TII. ACME'S FUTURE DIRECTION

projects. In fact, several ACME users have written their own. Since
the data bases have been established independently, the result is a
variety of data base designs and a duplication of information and re-
trieval methods. A future system should include a unified patient data
base, serving all patient care and related research programs of the
Stanford Hospital and Medical School. The data base structure might be
designed as a master file of common patient elements and a number of
subfiles for information unique to a medical descipline. Retrieval might
be along commonly used paths of the data base structure or it may be
unpredictable. That is, the questions cannot be determined in advance.
Consequently, the retrieval mechanism should honor both predetermined
and unpredictable requests. One implementation scheme would incorporate
the generalities of the PL/ACME language with new commands for executing
predefined functions.

c. Realtime Data Collection

ACME service in this area was a pioneering effort at the time it was imple-
mented. For today's technology, ACME's data rate capability is low and in-
capable of handling many applications. The 1800 can accomodate an

ageregate of only 10,000 samples per second, and many users claim diffi-
culty whenever aggregate data rates exceed 2,500 samples per second. Users
must obtain realtime input/output lines from an ACME operator. This service
should be automated. The 1800 is used only as a traffic control device and
not for any pre-processing of data. The Model 50's bulk core is 8 micro-
second core and this is a fundamental Limitation in the speed of the system.
One user in Nuclear Medicine recently proposed purchase of a dedicated

small machine with disk and tape peripherals to handle bursts of 40,000 samples
per second. The role of the "smart" or HIQ terminals applies to this realtime
data collection requirement. ACME must determine the best way to meet the
need for this type of service. A small machine pool may be a solution,
especially if a few of these units can be equipped with disk and/or tape.

4. I/0 Devices

Voice Drum. A private corporation gave Stanford one voice drum. We are
eager to find a potential user of the voice drum so that we can work with
him to develop this hardware. The drum has a limited vocabulary of 50
words. The hardware interfacing the voice drum to the PDP-11 should be
finished by the end of the current fiscal year. We anticipate using voice
output for a warning system or in other situations to call attention to a
typewriter or graphics terminal.

Tape Cassette. We would like to explore the use of the tape cassette for
storage of user data. Users might find that tape cassettes could replace
disk storage for many applications. We would like to make it possible for
the user to enter data on line, analyze it to some extent, record it on
tape cassettes, and retrieve it for future analysis as needed. The poten-

(26)
TIT. ACME'S FUTURE DIRECTION

tial reliability added by collecting all data on tape cassette at the time
of data input is also significant.

Alphanumeric Displays. The alphanumeric displays added to ACME for the
Drug Interaction Program during this past year have been interfaced to the
PDP-11 which in turn has been interfaced to the 360/50. ACME software has
been and is being modified to better support these devices. For a signifi-
eant number of ACME users (10-20) alphanumeric displays would be better
than the current 2741 typewriter terminals because they are quieter, offer
one page at a time rather than one line at a time, and are very convenient
especially for fixed formats of data entry. We consider special support
for this type of hardware highly desirable. Such support may include new
commands for handling full pages of data and revise editing features, and
other commands designed to take advantage of the &-dimensional nature of
CRE's.

 
PART JI. ACME USER DEVELOPMENTS

(28)
TV. PROJECT DESCRIPTIONS: CORE RESEARCH AND MAJOR USERS

A. Core Research Descriptions

During the past year ACME has identified several user projects which
we consider core research application work under the ACME grant. One
specific example is Dr. Stanley Cohen's drug interaction project. For
this project ACME pays for the computing services including some of the
programming effort.

Another kind of core research application consists of Dr. Lederberg
and Dr. Feigenbaum's DENDRAL project. The DENDRAL project intends to
pay for its computing services according to the grant request, but ACME
will support new systems work specifically for DENDRAL.

A third type of core research is one in which ACME pays for all
computing services and the user pays for all other associated costs.
The projects of Dr. Bellville (G SWANSO.THESIS) and Dr. Sussman fall into
this category. Dr. Sussman developed a Clinical Laboratory Information
System on ACME. When development was complete, he requested financial
support from the Hospital for its operation. Therefore, the LABSYS project
has been completed as a core research task.
 

