Artificial-intelligence-augmented clinical medicine

Klaus-Peter Adlassnig

Section for Medical Expert and Knowledge-Based SystemsCenter for Medical Statistics, Informatics, and Intelligent SystemsMedical University of ViennaSpitalgasse 23, A-1090 Viennawww.meduniwien.ac.at/kpa

Einführung in Medizinische Informatik, WS 2013/14, 30. Oktober 2013

Artificial Intelligence (AI)

• Definition 1: AI is a field of science and engineering concerned with the computational understanding of what is commonly called intelligent behavior, and with the creation of artifacts that exhibit such behavior.

from: Shapiro, S.C. (1992) Artificial Intelligence. In Shapiro, S.C. (ed.) Encyclopedia of Artificial Intelligence, 2nd ed., vol. 1, Wiley, New York, 54–57.

• Definition 2: AI is the science of artificial simulation of human thought processes with computers.from: Feigenbaum, E.A. & Feldman, J. (eds.) (1995) Computers & Thought. AAAI Press, Menlo Park, back cover.

Artificial Intelligence—applicable to clinical medicine

• It is the decomposition of an entire clinical thought process and its separate artificial simulation—also of simple instances of “clinical thought”—that make the task of AI in clinical medicine manageable.

• A functionally-driven science of AI that extends clinicians through computersystems step by step can immediately be established.

⇓artificial-intelligence-augmented clinical medicine

Computational intelligence in medical research

biomolecular researchmedical statistics

clustering and classification data and knowledge mining

consensus conferences

definitional, causal, statistical, and heuristic knowledge

molecular biomedicine

medical knowledge

facts consensus generalization

medical studies

evidence-based medicine

Computational intelligence in patient care

decision-oriented analysis and interpretation

of patient data

definitional, causal, statistical, and heuristic knowledge

human-to-human human-to-humanAI-augmentation

patient-physician encounter

patient-physician follow-up

medical knowledge modules

medical knowledge

Computers in clinical medicine—steps of natural progression

• step 1: patient administration• admission, transfer, discharge, and billing

• step 2: documentation of patients’ medical data• electronic health record: all media, distributed, life-long (partially fulfilled)

• step 3: patient and hospital analytics• data warehouses, quality measures, reporting and research databases,

patient recruitment… population-specific

• step 4: clinical decision support• safety net, quality assurance, evidence-based

… patient-specific

Clinical medicine

medical guidelines

history data

symptoms

signs

laboratorytest results

biosignals

images

genetic data

prognosispatient patient

symptomatic therapy

differentialtherapy

differentialdiagnosis

symptoms

signs

test results

findings

examination subspecialities clinic

medication history

radiological diagnosis

laboratorydiagnosis

…

Clinical medicine

medical guidelines

history data

symptoms

signs

laboratorytest results

biosignals

images

genetic data

prognosispatient patient

symptomatic therapy

differentialtherapy

differentialdiagnosis

symptoms

signs

test results

findings

examination subspecialities clinic

medication history

radiological diagnosis

laboratorydiagnosis

…

QMRDxPlainCADIAG

MYCIN

…

ANNs

SVMs

…

…

personalizedmedicine

+

+

Clinical medicine: high complexity

• sources of medical knowledge‒ definitional‒ causal‒ statistical‒ heuristic

• layers of medical knowledge‒ observational and measurement level‒ interpretation, abstraction, aggregation, summation‒ pathophysiological states‒ diseases/diagnoses, therapies, prognoses, management decisions

• imprecision, uncertainty, and incompleteness‒ imprecision (=fuzziness) of medical concepts

* due to the unsharpness of boundaries of linguistic concepts‒ uncertainty of medical conclusions

* due to the uncertainty of the occurrence and co-occurrence of imprecise medical concepts‒ incompleteness of medical data and medical theory

* due to only partially known data and partially known explanations for medical phenomena• “gigantic” amount of medical data and medical knowledge

‒ patient history, physical examination, laboratory test results, clinical findings‒ symptom-disease relationships, disease-therapy relationships, …‒ terminologies, ontologies: SNOMED CT, LOINC, UMLS, …

specialisation, teamwork, quality management, computer support

• studies in Colorado and Utah and in New York (1997)

– errors in the delivery of health care leading to the death of as many as 98,000 US citizens annually

