NATIONAL INSTITUTES OF HEALTH
DIVISION OF RESEARCH RESOURCES
BIOTECHNOLOGY RESOURCES BRANCH

RESOURCE IDENTIFICATION

 

Report Period: Grant No.
From: August 1, 1972 To: July 31, 1973 RR - 311 = O6S1

Date of Report Preparation:

 

 

May 1973
Name of Resource Resource Address Resource Telephone No.
Advanced Computer for Medical Stanford University (415) 321-1200 Ext. 6121

Research (ACME) School of Medicine
Stanford, California 94305

 

 

 

 

Principal Investigator Title Academic Department
Lederberg, Joshua Professor Genetics
Grantee Institution Type of Institution investigator's
Telephone No.
Stanford University Private Non-Profit
University (415) 321-1200 Ext. 5801

 

 

 

Name of Institution's Biotechnology Resource Advisory Committee:

ACME Policy Committee

 

Membership of Biotechnology Resource Advisory Committee:

 

Name Title Department Institution
*Elliott Levinthal Assoc. Dean for Research Genetics SU Med School
Stanley Cohen Associate Professor Clin. Pharmacology "
Joseph DeGrazia Assistant Professor Nuclear Medicine "
Donald Harrison Professor Cardiology "
Summer Kalman Professor Pharmacology "
Howard Sussman Assistant Professor Pathology "
Name and Title of Principal Investigator: Signature Date

 

Joshua Lederberg, Prof. of Genetics

 

Name and Title of Grantee Institution Official Signature Date

 

 

 
I. SUMMARY

A. Brief Evaluation.

The ACME facility is now approaching the end of a six-year grant plus
a one-year administrative extension. The ACME experiment has proven highly
successful as demonstrated by the following points:

1. Teachability:

Medical researchers have been taught to do their own programming for
non-trivial tasks. More than 1700 persons on the Stanford medical scene
have been trained in the use of computers.

2. Strong Educational Tool:

This facet of ACME encourages many persons to become involved in
computing.

3. Data Acquisition:

The ACME system combines moderate rate data acquisition service with
timesharing. A relatively sophisticated group of realtime data acquisition
users has been developed.

4, User Community:
More than 210 user projects, exclusive of ACME staff, are current users

of the system. They enter the system from 55 terminals spread throughout the
Medical Center.

5. Programming Effort :

The ACME disk packs and tapes hold programs representing over 250 man
years of programming effort.

6. Publications:

A list of recent publications by ACME users and an index of ACME Notes
prepared by ACME ataff are presented in Section VIII.

T. Dedicated Systems:

Several groups are now using dedicated computer systems which reflect
an outgrowth of pilot projects performed on the ACME facility. We observe

& propensity of large or clinical projects to become autonomous from large
central facilities.
The ACME experiment was initiated nearly seven years ago to provide
a timesharing system on an IBM 360 hardware system concurrent with some
realtime data handling. Applying hindsight to the choices made we can
now see that our successes noted above are mitigated by some trends which
were not predicted by us in 1965. Specifically, technological and price/
performance changes have occurred since 1965 which make mini-computing
systems dedicated to specific tasks much more competitive with centralized
resources than was true in 1965. Based upon our experiences, we at Stanford
Medical Center are moving to establish an improved large central computing
system while, at the same time, the number of dedicated mini-systems is
steadily increasing. In other words, the clear advantages of a large
central system for certain applications are offset by other goals in a
number of realtime and control situations where dedicated mini-systems
offer the only realistic solution.

Some of the more significant changes over the past seven years have
included the following:

(1) Both logic and, somewhat later, main memory units have dropped
in cost by 1 to 2 orders of magnitude for devices of roughly
equal performance.

(2) Disk systems have increased in capacity, speed of access, and
reliability while costs have dropped markedly. The figures
below demonstrate this trend.

 

Hardware Devices Speed Capacity MByte/Month
1. Core or semiconductor 200 x 1079 8MB $5 ,000
2. Fixed Head 2 to 5 x 1073 22MB $500-1 ,000
3. Moving Head 26 to 30 x 1073 ~— BOOMB < $10

DISK PERFORMANCE DATA FOR IBM HARDWARE*

 

Item 1956 1966 1970 1973 Est. 1977-80

1. Capacity 5 233 800 800+ 3 x800
2. Bits/Inch 100 2200 Loko 5600 4 x 5600
3. Tracks/Inch 20 100 192 300 600 to 900
hk, RPM 1200 2400 3600 3000
5. Data Rate 9.7KB 312 806 885 4 to 10 x 885
6. Disk Coating

in micro inches 1000 100 50 50
7. Flying Height

in micro inches 1200 100 50 50
8. Gap Width 1000 100 50 50

* Data received in IBM presentation in San Jose, California held in
May, 1973.
(4)

(5)

The prices for mini-computers have dropped an order of magnitude
while capability and level of software support have increased
Significantly. Also, the variety and flexibility of adding
various peripherial devices has risen sharply.

Communications terminals and supporting equipment has improved
in terms of speed and reliability.

The user community has grown in numbers and level of sophistication.
It also demands increased availability and reliability. Response
times must now be measured in fractions of seconds rather than
seconds which imposes high overhead costs on large time shared
systems.

In 1965 a handful of computer users existed in the Medical Center.
Today, there are more than 200 active research projects on the
ACME system plus about 25 mini-systems in laboratories. Computers
have become accepted for production use in many applications.