IV. PROJECT DESCRIPTIONS: CORE RESEARCH AND MAJOR USERS
Cross Reference Core Research Table
Project
Principal File Description
Investigator Project ACME User Pageminutes Storage Page
Bellville, J. THESIS Swanson, G. 770, 609 27, 342 41
Cohen, S. DRUGALRT Cohen, S. 492, 858 20, 141 22
Feigenbaum, E. DENDRAL Reynolds, W. 55, 220 300 33-36
Lederberg, J.
Djerassi, C.
Feigenbaum, E. DENDRAL Ross, R. 128, 176 2, 125 3% = 36
Lederberg, J.
Daerassi, C.
Feigenbaum, E. DENDRAL Stefik, M. 216, 957 2, 088 45-36
Lederberg, J.
Djerassi, C.
Feigenbaum, E, DENDRAL Stillman, R. 88, 896 400 33-36
Lederberg, J.
Dgeragsi, C.
Feigenbaum, BE. DREAMS Reynolds, W. 346, 971 18, 445 33-36
Lederberg, J.
Djerassi, C.
Feigenbaum, E. DREAMS Stillman, R. 199, 650 10, 430 33-36
Lederberg, J.
Djerassi, Cc.
Feigenbaum, £. CHEM Ross, R. 633, 855 15,123 33-30
Lederberg, J.
Djerassi, C.
Feigenbaum, E. GAME Bacon, V. 1, 030, 239 2, 850 a?
Lederberg, J.
Djerassi, C.
Sussman, H. LABSYS Sussman, H. 43, 083 182 38
Sussman, H. LABENG Sussman, H. 924 9 89

(30)
TV. PROJECT DESCRIPTIONS: CORE RESEARCH AND MAJOR USERS
Respiratory Studies

Name: Swanson, G. (P.I.: Bellville, W.) Project: THESIS

Department: Anesthesia

Project Description: The precise interpretation of the drug action mechanism
on the human respiratory system is critically important for the evaluation
of new pain relieving drugs. The improvement of analgesics and antagonists
depends in part on the precision and specificity of this interpretatior.
The classical experimental methods are restricted to assessing drugs in
terms of an integrated respiratory effect. One method cf improving the
specificity is to model the respiratory system mathematically and interpret
a drug effect as a parameter change. This project involves the development
of an experimental computer-aided instrumentation system for accumulating
and interpreting human respiratory response data in terms cf a mathematical
model. The model quantifies the function of the peripheral and central
chemoreceptors, and the effect of oxygen tension on carbon dioxide response.
A parameter estimation scheme estimates the model parameters from input-
output respiratory data. The model input (experimental end-tidal cO67o.
time history) can be specified to minimize the uncertainty in a parametér
estimate,

The system incorporates three important features: (1) An on-line nybrid
computing system for real-time data acquisition of human respiratory CC,
response data, (2) A digitally-controlled breathing chamber in which the
computer dictates the subject's inspired Co, concentration for the ccurse
of an experiment, and (3) A digitally-controlled breathing trainer to st
the subject's voluntary interaction with his involuntray CO, response,
The digital computer dictates the subject's inspired CO, concentraticn Tor
the course of an experiment. This flexibility allows us to design the
dynamic variation in end-tidal CO, so that the experiment yields specific
information about the properties of the human CC, regulator.

The digital computer also controls a device which generates a sound very
similar to human breath sounds. By having @ subject Listen to this device
and try to duplicate the breathing pattern being dictated by the computer,
we can study the subject's voluntary interaction with his involuntary CC.
response, -

This system is presently in use in ongoing studies of the normal and ¢crug-
altered respiratory control system. <A suggested application of this system
is to improve dose-effect sensitivity and drug-effect specificity. Another
application of the system could be for the study of exercise physiology.
IV. PROJECT DESCRIPTIONS: CORE RESEARCH AND MAJOR USERS

Drug Interaction Program

 

Name: Cohen, 8. Project: DRUGALRT

Department: Clinical Pharmacology

Project Description: The project involves the establishment of a computer-
based program aimed at preventing undesirable drug interactions and
reducing drug toxicity at the Stanford University Medical Center. &
data bank dealing with drug interactions of clinical significance will
be compiled utilizing already available information present in the
pharmacological literature. When prescriptions are filled by the
Stanford pharmacists, the pharmacists will type the name of the drug
and the dosage regimen into a terminal located in the Hospital pharmacy.
When a new drug added to a patient's regimen interacts with any cne cf
the several drugs the patient may already be receiving, the computer will
print out an appropriate drug interaction alert accompanied by a litera-
ture reference, which will then be sent to the nursing unit by the
pharmacist -- together with the drug. Prior to administering a drug
accompanied by such an "alert", the nurse will contact the physician in
charge, wno will retain the prerogative of deciding whether or not the
drug should be administered. This program will provide considerable
teaching benefits to students and house staff, in addition to providing
benefits of major importance to patient care. In addition, it will be
possible to assess the impact of providing ohysicians with drug inter-
action information, and also to learn in a prospective way about the
clinical consequences of drug interactions.

To date, a major part of data acquisition and programming has been
completed. It is expected that the program will be operational on a
trial basis during the next several months (April, May L971).

In the future, it is anticipated that the program will be extended to
other hospitals, and perhaps to non-hospital pharmacies in nearby
communities. It is also expected that the program will interface with

the laboratory test program which has recently been developed by Dr.
Howard Sussman. Once this interface has been accomplished, it will be
possible to utilize laboratory information in providing drug interaction
alerts, and also to utilize prescription information in evaluating labora-
tory test results, Thus, laboratory evidence of inadequate renal funeticn
might serve to alert the physician not to administer usual doses of a

drug that is excreted entirely by the kidney. Conversely, the arti-
factual effects of certain drugs on laboratory test results can be
detected and appropriate warnings provided in the clinical laboratory.