• causes of errors– error or delay in diagnosis– failure to employ indicated tests– use of outmoded tests or therapy– failure to act on results of testing or

monitoring– error in the performance of a test, procedure,

or operation– error in administering the treatment – error in the dose or method of using a drug– avoidable delay in treatment or in responding

to an abnormal test– inappropriate (not indicated) care– failure of communication– equipment failure

• prevention of errors– we must systematically design safety into

processes of care

errors

prevention

Medical information and knowledge-based systems

symptomssigns

test resultsclinical findings

biosignalsimages

diagnosestherapies

nursing data

•••

standardizationtelecommunication

chip cards

anatomybiochemistryphysiology

pathophysiologypathologynosology

therapeutic knowledgedisease management

•••

subjective experienceintuition

knowledge-based systems

patient’s medical data physician’s medical knowledge

medical statisticsclustering & classificationdata & knowledge mining

machine learning

clinical decision supportmedical expert systems

manypatients

single patient

diagnosistherapy

prognosismanagement

generalknowledge

•••

generalknowledge

telemedicine telemedicineintegration

information systems

induction

deduction

Clinical decision support and quality assurance (in general)

patients’ structured medical data

patient management guidelines & quality assurance• evidence-based reminders and processes• computerized clinical guidelines, protocols, SOPs• healthcare-associated infection surveillance

prognostic prediction• illness severity scores, prediction rules• trend detection and visualization

therapy advice• drug alerts, reminders, calculations

– indication, contraindications, redundant medications, substitutions

– adverse drug events, interactions,dosage calculations, consequent orders

• management of antimicrobial therapies, resistance• (open-loop) control systems

diagnostic support• clinical alerts, reminders, calculations• data interpretation, (tele)monitoring• differential diagnostic consultation

– rare diseases, rare syndromes– further or redundant investigations – pathological signs accounted for

• consensus-criteria-based evaluation– definitions– classification criteria

according toKensaku Kawamoto,University of Utah, 2012:

“A Holy Grail of clinical informatics is scalable, interoperable clinical decision support.”

What have we done?

Interpretation of

hepatitis serology test results

test results

interpretation

ORBIS Experter: Hepatitis serology diagnostics

Interpretation of

hepatitis serology test results

Differentialdiagnose rheumatischerErkrankungen

ComputergestützteEntwöhnung vom

Respirator

Personalized clinical decision support

history physicalsigns

lab tests

clinicalfindings + genomic

data

patient medical data

present (personalized) CDS

future personalized CDS

unavoidable, more specific diagnostics, extends the realm of therapy

Solution at the Vienna General Hospital

Arden Syntax development &

test environment

data & concept mining

data & knowledge

services center

Arden Syntax rule engine

knowledge

knowledge

resultsdatain

terf

aces

Arden Syntax server

wards

HIS

departments

HIS

units

PDMS

clinical laboratories

LIS

genomic data laboratories

LIS

hospital

extended documen-

tation& research

data base

Arden Syntax and Health Level Seven (HL7)

• A standard language for writing situation-action rules that can trigger alerts based on abnormal clinical events detected by a clinical information system.

• Each module, referred to as a Medical Logic Module (MLM), contains sufficient knowledge to make a single decision.

extended by packages of MLMs for complex clinical decision support

• The Health Level Seven Arden Syntax for Medical Logic Systems, Version 2.9—including fuzzy methodologies—was approved by the American National Standards Institute (ANSI) and by Health Level Seven International (HL7) on 14 March 2013.

continuous development since 1989

General MLM LayoutMaintenance Category Library Category Knowledge Category Resources Category

Identify an MLMData TypesOperators

Basic OperatorsCurly Braces List OperatorsLogical OperatorsComparison OperatorsString OperatorsArithmetic OperatorsOther Operators

Control StatementsCall/Write Statements and Trigger

Sample MLM (excerpt)

Arden Syntax, Arden Syntax server, and health care informationsystems

functionality

integration

HIS, MIS, PDMS, LIS, medical practice SW, web-based EHR, telemedicine applications,health portals,…

reminders and alerts, monitoring, surveillance, diagnostic andtherapeutic decision support, …

*operational:- harmonized input data- Arden Syntax MLMs - collected reasoning dataexploratory:- rule learning/tuning- data and concept mining

*data & knowledge services center

service-oriented

web-based

Arden Syntax server and software components

• web-services-based Arden Syntax server including

‒ Arden Syntax engine‒ MLM manager‒ XML-protocol-based

interfaces, e.g., SOAP, REST, and HL7

‒ a project-specific dataand knowledge servicescenter may be hosted

• Java libraries ‒ Arden Syntax compiler ‒ Arden Syntax engine

• Arden Syntax integrated development and test environment (IDE) including‒ Medical logic module