The first year and a half of the ACME grant was spent assembling
staff and hardware and developing the PL/ACME system. The result
was one of the first timesharing systems with concurrent realtime
support. The system we mounted is remarkedly easy to learn and
use. Since the system has not been exported, we assume that a
shift back to the mainstream of software systems as provided by
vendors will become necessary. The relative cost of people versus
hardware has grown to a point where " home brew" systems cannot be
afforded over the long term. This situation is regretted since
many of the newly announced systems fail to deliver to the end-
user the convenience and power of our existing system. At Stanford,
the conversion to vendor-supplied systems might be expected to
occur in three or four years.

The series of developments in mini-systems has relieved some bur-
dens but created new needs for small machine support from a central
system. Stanford expects to expand upon the currently available
intercommunication systems. Since mini-systems are frequently
being used as data collection controllers, the development of
shared data base systems on the central machine requires inter-
machine communication capability.

We have offered graphics services on the ACME system over the past
five years. There are now six graphics CRT's plus five hard copy
plotters attached to the ACME system. The growth in graphics usage
has been slower than one might have expected. We suspect that this
reflects a lack of description tools, the relatively high cost of
graphics terminals, plus the high cost of running graphics soft-
ware. Perhaps, the growth spurt will come in a few years after
the costs are dropped another factor of 5 or so, and description
tools are improved.

Our dependency upon NIH in establishing a user community and
developing a financial base with which to support computing ser-
vices warrants special mention. It is clear to me that the
venture capital to establish a system, train users, and form

a critical mass of support could not have been raised by incre-
mental growth through charges to individual research grants.

A central facility development grant, such as ACME grant, is the
only available counterpart to venture capital in a cash-accounted
grants system. This investment has now paid off: We have a
common language in use by nearly every department in the Medical
School, a cadre of trained programmers, and a strong momentum

in the direction of shared data bases and shared programs.

Some continuing incremental development effort will be needed

to prevent atrophy.

The passage of time has brought about needs for realtime support
systems which exceed the capabilities of our local ACME system
by one or two orders of magnitude. Rather than attempt to build
realtime systems which IBM hardware has not been designed to per-
form, we prefer to rely on alternative systems specifically
designed for this use and build improved communications into

the central site. Further efforts will therefore constitute
another generation of system planning, based on vendor-furnished
modules.

We have all heard of TSS, TSC, TORTOS, CPS, CMS, and other time-
sharing systems built on IBM hardware systems. It is a credit
to the small staff which built PL/ACME that their system can
compare so favorably with the other systems which have clearly
had far more effort spent on development. The use of PL/ACME as
a research tool by so many local groups is a tribute to the system
designers and implementers. Gio Wiederhold deserves special
credit as the principal creator of the system. He would also be
the first to point out some of the design features that might
have been improved with the benefit of hindsight. The lists of
publications and technical notes appended to this report attest
to the productivity of the ACME Facility staff and to the effec-
tiveness of the tools provided to the user community.

J. Lederberg
B. Highlights of FY73,
1. Planning for the Future.

During the past twelve months the ACME computing facility has passed
through an identity and existence crisis. The sizeable effort expended
by faculty, Hospital staff, and ACME facility staff has led to a decision
to maintain the PL/ACME system on new hardware. The computer services for
Hospital administration and ACME time sharing and realtime data acquisition
services are being merged onto a new facility to be installed in August
1973. Numerous studies and presentations have been required to bring about
the decisions which make this possible. This subject is discussed in detail
in Part C of this section and in Section II.

a. A Generalized Time Oriented Database System.

Special attention should be focused on the transition which has begun
to occur at Stanford with respect to faculty attitude toward the need for
sharing of data. The awareness of shared database concepts has increased
markedly. Evidence of this can be seen in the teamwork demonstrated in
preparation of a health care resources research proposal. Other evidence
can be seen in the attendance at seminars concerning the Time Oriented
Database system (TOD), developed this year by the ACME staff in conjunc-
tion with Dr. James Fries of the Division of Immunology. For more on this
subject, see Section V.

3. Software.

System programming development activities during the year resulted
in new data compression routines, file system improvements, mounting the
COBOL compiler, studies and planning concerning the new VS2 system announced
by IBM, support for the small machine multiplexor, and a PDP-1l simulator.

h, Hardware.

The small machine multiplexor was completed, allowing for inter-machine
communications. Other hardware projects included work on terminal light-
boxes and several new interfaces for users of the 1800 and the multiplexor.
ACME acquired several 300 and 1200 baud terminals during the year. In April
1972, we installed a Memorex terminal controller which has performed very
well.

5. Core Research.

Support of core research and development effort included programming
and computer service support for the DENDRAL project, assistance for the
Drug Interaction project, direct support of the initial application of the
Time Oriented Data (TOD) system, extension of small machine support to
GC/MS activity, a joint development effort on communication hardware
development, and a core project to develop new statistical analysis
techniques.
6. Utilization.

Utilization of the ACME system in terms of terminal hours has remained
relatively constant during the past year. One exception to this is the Drug
Interaction project in the Pharmacy which used ACME extensively through
February 1973, after which time the system was moved to a dedicated dual
mini-computer system.

Since the follow-on to the ACME system was not resolved until March
1973, the rate of new user signups has dropped from normal levels and there
have been essentially no new realtime users of the system. This is most
understandable since many users felt that ACME might not survive beyond the
end of the grant period. It is noteworthy that the user community has con-
tinued to use the system in the absence of (prior to March 1973) any Medical
Center commitment to retain the PL/ACME system beyond July. Now that such
assurance has been made, new users are again expressing interest, evidenced
by the number of signups for the introductory classes in use of PL/ACME.