(32)
IV. PROJECT DESCRIPTIONS: CORE RESEARCH AND MAJOR USERS

DENDRAL

Reynolds, W., Ross, R., Stefik, M., Project: DENDRAL
and Stillman, R. (P.I.: Feigenbaum,
E., Lederberg, J. and Djerassi, C.)

Department: Chemistry and Genetics

Project Beseription: The DENDRAT, project involves collaboration between the

Instrumentation Research Laboratory operating under NASA grant NGR-0C5-
020-00h, investigators operating under NIH grant GM O00612-01, and ACME.

The emphasis of the DENDRAL-ACME efforts is computer science while that
of IRL-ACME endeavors are data acquisition and computer-instrument
control.

The DENDRAL project aims at emulating in a computer program the induc-
hive behavior of the scientist in an important but sharply limited area
of science, organic chemistry. Most of the work is addressed to the
following problem: Given the data of the mass spectrum of an unknown
compound, induce a workable number of plausible solutions, that is, a
small list of candidate molecular structures. In order to complete ‘the
task, the DENDRAL program then deduces the mass spectrum predicted by
the theory of mass spectrometry for each of the candidates, anc selects
the most productive hypothesis, i.e., the structure whose predicted
spectrum most closely matches the data.

The project has designed, engineered, and demonstrated a computer
program that manifests many aspects of human problem-solving techniques.
Tt also works faster than human intelligence in solving problems cnosen
from an appropriately limited domain of types of compounds, as illus-
trated in the cited publications.

Some of the essential features of the DENDRAL pregram include:

Conceptualizing organic chemistry in terms of topological graph
theory, i.e., a general theory of ways of combining atoms.

Embodying this approach in an exhaustive HYPOTHESIS GENERATOR.
This is a program which is capable, in principle, of “imagining”
every conceivable molecular structure.

. Organizing the GENERATOR so that it avoids duplication and
irrelevancy, and moves from structure to structure in an orderly
and predictable way.

The key concept is that induction becomes a orocess of efficient
selection from the domain of all possible structures. Heuristic
search and evaluation is used to implement this "efficient selection”.
IV. PROJECT DESCRIPTIONS: CORE RESEARCH AND MAJOR USERS

Most of the ingenuity in the program is devoted to heuristic modifi-
cations of the GENERATOR. Some of these modifications result in early
pruning of unproductive or implausible branches of the search tree.
Other modifications require that the program consult the data for cues
(pattern analysis) that can be used by the GENERATOR as a plan for a
more effective order of priorities during hypothesis generation. The
program incorporates a memory of solved sub-problems that can be
consulted to look up a result rather than compute it over and over
again. The program is aimed at facilitating the entry of new ideas by
the chemist when discrepancies are perceived between the actual func-
tioning of the program and his expectation of it.

The DENDRAI research effort has continued to develop along several
dimensions during this period. The mass spectra of some previously
uninvestigated compounds were recorded. The computer program has been
extended to analyze the mass spectra of a more complex class of
compounds, using new kinds of data. The artificial intelligence work
on theory formation and vrogram generality has also progressed.

Many mass spectra were taken to gather more data for the DENDRAL Program.
The analysis of the mass spectra of carbamates and methoxyoximes pro-
vided general mass spectrometry rules for the computer program. The
spectra of many steroids were taken to elucidate the mass spectrometry
of steroids and to provide data for a problem area new to DENDRAL.

The steroid problem is new in several respects: First, in working with
steroids, the program deals with much more complex molecules than ever
before; second, the computer program uses element mans from high reso-
lution data to resolve ambiguities; and third, the program uses
metastable peaks to determine parent-daughter relationships between ions
and thus to distinguish molecular ions and their primary fragments.
Programming for the preliminary analysis of steroid svectra is nearing
completion, and will be useful in the laboratory even though the ccmplete
computer program for analyses of this complexity has not been finished.

The artificial intelligence interests of the DENDRAL group are reflected
in recent work in vrogram generality, partly described in reference ‘4%,
and in the program we call meta-DENDRAL described in reference {¢.,
Which will infer mass spectrometry rules from collections of data.

varts of the meta-DENDRAL program have been written which codify
observations about mass spectrometry, and work has started on the
succeeding phase of the program which will generalize these observations
into tentative rules.

A. Feigenbaum, E.A., Buchanan, B.G., and Lederberg, J. "On
Generality and Problem Solving: A Case Study Using the
DENDRAL Program", in Machine Intelligence 6, 3. Meltzer and
D. Michie (eds), Edinburgh University Press, 19/1.

