0 descision support systems in medicine - shojaee

In the Name of

the promoter

of

Wisdom and Beauty

BY: DR. ABBAS SHOJAEE M.D.

An Introduction to Decision Support Systems in

Medicine

This presentation uses works of:• Joseph Tan – Ehealth care information systems (book)• J.H. Van Bemmel - Medical informatics (book)• Edward H. Shortliffe – Biomedical Informatics (book)• Kate Farrell – Intro expert systems test (ppt file)

Introduction

Every human action is the next step to data acquisition and interpretation.

Practicing medicine is making decisions indeed.Computers are presumed that will help in these three steps

and so decision Support in medicine.Due to several limitations in storage, representation and

interpretation of information, it is yet limited to well structured and actionable data and knowledge.

Questions exist: What? How? When? At what quality? How they affect the field?

Decision making in medicine

Acquisition of data

Linking to data context to form

meaningful information

Deciding on next required

information and requesting them

Trying to reduce possibilities to

form an actionable hypothesis

Action: making therapy orDeciding on

next diagnostic step Interpretation

Itera

tes

Types of clinical decisions

Clinical decisions: Diagnosis

What is true? Diagnosis process.

Which information is needed to make right decision Treatment

Which is the best pass way through the jungle of possibilities

Requirements: Accurate and adequate data. Applicable knowledge. Appropriate problem solving skills. (inference engine)

Some obstacles to use computers as DSS in medicine

Abundance of rules and interacting data.Computational models are not as good as

human inference modelsYet we do not know the model(s) that brain

uses to infer.Brain computational power is higher than our

computers.

Additional terms

Inductive and deductive reasoningClassificationFeature setClusteringDecision modelSupervised and Unsupervised learningDifferential diagnosis

Expert Systems

An expert system is a computer program that is designed to hold the accumulated knowledge of one or more domain experts

Applications of Expert Systems

PROSPECTOR:Used by geologists to

identify sites for drilling or mining

PUFF:Medical system

for diagnosis of respiratory conditions

Applications of Expert Systems

DESIGN ADVISOR:Gives advice to designers of

processor chips

MYCIN:Medical system for diagnosing blood

disorders. First used in 1979

Applications of Expert Systems

DENDRAL: Used to identify the structure of chemical compounds.

First used in 1965

LITHIAN: Gives advice to archaeologists examining

stone tools

Components of an Expert System

The knowledge base is the collection of facts and rules which describe all the knowledge about the problem domain

The inference engine is the part of the system that chooses which facts and rules to apply when trying to solve the user’s query

The user interface is the part of the system which takes in the user’s query in a readable form and passes it to the inference engine. It then displays the results to the user.

Why use Expert Systems?

It takes long time and great costs to train an expert.

Experts are not always available. An expert system can be used anywhere, any time.

Human experts are not 100% reliable or consistent

Experts may not be good at explaining decisions

Cost effective

Some Problems with Expert Systems

Limited domainSystems are not always up

to date, and don’t learnNo “common sense”Experts needed to setup

and maintain system

Legal and Ethical Issues

Who is responsible if the advice is wrong? The user? The domain expert? The knowledge engineer? The programmer of the expert system shell? The company selling the software?

Medical Expert Systems

Codifying Human Knowledge

• Decomposition into “chunks” of knowledge, chaining of inferences

• Matching of case data to prototypical situations

• Using causal models (pathophysiology) to figure out cases

Mycin: Rule-based Systems

Task: Diagnosis and prescription for bacterial infections of the blood (and later meningitis)

Method: Collection of modular

rules Backward chaining Certainty factors

RULE037

IF the organism1) stains grampos2) has coccus shape3) grows in chains

THENThere is suggestive evidence (.7) that the identity of the organism is streptococcus.

Mycin Consult

--------PATIENT-1--------1) Patient's name: FRED SMITH2) Sex: MALE3) Age: 554) Have you been able to obtain positive cultures from a site at which Fred

Smith has an infection? YES--------INFECTION-1--------

5) What is the infection? PRIMARY-BACTEREMIA6) Please give the date when signs of INFECTION-1 appeared. 5/5/75The most recent positive culture associated with the primary-bacteremia will

be referred to as:--------CULTURE-1--------7) From what site was the specimen for CULTURE-1 taken? BLOOD8) Please give the date when this culture was obtained. 5/9/75The first significant organism from this blood culture will be called:

--------ORGANISM-1--------9) Enter the identity of ORGANISM-1. UNKNOWN10) Is ORGANISM-1 a rod or coccus (etc.)? ROD11) The gram stain of ORGANISM-1: GRAMNEG. . .

