphasis on the psychologic
state of the patient. The Index has been widely used and has
been automated to the extent that scoring can be done by
machines using mark sense techniques.
Since publication of the Cornell Medical Index, many
other medical questionnaires have been developed for special
purposes. For example, industrial organizations have devel': ,:':~:,-;: ~':':

. '" " .... '

1969 issue of



The winning entry will appear on the cover
of our August issue - more than 25 entries
will be published inside. The 1968 first
prize winner, "Hummingbird", is shown
here at the left.

1. Any interesting and artistic drawing, design or sketch

made by a computer (analog or digital) may be entered
2. Entries should be 'submitted on white paper in black ink
for best reproduction.Col0r entries are acceptable,
but they may be published in black and w1Iite.
3. Each entry should be accompanied by an explanation in
three or four sentences of h~wthe drawing was programmed for a computer, th~'type of comp-q,ter used,
and how the art' was produced by tb~ computer.
There are no formal entry blanks; any letter submitting
and describing the entry is acceptable. We cannot undertake to return artwork, and we ask that you not send orignals if good copies are available.

Deadline for receipt of entries in our office is July 3, 1969.



Source Exif Data:
File Type                       : PDF
File Type Extension             : pdf
MIME Type                       : application/pdf
PDF Version                     : 1.3
Linearized                      : No
XMP Toolkit                     : Adobe XMP Core 4.2.1-c043 52.372728, 2009/01/18-15:56:37
Producer                        : Adobe Acrobat 9.1 Paper Capture Plug-in
Modify Date                     : 2009:03:20 21:26:35-07:00
Create Date                     : 2009:03:20 21:26:35-07:00
Metadata Date                   : 2009:03:20 21:26:35-07:00
Format                          : application/pdf
Document ID                     : uuid:739fe5dc-4c08-418b-9392-b43bb8c52a68
Instance ID                     : uuid:097d697a-67c6-428f-a013-6c9b50af3c5a
Page Layout                     : SinglePage
Page Mode                       : UseNone
Page Count                      : 56
EXIF Metadata provided by EXIF.tools

Navigation menu