(34)
IV. PROJECT DESCRIPTIONS: CORE RESEARCH AND MAJOR USERS

B. Buchanan, G. and Lederberg, J. "The Heuristic DENDRAL
Program for Explaining Empirical Data", to be presented
at the 1971 Congress of the International Federation of
Information Processing Society (August, 1971) and published
by North Holland Publishing Co. (in press).

C. Buchanan, G., Feigenbaum E.A., and Lederberg, J. "Beyond
Heuristic DENDRAL", to be presented at the International
Joint Conference on Artificial Intelligence (September,
1971) and published in the Proceedings.

An evaluation of new high performance mass spectrometers was carried
out to plan new instrumentation for this program.

This system plans to incorporate a high degree of computer control.

The goal of the instrumentation project will be to combine the analysis
of the DENDRAL computer program with the data acquisition and control
capability of the computer. It is planned to do a fast preliminary
data acquisition, let the DENDRAL program determine what additional
data and data mode is desirable, have the computer control the instru-
ment mode and data scan, and return the pertinent data to tne DENDRAL
program. Further iterations of this cycle can be repeated as long as
the sample persists.

It is planned to connect a GLC (gas chromatograph) to the inlet of the
mass spectrometer. The persistence of a given sample is determined by
the duration of a GLC peak, a few seconds to a very few minutes. The
mass spectrometer of the kind we are considering could usefully take
data in many modes; low, high, ultra-high resolution and meta stables,
high or low ionization potential, ete, It cannot acquire all this data
in the time span allowed by a single GLC peak. Hence it is required
that the computer determine, during the limited sampling time the most
useful mode of operation, and then implement this optimum mode.

Dispersed Computer System for Instrumentation-HIQ Terminal:

The project for development of dispersed computers is the HIQ (High IQ)
terminel project being carried out in collaboration with Professor
Melvin Schwartz of the Physics Department, with joint NASA support and
Air Force support under contract AF F 44620 67C 0070. Frofessor
Schwartz's interest concerned a Direct Memory Access (DMA) for a PDP-11.
This forms an important portion of our "HIQ"” terminal concept. This
terminal development will provide the instrument-computer interface

for the DENDRAL project.

The HIQ terminal is the result of an experience in connecting scientific
instruments to various computers and the current supply of powerful but
economical mini-computers. These computers have desirable attributes
for dedicated real-time instrument interaction. However, when used

(35)
IV. PROJECT DESCRIPTIONS: CORE RESEARCH AND MAJOR USERS

alone they can require costly software developments. Current time-
shared computers such as ACME have very efficient programming generat-
ing capabilities. In the present configuration the HIQ terminal
utilizes a PDP-1l mini-computer. The mini-computer is tied to the

ACME time-shared computer. The system facilities for the mini-computer
will lie wholly in the larger time-shared computer: program manuscrivot
editing, compiling or assembly, and the handling of files. All these
functions benefit by the programming efficiency of the larger computer.
In addition the time-shared computer replaces local tape and disc units.

By the end of 1970 the prototype HIQ using the PDP-11 as a Local
Processing Unit was connected to ACME via our existing data lines.
Program writing, code assembly, and a degree of control by ACME, the
time-shared computer, has been achieved.

Next in development will be the use of the HIQ to preprocess data from
the existing mass spectrometer data channels in our Chemistry Department
(DENDRAL). After that we will integrate the Direct Memory Access (DMA)
unit for use with the next generation of mass spectrometers, develop
telephone line communication, and other "front end” modules.

Note: The following names and projects are associated with the DENDRAL
project: W_REYNOL. DREAMS, W_REYNOL.DENDRAL, R_ROSS.CHEM, R_ROSS,.DENDRAL,
M_STEFTK, DENDRAL, R_STILLM. DREAMS, and R STTLLM. DENDRAL. They appear
under categories 5 and 6 in the "Summary of Resource Usage Table".

(36)
IV. PROJECT DESCRIPTIONS: CORE RESEARCH AND MAJOR USERS

Mass Spectrometry

Name: Duffield, A. and Lederberg, J. Project: GAME
(see Bacon, V. on Summary of Resource
Usage table)

Department: Genetics

Project Description: This project involves data analysis from a Finnigan 1015
mass spectrometer, In this on-line project, the decision-maxing capa-
bilities of the computer are coupled with those of an operator to direct
the operation of a Finnigan 1015 quadrupole mass spectrometer.

The computer is used to actively direct the operation of the mass spec-
trometer by controlling the mass filtering system of the instrument. It
is used to recognize and control the voltage changes which define mass
peaks and enable the rapid collection and presentation of data,

The computer traces out peak shapes of the known masses in a reference
gas, allowing the operator to determine correct mass positions and to
enter any shifts in calibration into the computer register for compensa-
tion automatically.