[. Minicomputers.

Other computing activity in the Medical Center includes the acquisition
of several minicomputer systems for various research and production projects.
Approximately thirty minicomputer systems are currently used within the
Medical Center. Some of the applications include data acquisition for mass
spectrometers, operation of the Drug Interaction programs, an information
system for the Clinical Laboratories, and research support in Nuclear
Medicine, Chemistry, Psychiatry, Cardiovascular Surgery, Cardiology, and
other divisions and departments. The growth and number of minicomputer
systems used for instrumentation control and data collection have pushed
the central facility to provide small machine communications and other
support activities.

8. Documentation & Conversion.
Throughout Fiscal 73, the staff has spent a great deal of time on
documentation of the existing ACME system. Since the decision to move to

a merged/158 facility, the conversion effort has been of central importance.

Cc. Planning and Reorganization.

 

l. Stanford Center for Information Processing (SCIP).

In the past Stanford operated five major service computing organiza-~
tions, each of which had its own loyalties to a specific user community.
The five were:

Stanford Linear Accelerator Center
University Administrative Computing Facility
Campus Facility

ACME Facility

Hospital Data Processing Facility
- T-

Computing at Stanford University was reorganized during the spring of
1973. The new organization, entitled "Stanford Center for Information Pro-
cessing (SCIP)", provides a unified structure for the five facilities
mentioned above. ACME and the Hospital Data Processing Facility will be
combined to form the Medical Center Computing Service (MCCS). The heads
of all the facilities will report to the director of SCIP. Along with the
reorganization of staff involved in managing the various computer facilities,
the policy committee structure comprised of faculty members is currently
being modified.

2. The Medical Center Planning Effort.

A description of the computer planning activity at the Stanford Medical
Center over the past eighteen months would fill many volumes. Several
different faculty committees and staff groups have reviewed alternatives
ranging from highly distributed interconnected minicomputer systems to
highly centralized large computing systems. The issues faced by the various
groups were:

1. Should PL/ACME service be continued?

2. Can the ACME users provide a critical mass of dollars required
for a stand-alone facility.

3. If a merger is required, who should be the parties to the merger?

4. Are the potential advantages of a shared database between Hospital
and Medical School strong enough to outweigh the disadvantages of
merging a production system with a research support system?

5. Should realtime computing services continue to be provided from
@ central computing source?

6. How should computing at Stanford University and the Medical
Center in particular be organized?

7. What computing services will be needed over the next several years?

8. How can we relate Medical Center computing planning to broad
University goals?

3. The 370/158.

These are among the many issues which have been considered during the
past year and a half. The solution selected entails the installation of an
IBM 370/158 hardware system using IBM's newly announced VS2 software system.
The services to be offered will include batch services in several languages,
time-sharing using PL/ACME, realtime data acquisition services using the
existing 1800 system, normal consulting and user services, and small machine
communications.
The current schedule calls for removal of the 360/50 system from
Stanford on July 28, 1973. A number of peripherals will be moved to the
new 370/158 site where systems programmers will have approximately 2-1/2
to 4 weeks to bring up the new system. We expect to resume PL/ACME services
for terminal users by September 1, 1973; realtime services will hopefully
be available approximately one month later.

Funding for use of computers within the Stanford Medical Center is
expected to drop over the next eighteen months due to cuts in federal budgets
as well as escalation of costs within fixed budgets. A tight dollar economy
coupled with multiple options for the users (e.g. outside time sharing service,
Campus computing facility, more powerful dedicated mini systems) will force
the new Medical Center Computing Service to perform very well to attract the
business of the Medical Center commuity.

4. The SUMEX Proposal.

A proposal has been submitted by Dr. Lederberg calling for the formation
of a Stanford University Medical Experimental Computing Facility (SUMEX). If
approved, this proposal would result in the acquisition of a PDP~10 to support
a national facility specializing in tools for the development of artificial
intelligence in medicine (AIM). The ACME experience has been invaluable in
demonstrating both the opportunities and the problems of community-shared
resources. In particular it has given us the technical expertise needed to
design realistic specialized instruments to serve geographically dispersed
but intellectually convergent users.

dD. Overview of ACME Experiment.

The ACME experience indicates that a large central resource can provide
a@ very valuable service for users requiring text editing, numeric calculations,
statistical analyses, and realtime data acquisition at relatively low rates.
Our experience has also demonstrated that a large central facility should not
undertake high data rate realtime data acquisition and closed loop control
functions if it intends to service a large number of time sharing users con-
currently. In addition we have learned that an extensive amount of "hand-
holding" is needed to serve the research scientists in a medical community.
This may change in the future when MD's will routinely receive more training
in computer science in the course of their college educations.

ACME's initial proposal included the following paragraph concerning
hardware selection and resource allocation:

"The IBM/360-50 has been selected for the initial realization of
ACME (1) as a machine technically appropriate to the immediate tasks
in mind and (2) for its system compatibility with the 360-67 already
selected for the eventual replacement of the 7090 by the Stanford
Computation Center. The 360-50 will be installed in ACME May 1966
and will run on three shifts under Operating System/360, subject

to review by the policy committee. These will be dedicated respec-
tively:
(A) A prompt access time-sharing mode - perhaps over most of
the working day.

(B) A scheduled, full-use, on line mode - to service development
work on high data rate and on line control applications, and
for similar systems development.

(C) Job-shop, especially longer runs for which overnight turnaround
is acceptable, and which cannot be serviced with comparable
effectiveness by SCC."