(MLM) editor and authoring tool

‒ Arden Syntax compiler(syntax versions 2.1, 2.5, 2.6, 2.7, 2.8, and 2.9)

‒ Arden Syntax engine‒ MLM test environment‒ MLM export component

• command-line Arden Syntax compiler

Arden Syntax development & test

environment

data & knowledge services center

Arden Syntax rule engine

knowledge

knowledge

resultsdatah

ealt

h c

are

info

rmat

ion

sys

tem

results

reporting toolsknowledge administration

data

inte

rfac

es2)

Arden Syntax server 1)

1) integrated, local, or remote2) local and web services, web frontend

Arden Syntax development & test

environment

data & knowledge services center

Arden Syntax rule engine

knowledge

knowledge

resultsdatah

ealt

h c

are

info

rmat

ion

sys

tem

results

reporting toolsknowledge administration

data

inte

rfac

es2)

Arden Syntax server 1)

1) integrated, local, or remote2) local and web services, web frontend

Fuzzy Arden Syntax: Modelling uncertainty in medicine

• linguistic uncertainty

‒ due to the unsharpness (fuzziness) of boundaries of linguistic concepts; gradual transition from one concept to another

‒ modeled by fuzzy sets, e.g., fever, increased glucose level

• propositional uncertainty

‒ due to the uncertainty (or incompleteness) of medical conclusions; includes definitional and causal, statistical and subjective relationships

‒ modeled by truth values between zero and one, e.g., usually, almost confirming

Crisp sets vs. fuzzy sets

yes/no decision

gradual transition

age

1

χY young

0 threshold

U = [0, 120]Y ⊆ U with Y = {(µY (x)/ x)x ∈ U}µY: U → [0, 1]

11 + (0.04 x)2

∀ x ∈ U0 age

1

µY young

threshold0

U = [0, 120]Y ⊆ U with Y = {(χY (x)/ x)x ∈ U}χ Y: U → {0, 1}

χ Y (x) = ∀ x ∈ U0 x > threshold1 x ≤ threshold

1 x ≤ threshold

x > thresholdµY (x) =

0

Crisp sets vs. fuzzy sets

age

1

χY young

0 threshold

0 age

1

µY young

threshold0

0

“arbitrary” yes/no decisions• cause of unfruitful

discussions• often simply wrong

“intuitive” gradual transitions

x50

0.50

1.00

µA (x)

100 150 200

µ↑ (x) = 0.82

highlydecreased

↓↓decreased

↓normal

⊥increased

↑

highlyincreased

↑↑

[mg/dl]

µ⊥ (x) = 0.18

µ↓↓ (x) = 0.00µ↓ (x) = 0.00µ↑↑ (x) = 0.00

glucose level in serum of 130 mg/dl

0.00

Degree of compatibility [= degree of membership]

Integration into i.s.h.medat the

Vienna General Hospital

SOP checkingin melanoma patients

receiving chemotherapy

Towards a science of clinical medicine

patient’s medical data and

healthcare processes for

machine processing

patient’s medical data and

healthcare processes for

human processing

≠

“Measure what is measurable, and make measurable what is not so.”Galileo Galilei

1564–1642

Crucial point in clinical medicine:

“Digitize what is digitizable, and make digitizable what is not so.”Klaus-Peter Adlassnig

observations measurementse.g., temperature chart e.g., CRP

skin color (jaundice, livid, …) color measurement… …

The medical world becomes flat

after Thomas L. FriedmanThe World is Flat.Penguin Books, 2006.

• in a local world‒ decision support will be part of clinical information systems

• in a global world‒ any activity—where we can digitize and decompose the value chain*,

and move the work around—will get moved around

* patient value chain: patient examination, diagnosis, therapy, prognosis, health caredecisions, patient care

Future of artificial-intelligence-augmented clinical medicine

predictable future

medical data collection, storage,

& distribution

clinical decision support

tomorrowtoday

043

21

41 1

HIS 1.0

HIS 2.0

personalized medicine

implants, prostheses,

robotics

Closing remark: formalism vs. reality

Pure mathematics is much easier to understand, much simpler, than the

messy real world!

Gregory Chaitin (2005) Meta Math!: The Quest for Omega,Pantheon Books, New York.

Clinical informatics deals with the “messy real patient”.