Davis, et al., Artificial Intelligence 8: 15-45 (1977)

How Mycin Works

To find out a fact◦ If there are rules that can conclude it, try them◦ Ask the user

To “run” a rule◦ Try to find out if the facts in the premises are true◦ If they all are, then assert the conclusion(s), with a suitable

certainty Backward chaining from goal to given facts

Dynamically traces out behavior of (what might be) a flowchart Information used everywhere

appropriate Single expression of any piece of

knowledge

Explore Mycin’s Use of Knowledge

** Did you use RULE 163 to find out anything about ORGANISM-1?RULE163 was tried in the context of ORGANISM-1, but it failed because

it is not true that the patient has had a genito-urinary tract manipulative procedure (clause 3).

** Why didn't you consider streptococcus as a possibility?The following rule could have been used to determine that the identity of

ORGANISM-1 was streptococcus: RULE033But clause 2 (“the morphology of the organism is coccus”) was already

known to be false for ORGANISM-1, so the rule was never tried.Davis, et al., Artificial Intelligence 8: 15-45 (1977)

Even Simpler Representation

Disease

s1s2s3s4s5s6s7s8s9s10s...

Disease

s1s2s3s4s5s6s7s8s9s10s...

Diagnosis by Card Selection

Disease

s1s2s3s4s5s6s7s8s9s10s...

Disease

s1s2s3s4s5s6s7s8s9s10s...

Disease

s1s2s3s4s5s6s7s8s9s10s...

Disease

s1s2s3s4s5s6s7s8s9s10s...

Diagnosis by Edge-Punched Cards

Dx is intersection of sets of diseases that may cause all the observed symptoms

Difficulties: Uncertainty Multiple diseases

“Problem-Knowledge Coupler” of Weed

Multi-Hypothesis Diagnosis

Set aside complementary hypotheses… and manifestations predicted by themSolve diagnostic problem among competitorsEliminate confirmed hypotheses and

manifestations explained by themRepeat as long as there are coherent

problems among the remaining data

Internist/QMR

Knowledge Base: 956 hypotheses 4090 manifestations (about 75/hypothesis) Evocation like P(H|M) Frequency like P(M|H) Importance of each M Causal relations between H’s

Diagnostic Strategy: Scoring function Partitioning Several questioning strategies

QMR Database

QMR Scoring

Positive Factors Evoking strength of observed Manifestations Scaled Frequency of causal links from confirmed

HypothesesNegative Factors

Frequency of predicted but absent Manifestations Importance of unexplained Manifestations

Various scaling parameters (roughly exponential)

Example Case

Initial Solution

More Expert Systems

Causality?What’s in a Link?Temporal reasoningQuantitative reasoningModel-based reasoningWorkflow

Meaning of Representation?

Always? probabilityMagnitude? severity; bad cold worse fever?Delay? temporalityWhere? spatial dependencyUnder what conditions? contextInteraction of multiple causes physical lawsCross-terms high-dimensional descriptions

SDcauses

Interpreting the Pastwith a Causal/Temporal Model

weak heart

heart failure

digitalis effect

retain

losediuretic effect

high

low

edemafluid therapy

water blood volume

low cardiac output

definite cause

possible cause

possible correction (not all shown)

Exploiting Temporal Relations

transfusion precedes both abdominal pain and jaundice implies transfusion-borne acute hepatitis B

as in 1, but only by one dayjaundice occurred 20 years ago, transfusion

and pain recentCan be very efficient at filtering out

nonsense hypotheses.

bloodtransfusion

abdominalpain

jaundice

?

?

The “Too Many Rules” Problem

One decision support rule may work very well.

10 rules may work better1000 rules may be unworkable!Need for a comprehensive “view” of the

medical status of the patientAlert based on a full set of diagnoses not

individual facts

Conclusions

Medical decision support can be helpful in reducing medical errors

Usually based on simple rule based systemsMore advanced medical decision support

systems can be use to help with diagnosisNot much use of advanced systems at present

in routine care