The following aims were added to the ACME charter at the time of the
Renewal Proposal in the Spring of 1969:

1. To improve hardware and software reliability for the benefit of the
medical users.

2. To provide small machine assemblers in PL/ACME so that code for small
machines can be written from an ACME terminal.

3. To achieve over time a state where income from user charges will match
operational costs for the ACME system.

All of the original objectives have been achieved to varying degrees
of satisfaction. Of special note is the development of PL/ACME as an inter-
active time sharing system which can be easily learned and used by medical
staff. On the other hand, the realtime support offered is inadequate due to
system instabilities and data rate limitations. Access to Campus Facility
is inconvenient for ACME users.

In terms of the items added at renewal time, hardware and software
reliability have been markedly improved. Small machine assemblers have
been added, but the user must write code in the assembly code for whatever
satellite he intends to run. At present, assemblers of this type exist for
PDP-11, PDP-8, and 1800. The income of the facility has been rising steadily.
Economic overlaps with NIH direct support for ACME have blurred the transition
to totally non-subsidized use. A major rate increase was initiated in April,
1972. With this change, income over the last 12 months reached roughly 55%
of direct operating costs (exclusive of development efforts). From the
vantage point of hindsight one could well ask whether the selection of the
360/50 hardware and the decision to promote a large central time sharing
and data collection resource were appropriate. Given the availability of
new third generation hardware and the promises of IBM or expectations of
its customers in 1966, the 360/50 hardware selection is defensible. However,
the development of low cost, fast, well-supported minicomputers was not
anticipated to proceed at the phenomenal pace that it has. This major
technological shift has strongly influenced our present thinking for the
future of computing in medicine and related research. The role of a large
shared resource has by no means been obviated by the minicomputer revolution.
We will continue to need powerful facilities beyond the scene of current mini
architecture.
-~10-

II. STANFORD MEDICAL CENTER COMPUTING PLANS

A. The Current Scene,

Between January and April 1973, the following significant events
occurred in the computing environment affecting the Medical Center:

1. The University reorganized the service computing management structure
to form the Stanford Center for Information Processing (SCIP). The SCIP
organization will manage and operate all major service computing functions
for the University.

2. The Board of Trustees authorized acquisition of an IBM 370/158 system
to service the needs of the Medical Center.

3. Personnel from the ACME Facility and Hospital Data Processing Facility
were assigned the task of converting the current systems to the new hard-
ware systems.

4, Users were notified of the changes scheduled to occur between July
and December 1973.

5. Planning of new faculty advisory groups for computing throughout
Stanford was done.

PL/ACME users had been warned that the time sharing service might
have to be disbanded at the close of the ACME grant. Therefore, the ACME
community was elated by the above series of decisions. Medical School
faculty and Hospital management were notified of the scheduled changes by
@ memorandum from Mr. Victor Barber dated April 23, 1973. A copy of this
memorandum has been reproduced on the pages which follow.

B. Shared Database Planning.

The need for planning of shared database effort is presented in
Appendix A. Given the need to which these memoranda attest, it is likely
that the central facility will assign key personnel to work on the problem
along with interested researchers. The near term development effort is
likely to be based on use of the Time Oriented Database (TOD) system.
Further information on TOD is presented in Section V of this report.

C. New Faculty Appointment.

A selection committee has nearly completed its deliberations with
respect to a new faculty member in the Medical School who will have
considerable responsibility for policies affecting computer services.
The new position will be located in the Department of Community and
Preventive Medicine. It is hoped that the new appointee will serve as
a focus and spearhead for development activity in the shared database
area.
From

Sussect

-ll-

Date: April 23, 1973

op Pras

V. H. Barber, Associate Director, Medical Center Computing Service
Stanford Center for Information Processing

Distribution

Computing Services for Medical Center

Stanford University announced in March of 1973 the complete reorgani-

zation of its general support computing facilities. The new organiza-
tion, Stanford Center for Information Processing (SCIP), is described

in the attached press release. The result of the reorganization pro-

vides a large general support computing facility in the Medical Center
environment that merges the services of the medical research community
with the business, administrative, and patient care activities.

My role as Associate Director of the Medical Center Computing Service

is to serve Medical Center users and represent their interests in the
service computing arena. Our goal will be to provide the required com-
puting services at the lowest possible cost. A new hardware system will
be available to serve the entire Medical Center in September 1973. The
new facility will have more than three times the compute power of exist-
ing service facilities and will make available improved services during
the year as the power of the system is harnessed with associated soft-
ware. These will include shorter response or turn-around time and
sharing of data bases.

A list of service goals for the MCCS is attached as Appendix A. Addi-
tional needs of the medical community will be established through inter-
action with the Hospital management group, individual users, and faculty
committees,

We intend to provide easy communications between you, the user, and the
staff of the new facility. We want a highly personalized service that

is responsive to the needs of the medical community. Madhu Bhide,x5151,
will be the primary liaison and coordination point for Hospital services,
especially those oriented toward financial applications. Ms. Karen
Richards, R.N., will continue as the Nursing Service Coordinator for com-
puting matters. She is available at x6084. Ron Jamtgaard, x6121, will
be the primary contact for users of timesharing services and realtime
support; he will respond to needs of the Medical School research and
education functions. B. J. Gaul will be Operations Manager of the
370/158 computing facility. He is available at x5880.

Ron Jamtgaard and his staff will be housed in the old ACME offices
(Tc101, temporary building). All other MCCS computing and management
personnel will be located in the Administrative Services Annex, just
north of the Medical Center, in the old Hospital Data Processing area.