While taking data, the information may be displayed on an oscilloscope
or recorded on magnetic tape. Once data is acquired, the structural
identification of organic compounds is made from orthogonal coordinate
or spiral base plots of mass spectra made by computer direction of a
calcomp plotter. The system is also used to analyze gas liquid chromato-
graph effluent, permitting the structural identification of mixtures of
organic compounds.

Stored data offer the future possibility of spectra matching of unknown
compounds.

It is being applied to problems supported by both NTE Grant AM 12797 -O1
and NASA Grant NGR-05-020-004. Useful applications have been in the
determination of optical purity of alcohols, ketones, phenylacetic acids
and phenoxypropionic acids. This analytical technique has also been used
to determine the absolute configuration of alloisoleucine present in the
serum of a patient suffering from "Maple Syrup Urine disease”, In
addition chemical methods are being developed for the analysis of the
basic and acidic constituents of urine. The bases are present in far
smaller amounts than are the acids and the isolation of the former
requires maximum sensitivity from the gas chromatography-mess spectro-
metry system. The computerization of this instrumentation aids in both
the data collection and analysis.
IV. PROJECT DESCRIPTIONS: CORE RESEARCH AND MAJOR USERS

Clinical Laboratory Information System

Name: Sussman, H. Project: LABSYS
Department: Pathology

Project Description: A computer-based laboratory information system has
been established at the Stanford University Medical Center. It is
a modern system of handling the data flow in the clinical labora-
tories and has been developed on the ACME facility. Specific
objectives of the system are to:

Ne

) Reduce human errors;

) Insure that processed data is available at a certain time each
days

Implement self-checking routines for monitoring accuracy of test
results and to flag results that are beyond normal range;

Reduce bookkeeping and clerical work costs;

Provide wider distribution of data;

Improve procedural work flows as a step towards establishing
a systems-oriented Laboratory.

\N
SS

NW fe
Noe Nee

To date, the laboratory information system that has been developed
handles the following functions of the clinical laboratory:

1) Ordering and processing of requests;

2) Accessioning of specimens into the laboratory;

2) Processing specimens within the laboratory for test analyses;

4) Recording test results both within the laboratory and externally.

All data is entered into the system by optical mark cards through a
Yewlett Packard 21-2761B optical mark sense reader, These cards include
cards to request that services be performed by the laboratory, cards to
indicate the disposition of specimens within the laboratory, and cards
to enter test results. ACME is used to control the process and report
back information to the Laboratory.

The first pilot study was conducted in the intensive care unit at the
Stanford Medical Center. The Logic of the system was tested and the
operational commitments of the various staffs were determined. A
current pilot study is being conducted using the entire Hoover
Pavillion as a test model. In the next few months the gains of the
automated laboratory system will be evaluated over the present manual
system. The study is designed to allow an operations research type
study of all transactions between the Hospital and the laboratory.
IV. PROJECT DESCRIPTIONS: CORE RESEARCH AND MAJOR USERS

Name: Sussman, H. Project: LABENG

Department: Pathology

Project Description: Four Electrical Engineering students are connecting

one of the large instruments in the clinical laboratory to the IBM
1800 for data collection.

This project will further the establishment of a modern system of
handling the data flow in the clinical laboratory. The software for

such a system is largely developed using ACME, with test results being
entered by means of cards.

A direct connection to the 1890 will enable results to be entered into

the system without the high probability of error inherent in tran-
scribing results onto cards.

(39)
TV. PROJECT DESCRIPTIONS: CORE RESEARCH AND MAJOR USERS

B. Major Users

Listed in alphabetic order by Principal Investigator are descriptions
of a few of ACME's major user projects. Additional descriptions of
projects using ACME are presented in Part IV of this report. All of
the projects described here are charged for all of their use of ACME
services.

The Biotechnology Resources Branch requested that the current year
annual report include more descriptive material on a few of the major
users' projects, We have selected the projects which follow with these
criteria in mind: amount of usage, written material available, and
varying types of usage of ACME. Dr. DeGrazia's work in Nuclear Medicine
involves the use of a model, realtime data acquisition, and graphics
terminals. The Immunology Project of Dr. Fries is an information re-
trieval project on a large data base. Dr. McConnell's project involves
instrumentation that drives a plotter. ACME reads the X-Y coordinates
via the 1800. Dr. Pauling's project involves realtime data collection of
time series data from gas chromatographs, plotting of data and analysis.
Dr. Petralli's work in the Infectious Disease Laboratory is an example of
a production program currently used to ensure good quality control in the
nandling of laboratory samples. The Stroke Registry Project of Dr. John
Wilson entails creation of a data base, manipulation of data, and report
generation.

(40)
IV. PROJECT DESCRIPTIONS: CORE RESEARCH AND MAJOR USERS

Name: DeGrazia, J. Project: RADIOREN
Department: Nuclear Medicine

Project Description: Renogram data are analyzed with a kinetic model
of renal function to determine whether (1) quantitative estimates
of renal function parameters could be obtained and (2) advantage
accrues from employing additional probes in performing hippuran
renograms.