GYOUNVILS e

ALISSTAING

WNONVYOW3IW 3D1ddO © ALISHJAINA GYOINVLS © WNGNVYOWIW ID1IO e

WNONVSOWIW J3D1ddO © ALISYFAINNA GYOINWIS
-~12-

We plan to serve you in the following manner: The new hardware facility
(IBM 370/158) will be available in September 1973 to serve the reasearch and
development interests of the Medical Center; by December 1973, the
business and finance computing will be merged onto the new system. The
services which will be offered when the facility opens include timesharing
(using the PL/ACME language) and batch services in those languages for
which a user need exists. Initially, batch services will be provided for
FORTRAN, COBOL, PL/1, and LISP. Services to the business and finance
community will continue as before; however, larger resources will be
available, and there will be new opportunities for service.

The transition from PL/ACME on the current Model 50 to the 370/158 will be
transparent for the terminal user except possibly in some realtime areas.
Our target is to hold service interruption times to a minimum. Standard
terminal services will likely be unavailable for about four weeks; realtime
services may be disrupted for six to ten weeks. Digital realtime data
acquisition services and graphics via the IBM 1800 as well as small machine
communications will be provided by the new facility as soon as possible.

The transition of Hospital services should be completed by December 1973.

The facility will be operated on income received through user fees. Our
goal will be to provide maximum service at the most cost effective rates.
Further policy on fees will be developed and released in the near future.

Persons who are new to the Stanford medical computing scene are encouraged
to contact me at x5998 so that staff can be assigned to assist in definition
and solution of computing needs.

MCCS exists to serve you and your computing needs. We hope to hear from many
of you regularly.

ee: Medical School Faculty
Hospital Department Heads
ACME Users
SCIP Associate Director
C. Rich

- Carpenter

. Gonda

Hofmann

Dickens

Roberts

Jamtgaard

BmavHyy
NST

-13-

MEDICAL CENTER COMPUTING SERVICE
FUNCTIONS, SERVICES, AND GOALS

 

valient Accounting. Hospital financial and administrative services

and patient accounting services. These services involve chiefly patient
accounting, patient billing for Stanford Hospital and Clinics, accounts
receivable, third-party allocations, payroll, personnel, general ledger,
accounts payable, census, financial and budgetary analyses for the
Stanford University Hospital.

As part of the patient accounting services, MCCS manages a large patient
financial data base. It is expected that this will form a nucleus for

a comprehensive patient data base in the future. One of the services
that the service facility will offer is building on and managing this
patient data base for both the Hospital administrative staff and medical
research personnel interested in the patient data base.

Interactive time-sharing -- PL/ ACME. PL/ACME is an easy-to-use time-
sharing service. The major user of this service is the medical community
at Stanford University. It is expected that the proposed facility wil]
continue to support PL/ACME in its present form and gradually enhance the
service to satisfy future requirements.

 

Data reduction and data control services. Currently a well-trained staff
of keyboard operators and data control personnel perform these services
for the Stanford University Hospital. This service will continue to be
performed, and the recipients of this service will expand from Hospital
financial data processing users to include other medical personnel.

 

Realtime services. Realtime service is currently provided by an TRM 1800
and ACME-built interfaces in the laboratories. The 1800 is programmed as
an integral part of the system and acts as a 360 control unit. These
services are currently used routinely by a number of investigators.

Small machine communications services, A multiplexor (MUX) for connec-
tion of mini-computers to the new 370/158 will be installed in 1973,
after more routine services are operational.

 

Data collection. The new facility hopes to collaborate in the jevelopment
of data collection systems. An example would be the development of an
automated Patient Admission, Discharge, and Transfer system. Opportunities
in this field will be vigorously pursued.

» Language support. Other such collaborative efforts are foreseen in the

area of language support. One such example is the MUMPS language. This
lansuage was developed at the Massachusetts General Vospital. Tt is used

in conjunetion with PDP-11 and PDP-15 computers. The llospital uses a PHP-1l
MUMPS system for the Pharmacy Drug Interaction Project. ‘The facility role
is not clear; we are open to new ideas here,
Ww

10,

ll.

- 1h -

Liaison with Forms Management. MCCS will work closely with the Yospital
Forms Management Section. The responsibility of the Forms Management
Section is to coordinate all the forms that are used at the Stanford
University Hospital. It takes on the responsibility of designing, print-
ing, ordering, and stocking of forms. This service is expected to con-
tinue and will closely interact with the information flow development

at the Medical Center.

 

Vrogramming and consulting. The proposed facility will offer programming
aud consulting service. These programming activities include fee-for-
service programming for users, design and development of production systeme.
and maintenance of public utility programs as well as existing projuct.ion
systems, Additionally, the facility will offer services in the areas of
procedures analysis and automation of procedures. One of the current ana-
lytical services in which we are participating is an automated work-mea-
surement study. [t is expected that the proposed t'acility will continue

to participate in such studies and offer services in these areas.

 

Library and grant assistance. Assistance in the identification of funding
opportunities, proposal preparation, and management of grants that include
the use of computer facilities will be available.

 

A library of current reference publications in the area of computer use in
the health care field will be maintained.

Educational activity. The facility is expected to be very active in con-

 

tinuing its current educational activities, specifically in teaching ACMEF's
interactive and timesharing usage, as well as education of nursing and
other Hospital staff in automated procedures. It is further expected that.
the proposed facility will extend its activities in continuing cducation
of llospital staff in data processing procedures, systems desiygm, and data
base management, and that it will also be active in the area ot current
awareness and dissemination of information to physicians and other medical
personnel. As an adjunct of this, it is expected that the proposed facil-
ity will have an internal awareness program to keep its staff abreast of
development in health care technology as applied to a computer service
facility of a major medical center.
-15-

D. Some Observations on Computer Planning.

It took the Medical Center and University eighteen months to perform
the planning activity leading to a decision for a course of action. A
chronology of this period is attached as Appendix B. By scanning the
chronology one can quickly observe that organizational and technical
issues involving computing become quite complex and require extended
timeframes to complete. Some of the major policy issues addressed by
the various study committees included the following:

1. Do we want a highly centralized computing environment or do we choose
a distributive minicomputer system with some inter-machine communications?
Would some middle ground between these two choices be most appropriate?

e. Can we successfully merge the research support computing of the
Medical School with the business and finance data processing of the
Hospital?

3. What advantages might be gained by merging with the central Campus
Facility of the University?

4, What investment does the PL/ACME user community have in the PL/ACME
system and language? How easily could they be converted?

5. How can we fund computing for medical students and researchers on
the faculty?

6. What computing needs are likely to dominate over the next five years?

These are some of the questions which the various committees have
addressed.
- 16 -

III. ACME FACILITY ACCOMPLISHMENTS - FY73

Accomplishments of ACME staff personnel are described here; core
research projects led by faculty members are included in Section V.

The primary accomplishment of the ACME facility during the past year
has been to hold its user commmnity largely intact during a period when
the future existence of PL/ACME services was highly in doubt. The doubt
stemmed from the fact that PL/ACME services had been subsidized by the
ACME NIH grant and that the paying users did not constitute a critical
mass to afford a facility which could duplicate these services.

A. Planning Studies.

Since October 1971 several members of the ACME staff have been actively
involved in planning methods to continue offering PL/ACME services beyond
the period of the ACME grant. There follows a list of some of these studies:

1. Merger of Hospital ADP and ACME facilities on a 360/65, PDP-10,
or 370/158.

2. Merger of University Administrative Computing Facility, Hospital,
and ACME facilities on a 370/158.

3. Merger of Campus Facility and ACME on a 360/67.
4, Conversion effort to mount ACME on various systems.

>. User surveys to determine user plans if PL/ACME services were
dropped.

6. Specification of users needs.
7. Review of potential need for time oriented database sub-systems.

8. Consideration of various organizational alternatives. The
results of most of these planning studies have been reported
in earlier sections of this report.

B. Time Oriented Database Development.

One of the major tasks of the ACME applications staff during FY73 has
been the generalization for ACME users of the Time Oriented Database system
originally designed by Dr. James Fries of the Division of Immunology. A
lengthy description of this system (TOD) is included in Section V of this
report.
-~17-

Cc. New and Continuing Applications Programs.
1. DENDRAL:

Support for the DENDRAL project during this fiscal year has consisted
of machine services both in interactive PL/ACME and batch LISP. Early in
the fiscal year an overnight version of batch LISP was mounted so that jobs
could be entered from terminals in the daytime and run when the PL/ACME
system was not needed. In addition the LISP interactive compiler was
markedly improved. The small machine multiplexor and other small machine
support has found limited use in the DENDRAL area. In addition the Loma
Linda graphic displays have been fully incorporated into the DENDRAL
closed loop control problem.

2. Drug Interaction Project.

This has been the year of transition for the Drug Interaction Project
from the ACME system to a dedicated dual PDP-11 system. The new software
written in MUMPS is now operational. This project has served as a classic
example of how a new idea is formulated by faculty, tested under pilot
project status on the ACME system, proposed for a research grant, and
finally implemented in a production form. Many of the computing applica-
tions in the Stanford Medical Center have followed this course of action.

3. Medical Student Admissions.

Programming is now being done to handle medical student admissions
needs. The system will assist the Admissions Office in screening applicants
and provide administrative support.

4, Time Series Data Analysis.

Last summer ACME helped to support the work of Dr. Will Gersch of
the University of Hawaii who used the ACME system to develop an automatic
decision procedure to calculate spectral density estimates. The result
of this effort is now available to all users in the form of a public
program.

Be Radioimmunoassay Programs.

A number of FORTRAN programs written at NIH have been moved to the
PL/ACME system to support research in radioimmunoassays.

6. New Realtime Users.
Very few new realtime users were recruited during the past year.

This reflects the doubt in the minds of many concerning the future of
PL/ACME realtime services. Two projects which were implemented:
~ 18 -

Dr. Don Perkel's project involved analog/digital processing
of two to four channels of nerve impulses recorded during
swimming of the leech. This is a study of nervous control
of movement.

The second project, headed by Dr. P. Sokolove was a study of
the role of the nervous system in production and maintenance
of circadian rhythms (data consisted of nerve spikes and EEG
records).

Note: Detailed descriptions of Items 1 and 4 above are presented
in Section V.

D. System Software Improvements.

In addition to its considerable planning effort, the systems staff
incorporated a number of improvements into the system during 1973. Some
of these are listed below:

1. Satellite Machine Support.

The primary effort here involved mounting the small machine milti-
plexor. This entailed software in the 360/50 as well as small machine code
to test the hardware. The new small machine multiplexor can accommodate up
to sixteen satellite machines serially passing data to or from ACME (See
Appendix C). In addition a PDP-1l simulator was completed. The simulator
can be operated in batch or interactively.

2. LISP.

Two tasks were undertaken for LISP users. The first was to mount an
overnight LISP batch service to which jobs could be submitted from terminals
during the day. The second was to improve the response time of the inter-
active compiler.