Four-probe renograms were performed on patients suspected of having
renal disease. The sitting hydrated patient was given an intravenous
bolus of 7-131 hippuran. Four balanced probes, equipped with thick-
walled collimators, continuously monitored activity over each kidney,
the heart and the bladder. Data was obtained for thirty minutes
directly through an IBM 1800 or recorded by audio tape.

Among the parameters derived are:

1) Effective renal perfusion, ag a fraction of cardiac output;
2) Fraction of effective renal perfusion to each kidney;

4) Exeretion rate constant for each kidney (a measure of tubular
function);

4) Urine transport time for each collecting system (a measure of
obstruction, dehydration, etc.).

The clinical value of parameter estimation is being evaluated and a
comparison is being made between the two-probe analysis and that

using four probes, Thus this is testing for improvement when areas

of interest are used to form the data base. This program will be a
model for diagnostic studies of other organs with radioactive isotopes.

(41)
TV. PROJECT DESCRIPTIONS: CORE RESEARCH AND MAJOR USERS

Name: Fries, J. Project: DATABANK
Department: Medicine - Immunology

Project Description: This project establishes a large databank of clinical
information and explores multiple uses of such stored information.
The databank currently includes 1,200 patient visits and approximately
half a million individual items of data. A variety of search orocedures
operate on this body of data to give answers to clinical questions,
provide administrative data, perform retrospective research procedures,
coordinate and evaluate ongoing prospective clinical trials, and provide
data for clinical correlations.

The project grew out of clinical problems in establishing significant
clinical and laboratory correlations in immunologic diseases and assess-
ing these correlations in regard to classification, pathogenesis, prog-
nosis, and response to therapy. Diseases in this area are cheracterizec
by involvement of a large number of organ systems, a Large variety of
associated laboratory abnormalities, and a course long in duration and
punctuated with periodic exacerbations and remissions. A massive amrcunt
of data is thus generated by each patient and meaningful correlations may
be obscure to the clinician and inaccessible to the clinical researcher,
The data cannot be well handled retrospectively under current systems
nor can they be handled manually.

An orderly comprehensive method of recording vatient data structured

with respect to time has been develoved and established. Information
recorded in this manner has been stored in the databank. Using the
databank as a central information processing vehicle, associated physi-
clans in the community have become active in clinical research projects
of the division and are obtaining significant clinical information rele-
vant to management of their patients. In the early stages, therefore, e
regional medical system of physicians interested in rheumatic disease is
being established with an upgrading and a monitoring of the quality of
care in the region. The system was developed for rheumatic disease but
is easily applicable to any disease area or to general medicine itsel?.
Other groups associating with the project are actively pursuing these
ramifications. The availability of the databank has greatly enhanced the
clinical program and has resulted in the establishment of clinical fellow-
ships and additional student learning opportunities. The teaching value
of available accessible clinical information cannot be overestimated and
these applicatioris are in daily use.
Ivy. PROJECT DESCRIPTIONS: CORE RESEARCH AND MAJOR USERS

Name: McConnell, H. Project: ABSORB
Department: Chemistry

Project Description: ACME is being employed for analysis of experimental
paramagnetic resonance spectra and calculation of theoretical spectra.
These paramagnetic resonance spectra arise from the application of the
"spin label" technique to problems involving biological macromolecules.
Biological problems which are currently being studied include cooperative
oxygen binding to the protein hemogiobin and the relation of molecular
orientation and motion to funetion in biological membranes and membrane
model systems.

The paramagnetic resonance spectra that are obtained in spin label studies
are recorded as the first derivative of an absorption curve. The area
under the absorption curve is proportional to the number of spin labels
giving rise to the signal. Spin label spectra give information on changes
in conformation or motion of a macromolecule. Frequently, these changes
are detected as a small change in the paramagnetic resonance spectrum.
Quantitative measurement of these changes requires normalization of spectra
from a series of experiments. ACME will be used to compute the double
integral of the experimental spectrum and then to regenerate a normalized
spectrum, Subtraction and addition of spectra will also be used to
analyze experimental data. In addition, a program for generation of
theoretically calculated spectra will be employed.