3. File Support.

The primary improvement was a data compression routine which permits
users to file their data in a compressed format. For some users, such as
those using the time oriented medical record files, this feature permitted
a factor of five savings in storage costs. In addition the file system
was documented extensively during the last year.

4. Reliability.

A number of bugs were found and fixed. It is a credit to the system
staff that we now operate three to four months between system crashes due
to software.
~19 -

5. Other System Tasks.

The number of terminal ports was expanded from 32 to 8. Accounting
programs were modified to capture data at hourly intervals. Batah account-
ing was added to the system. Release 20 of OS was implemented. A COBOL
compiler was mounted for batch running. The file system directory was
rewritten. A terminal survey was conducted to determine the best terminals
for ACME to support in the future. A hardware monitor (called the SUM
monitor) was attached to the system by Lee Hundley for a series of system
measurements. Special programs were provided for ACME-to-OS dataset con-
version.

E. Education and Training.

Over the past six and a half years more than 1700 members of the
Stanford Medical community have been trained in the use of PL/ACME. This
number includes only those who have enrolled in a formal training class.
During the past twelve months, nine introductory classes have been offered
with a total enroliment of 84. This is less than the normal annual enroll-
ment, primarily due to the fact that classes were not held during three
months of the year when the future of ACME was unresolved. Of the 84
persons enrolled, 23% had a Ph.D. or M.D. degree, another 45% had a Bachelors
or Masters degree, and 32% indicated no degree. ‘Two years ago the corres-
ponding percentage of Ph.D. and M.D. participants was 35%. Approximately
one-haif of those who signed up for the introductory class had no prior
programming experience; most of the balance had only slight experience.
When asked why they sign up for the course roughly 60% indicate their
intent to use the computer for numeric calculations and statistical
analyses; approximately 20% plan to handle large data files; 10% indicate
an interest in realtime applications; the balance require graphics dis-
plays and text editing. More than half plan to participate in a project
currently using ACME; about 20% intend to start a new project.

It is interesting to note that 100% of the participants report that
they have access to an ACME typewriter terminal. There are currently 55
terminals in our network.

In addition to the introductory course, ACME staff have prepared and
offered seminars and advanced classes. The seminars have dealt with general
medical computing topics. A special series of seminars was held concerning
time oriented database work.

Some persons learned to use ACME without enrolling in the formal
classes. One might estimate that 30% or more of the current users did this.
An aid to the persons who preferred to learn by doing is a new program on
the public file called "TEACHER". This is a question-and-answer course
designed to be used from an alphanumeric CRT terminal.
~ 20 -

F. Hardware Changes.
1. Multiplexor.

The satellite machine multiplexor is the most complex and costly
equipment item designed and fabricated by the engineering group during
FY73. The device, which can connect up to 16 satellite machines to the
ACME system, provides a data path to and from the 360/150 in a demand-
response mode. Data rates of 40,000 bytes per second are available using
4 twisted pair; rates of 250,000 bytes per second are available using
coax. The multiplexor is connected to the IBM hardware through a 16-byte
parallel data adaptor on the 2701 which in turn is connected to a selector
channel (See Appendix C).. Only three computers have been connected through
the multiplexor. Additional customers will not be urged to connect until
the new 370/158 system and its saftware become operational.

2. Terminal Controller.

All terminals are connected to the ACME system via a Memorex 1270
terminal controller which was installed in April 1972. This device has
performed very well. It has 32 ports capable of automatic speed recogni~
tion up to 1200 baud.

3. Standard Analog Interface Card.

A new standard analog card has been developed for use in the laboratory.
Since the 1800 is being moved to a distance 2000 feet further from most
users, we will be encouraging the user community to convert to digital
signals at the laboratory end rather than in the 1800. As a result, the
new standard analog card is not likely to be used extensively.

4. Lightbox.

The standard ACME lightbox which has been used on IBM 271 typewriter
terminals was designed to operate off the 2741 power supply. Our shift to
G.E. Terminet 300 terminals and Beehive CRT's has made the lightbox unusable.
A new one has been designed and will be placed in use in June 1973.

5. Interfaces for Users.

The engineering group has designed and maintained a number of inter-
faces for user instrumentation. In general these are paid for directly
by the customer. Some of the devices interfaced during the past year
include scintillation counters, a paper tape reader-punch, and a Houston
plotter.
-21-

G. Operations.

The annual average meantime between failures due to all causes
(hardware, software, power, and human) reached a new high in FY73:
87.7 hours. A chart presenting additional information on meantime
between failures is on the following page.

ACME's Operations Manager, Charles Class, spent a great deal of
effort on planning support for the 370/158 facility, providing assis-
tance in the areas of hardware, physical space and communications.
Here at ACME he was active in the work of installing the new 300 baud
terminals.