Name: Pauling, Lb. Project: MENTLRES

Department: Chemistry

Project Description: This project involves research on the molecular
basis of mental disease. Samples of urine and breath are taken
before and after a synthetic diet, and analyzed by gas chromatography
for abnormal patterns. Computerization is necessary for collection
and rapid analysis of large volumes of data, and the project has
recently hooked into the ACME system. A small project on mental
retardation has been run and future plans include a large study of
schizophrenia with all analysis done within ACME.
IV. PROJECT DESCRIPTIONS: CORE RESEARCH AND MAJOR USERS

Name: Petralli, J. Project: MED DATA
Department: Medicine - Infectious Diseases

Project Description: Antibiotic-sensitivity testing gives physicians impor-
tant information about treatment of specific infections. The fact that
the Food and Drug Administration recently decided to regulate the formu-
lation of antibiotic-sensitivity disks reflects the increasing awareness
ef the critical nature of this testing procedure. To improve the qualivy
of antibiotic-sensitivity data (high potency single disc method) and to
guide the interpretation of results and antibiotic selection, a computer
program has been developed. Clinical information and zone sizes are
entered into the ACME computer each day. As the information is given to
the computer, the quality-control program immediately detects and challenges
unusual results and directs the laboratory technician to appropriate
restudy of the organism in question. This system converts zone sizes to
resistant, intermediate, or sensitive and prints final reports from its
memory. Decreased potency of antibiotic disc is detected by comparison
of periodically determined mean zone sizes. Limits of confidence cf a
single reading are established by review of zone sizes observed with a
standard organism tested on different occasions. Knowledge of antibiotic
sensitivities of organisms isolated from a specific site such as blood
or urine will help to guide the selection of antibiotics before specific
sensitivities are known, Such information is of value in selection of
antibiotics in treating rarely encountered organisms with less well-
known sensitivity patterns or in selection of alternate antibiotics when
the first choice drug is hazardous. Yearly comparison of antibiotic
sensitivity patterns obtained will give information about major trends
and suggest appropriate changes in treatment of various infections.

Thus, a new weapon in the battle against disease has been developed --

a computer to keep tab on dangerous bacteria and to helo select which
antibiotic drugs should be used to treat infections. Through antiobiovic-
sensitivity testing, the computer has completely eliminated errors in
conversion from zone size to sensitivity, which otherwise occur once in
every 12 reports.

In addition to improving the accuracy of laboratory results for the benefit
of patients, the computer has proved valuable in checking the work of
laboratory technicians and students undergoing training. With the initial
goal accomplished, that of computer quality control of antibiotic-
sensitivity testing, future plans call for the study of when and why
organisms become resistant to antibiotics.
IV. PROJECT DESCRIPTIONS: CORE RESEARCH AND MAJOR USERS

Name: Wilson, J. Project: STROKE
Department: Regional Medical Program

Project Description: This project involves the development of a county wide
registry on stroke patients for a period of one year. The goal is to
develop a population base for study and analysis of descriptive para-
meters of stroke, correlations and relationships cf parameters with
resultant predictive output for improvement of treatment and care of
patients.

To date, two programs have been written for input and storage of data
on stroke patients in acute facilities. These can be used to update

and correct data as well as in checking the accuracy of the data. Once
the data has been input, two output programs can be run which give an
evaluation of the patient's status upon admittance and discharge. For
the period July 1, 1970 through February 28, 1971, data on approximately
300 patients has been stored in the computer file. Each month, data on
30-4O stroke patients is entered into this file via the teletype terminal
located at the Dominican Santa Cruz Hospital. On the same terminal,
evaluations of the patient's status on admission and discharge are ob-
tained based on the computer data, and a printout of the data regarding
the patient is placed in the patient's medical records. Preliminary
statistical analysis (in the form of a monthly report) of the data
obtained between July 1, 1970 and February 28, 1971 has also been com-
pleted,

Future plans include the continuation of data collection and processing,
Enlargement of the stroke registry in the computer is anticipates to
include data on long-term care of the stroke patient based on evalua-
tions at three months, six months and a year after discharge from the
acute facility. More sophisticated and definitive analysis of data will
aid in improvement of care. Possible predictive relationships and
correlations will be sought from the data.

(45)
V. USER CORE RESEARCH - FY197e2

The user-related core research activities for the coming year will include
the DENDRAL Program, Drug Interaction Program, Clinical Laboratory Data
Collection and Analysis, Arrhythmia Detection, and other programs to be
identified. Only the Arrhythmia Detection is a new project.

A. Current Core Research Projects

1. DENDRAL

ACME will be used to support DENDRAL over the next 12 to 18 months primar-
ily by improving the LISP compiler and relating realtime services to DEN-
DRAL needs. Some graphics support is contemplated for scanning search
trees.

2. Drug Interaction Program

The Drug Interaction Program will continue but on new problems. The goal
in FY1971 was to get prescription data into the system along with a patient
census and to launch use of the system by the Pharmacy. For next year, Dr.
Cohen is seeking a new grant for further work in relating drug interactions
to more of the known factors about each patient (i.e., clinical lab data,
admissions data).

3, Clinical Laboratory

Additional instruments in the laboratory will be connected to ACME for
realtime data collection.