Mr. Class was the ACME representative in Co-op, an organization
formed by the operations managers of the University's service computing
facilities to increase communication and cooperation among the several
operations groups.
I°Tg
9° TL
L°08
H°ne

esei1say
Tenuuay

- 22

T’OeT

Cet

 

g° HT
e* 49

aBeisay
Tenuuy

"E€L6T Trady
ysgnorm} ZL6T ysn8ny ‘syquou sutu 4satz uodn peseq emm{TBy 07 suTy ues pazyoeforg (ECL6T SATNLC - LEW) »

*‘zeak yore Jo potazed ames of paredmoo sv amNTTe@y 07 SWTY Uva 4Yseagq = SamnsTy peutTsspuy

‘susqsfs zay10 IO STT-ddd SOO0QT JO saanTTBy YOSTJer you op SsamI3tTj esau,

 

¥L° LQ xL°LQ x L°LQ O'OHT Q°HHT S°TR E°aG €°ey «= a" HLT «2°08 O°Th O'nHe %ELET-2LET

 

 

c°On 0°6E O°On L°9TT 6°61 0°2S co Ot 9°HS §=6 QTEE EL Rent Q°OnT :eLé6éT-TL6T

 

 

0°88 4°08 TLE 2°99 6°08 L* 4S €°ge L'veée O°TSt THs €°ne 0°62 ‘+ TLET-OLET
9° Se 1°09 9°62 O° Le 2°Se 0°6€ O* ht o’er 86S'9e)) «L*LT Z°c€ =0°0h «= OLET-696T

Tor une AoW ady Tey] qed wer oeq AON 20 qdeg any

 

WUVMCUVH *TONI SHYNTIVA TIV

 

 

xT°O?T xT°O2T xT°O2T 8 O°ORT O°TET OTEOT Q°6TT Hel E°eES E€°06 €°gS §=6O* tHe «=F ELOT-2LET

 

9°TS Q°lLS -E° aH O°SLT Q’eHt 2°95 €°09 Zng = on'6E «OO'RDE «OBIT Thhe :2L6T-TL6T

 

 

O°OLT = °EOT «6° gh O'egt O'NGE Tete o'Hr0tT O'HOL OTZOE O'OLT L°HE 4°2h ‘?TL6T-OLET
SE 0°eg 9°66 L°9S G°6E helt e°l9oT O°9S E°TE 3°Se €°SG = EE TT *OLET-696T

Tur une key ady Tey qeg uer 2a¢ AON 20 qdag any

 

HaVMCYVH

(SHNOH NT)
ElL6T *O€ TIudv - 696T SLT XING
SHUNTIVA NAAMLEG ANIL NVGN JO NOSIYVANOD
0S/09€ FNOV
~ 23-

IV. ADMINISTRATIVE ORGANIZATION

The Stanford Center for Information Processing is a new organization
for service computing at Stanford University. The SCIP organization is
shown schematically below.

 

PresiIdant -- R, W. Lyman

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

al {
Provost - W. F. Milter University Committoe on
Computation Facilities -
Assoc. Provost E, J. McCluskey, Ch.
for Computing - G. F. Franklin
Director - C. R. Dickens Executive Committee
Deputy Director ~ M. M. Roberts
Assoc. Director for Technical Planning - T. 0. Phittips Bus/Fin - Creighton
Assoc. Director for Systems Software - E. Williams Med.Ctr. - Carpenter
Assoc. Director for Hardware Operations - M, D. Ray Academic - Franklin
Asst. Olrector for @udget and Accounting - W. H. Yundt SLAC - Taylor
Associate Oirectors
] | I |
SLAC Center CAMPUS Center MEDICAL CENTER ACMINISTRATIVE LIBRARY PROJECT
Center COMPUTING Center
C. Dickens T.0. Phillips Vv. H. Barber M. M, Roberts A. H. Epstein

 

 

 

 

 

 

 

 

 

 

 

 

STANFORD CENTER FOR INFORMATION PROCESSING
(SCIP):

The ACME computer facility is being merged with the Hospital Data
Processing Facility and will now be represented by Mr. Victor Barber as
Associate Director of SCIP for Medical Center computing. Until April
1973, the ACME facility was one of three facilities comprising the Com-
putation Center.

The staff of the ACME facility is listed along with the percent of
full time equivalent effort on the following page. Major personnel
changes which occurred during fiscal year 1973 are as follows:

i. Lee Hundley transferred to the SLAC computing facility where
he will be working on realtime applications.

2. Linda Crouse transferred to the Pharmacology Department as
a acientific programmer.
T.

~ Chow

Rich Cower transferred to the SLAC facility as a computer
operator.

Jane Whitner and Ying Lew were terminated in view of the end
of the ACME grant. The University will consider rehiring
them as future needs develop.

Chuck Granieri transferred to SLAC as a systems programmer in
the spring of 1973.

Russell Briggs was assigned full time to the Drug Interaction
project.

Madeline Aranda, the ACME Secretary, transferred to the
Financial Aids Office.

The balance of the staff will likely be assigned either to new con-
puting facilities within the Medical Center or to other service computing
facilities on the Stanford campus.
25
CURRENT ACME PERSONNEL

NAME % FTE JOB TITLE
Jamtgaard, R, 100 Director
Wiederhold, G. 40 Consultant
Rindfleisch, T. 100 Systems Analyst
Levinthal, E. 18 Computer Planning Faculty Representative
Frey, R. 100 Systems Programmer
Heathman, M. 60 Systems Programmer
Levitt, R. 50 Systems Programmer
Lipkis, J. 75 Systems Programmer
Miller, S. 100 Systems Programmer
Schroeder, J. 100 Systems Programmer
Stainton, R. 100 Systems Programmer
Williams, E. 50 Associate Director for Systems Software
Bassett, R. 100 Scientific Programmer
Germano, F. 100 Scientific Programmer
Baxter, &£. 100 Administrative Asst.
Class, C. 100 Operations Manager
Billger, G. 50 Computer Operator
Sutter, J. 80 Computer Operator
Matous, J. 100 Computer Operator
Rieman, J. 60 Computer Operator
Duffield, A. 80 I.R.L. Support Personnel
Hwang, d. 100 1.R.L. Support Personnel
Pereira, W. 100 I.R.L. Support Personnel
Veizades, N, 100 I.R.L. Support Personnel

Total FTE 19.6