B. Proposed Core Research

1. Arrhythmia Detection

The Arrhythmia Detection System is planned as a joint effort between ACME
and Dr. Don Harrison in the division of Cardiology. Dr. Harrison expects
to use a small Hewlett-Packard computer as the control device for an arr-
hythmia detection system. The Hewlett-Packard computer would be connected
to the ACME system for large file storage and data analysis. The detection
system will be based largely on the work done at Washington University in
Saint Louis by Dr. Jerry Cox and others, and will incorporate a number of
new design concepts from a local group. This is viewed as the first step
in a Cardiac Care Unit Monitoring Package.
V. USER CORE RESEARCH - FY1972

This project is included as ACME core research with the following conditions:

® This project is to be funded from ACME's income or other-
wise as a cost above the basic budget of $800,000 in direct
costs for FY1L972.

@® We are not attempting to compete with the study section re-
view process. The basis for ACME's interest in this process
is a future convergence of ACME's realtime goals with the
overall Cardiac Care Unit Monitoring System.

At the present time the detection of arrhythmia depends upon two techniques:

@ Analog devices that detect gross changes in heart rate.
@ Visual EKG monitoring by trained nurses,

Neither of these techniques, alone or in concert, are wholly satisfactory.
Devices that detect gross rate changes give an indication only after the
patient has entered a life-threatening condition. By visual monitoring,
trained nurses should be able to detect subtle arrhythmias which are often
precursors to a life-threatening situation. However, it is obviously un-
realistic to expect anything close to 100% effective visual monitoring.
Limited attention span, fatigue, a number of tracings to watch simultan-
eously, plus the normal distractions of the CCU environment all vitiate the
nurse's effectiveness.

What is needed is a technique that would be able to detect subtle arrhythmias
and to provide constant monitoring.

This would be some sort of device that would be able to monitor EKG's on a
beat-by-beat basis, detect the subtle arrhythmias that are precursors to
life-threatening situations, and give an alarm that would allow a nurse to
initiate proper treatment.

The heart of the arrhythmia monitoring system will be a high-performance
mini-computer. The exact size of the computer, as well as the number of
patients thet can be monitored are yet to be determined. From past experi-
ence we expect to monitor 8 patients using a machine with an 8k memory. The
execution speed of the machine is of critical importance since the mumber of
patients monitored, and hence the cost-per-patient of the system, is primarily
limited by this factor.

Input of FKG's to the system will be performed by a high-speed analog-to-
digital converter. The analog-to-digital converter will interface to the
normal central station cable connectors. Qutput of alarms will be aug-
mented by additional alarm indicators to the normal central station alarm
devices. This will allow for the normal operation of such devices as tape
delays or graphics recorders.
V. USER CORE RESEARCH - FY1L972

2. Other Projects

Other core research applications will be identified in the course of the
year. Potential candidates at this time include Dr. Eugene Dong's non-
invasive arrhythmia detection system development, and some of Dr. De-
Grazia's kidney function work in Nuclear Medicine. The principal invest-
igator for ACME will issue new criteria for potential users to consider
if they wish to become classified as ACME core research. Further dis-
cussions with NIH Biotechnology Rescurces Branch are needed on this sub-
ject.

43. Realtime Programs

We will attempt to expand the number of realtime users. These users
have received a priviledged rate for ACME use because they have been
pioneering a new area and are sharing the problem solving techniques
developed with the community. As part of our move to become self-

sufficient through fee for service, we expect to reduce the subsidy.
PART 3: UTILIZATION AND COST DATA

(49)
Vi.

A. Resource Expenditures

SUMMARY

10,

ii.

Personnel:
a, Salaries & Wages
b. Fringe Benefits

Subtotal
Consultant Services

Eauipment:

a, Main Resource-Rented

b. Main Resource-Purchased
e. Supporting Equipment

d. Equipment Maintenance

Subtotal
Supplies
Travel
Engineering Services
Publication Costs
Other (1)

a, Computer Services
b, Other

subtotal
Subtotal - Direct Costs
(2)

Indirec® Costs

Total Costs

 

 

 

\

BUDGET
Total Resource Expenditures
Actual Estimatea
Previous Current Next
Budget Budget Budget
Period Period Period
201,673 237,973 247 735
24,710 _ 32,942 37.2384
226, 383 270,915 285,119
1,500 800 1,000
367 , 394 375,254 365,825
27 5453 39,251 61,136
10,876 10,120 8,362
4225 6,559 CG, 670
409,948 431,184 bh O13
12,939 13,500 16,250
3,896 3,600 4900
20,508 17,810 30,900
4 262 4,000 L000
12,177 9,660 7,000
(2915 (2900 8,618
20,092 17,560 15,618
699,528 759,369 800,000 ‘9
114,954 iho, bok 171,598
S14 482 899,773 971,595

(1) Includes IBM education courses

(2) Based on salary and wages in years O4 and 05; based on
net ($800,000 - equipment rentals and purchases totalling
$428,077) total direct costs in year 06 (will be reduced
by netting revenue from service fees from direct costs)

(4) This budget level does not include the efforts we would
like to fund using ACME's income.
mate for arrythmia detection work on page 52.

(50)

For example,

see esti-