“victor babes” university of medicine and pharmacy timisoara department of medical informatics...

24
Victor Babes” Victor Babes” UNIVERSITY OF MEDICINE UNIVERSITY OF MEDICINE AND PHARMACY AND PHARMACY TIMISOARA TIMISOARA DEPARTMENT OF DEPARTMENT OF MEDICAL INFORMATICS AND BIOPHYSICS MEDICAL INFORMATICS AND BIOPHYSICS Medical Informatics Division Medical Informatics Division www.medinfo.umft.ro/dim www.medinfo.umft.ro/dim 2004 / 2005 2004 / 2005

Post on 19-Dec-2015

228 views

Category:

Documents


5 download

TRANSCRIPT

Page 1: “Victor Babes” UNIVERSITY OF MEDICINE AND PHARMACY TIMISOARA DEPARTMENT OF MEDICAL INFORMATICS AND BIOPHYSICS Medical Informatics Division

““Victor Babes” Victor Babes” UNIVERSITY OF MEDICINE UNIVERSITY OF MEDICINE

AND PHARMACY AND PHARMACY TIMISOARATIMISOARA

““Victor Babes” Victor Babes” UNIVERSITY OF MEDICINE UNIVERSITY OF MEDICINE

AND PHARMACY AND PHARMACY TIMISOARATIMISOARA

DEPARTMENT OFDEPARTMENT OF

MEDICAL INFORMATICS AND BIOPHYSICSMEDICAL INFORMATICS AND BIOPHYSICS

Medical Informatics DivisionMedical Informatics Divisionwww.medinfo.umft.ro/dimwww.medinfo.umft.ro/dim

2004 / 20052004 / 2005

Page 2: “Victor Babes” UNIVERSITY OF MEDICINE AND PHARMACY TIMISOARA DEPARTMENT OF MEDICAL INFORMATICS AND BIOPHYSICS Medical Informatics Division

MEDICAL DECISION MEDICAL DECISION SUPPORT (II)SUPPORT (II)

COURSE 12COURSE 12

Page 3: “Victor Babes” UNIVERSITY OF MEDICINE AND PHARMACY TIMISOARA DEPARTMENT OF MEDICAL INFORMATICS AND BIOPHYSICS Medical Informatics Division

4. ELEMENTS OF LOGICS4. ELEMENTS OF LOGICS 4.1. GENERAL NOTIONS4.1. GENERAL NOTIONS

• a) SENTENCEa) SENTENCE– expressing ideas about different objects or expressing ideas about different objects or

events features or causes events features or causes – TYPES: cognitive, interogative, imperative...TYPES: cognitive, interogative, imperative...– TRUTH VALUE : TRUTH VALUE :

• True (T), False (F), Uncertain (?)True (T), False (F), Uncertain (?)

• b) LOGIC FORMS :b) LOGIC FORMS :– notion, sentence, inference– logic form of a declarative affirmative

sentence: S is P (S=subject, P=predicate)

Page 4: “Victor Babes” UNIVERSITY OF MEDICINE AND PHARMACY TIMISOARA DEPARTMENT OF MEDICAL INFORMATICS AND BIOPHYSICS Medical Informatics Division

• c) PRINCIPLES OF LOGICSc) PRINCIPLES OF LOGICS– identity principleidentity principle

– non-contradiction principlenon-contradiction principle

– excluded tertiary principleexcluded tertiary principle

– sufficient rationale principlesufficient rationale principle

• d) CATEGORICAL SENTENCES:d) CATEGORICAL SENTENCES:– Universal - affirmative/negativeUniversal - affirmative/negative

• All S are P. None S is P.

– Particular - affirmative, negativeParticular - affirmative, negative• Some S are P. Some S are not P.

Page 5: “Victor Babes” UNIVERSITY OF MEDICINE AND PHARMACY TIMISOARA DEPARTMENT OF MEDICAL INFORMATICS AND BIOPHYSICS Medical Informatics Division

4.2. COMPOSED SENTENCES4.2. COMPOSED SENTENCES• Applying an operator on one or two simple sentences (unary or binary operators) • Truth value of composed sentences:• NegationNegation pp 11 00

(not)(not) ~p ~p 00 11

• ConjunctionConjunction q q pp 11 00 (p (p q) q) 11 11 00 (and)(and) 00 00 00

Page 6: “Victor Babes” UNIVERSITY OF MEDICINE AND PHARMACY TIMISOARA DEPARTMENT OF MEDICAL INFORMATICS AND BIOPHYSICS Medical Informatics Division

• DisjunctionDisjunction q q pp 11 00

(p (p q) q) 11 11 11

(or)(or) 00 11 00

• ImplicationImplication q q pp 11 00

(p (p q) q) 11 11 11

(if…then...)(if…then...) 00 00 11

Page 7: “Victor Babes” UNIVERSITY OF MEDICINE AND PHARMACY TIMISOARA DEPARTMENT OF MEDICAL INFORMATICS AND BIOPHYSICS Medical Informatics Division

• Exclusive Exclusive q q pp 11 00

disjunction disjunction 11 00 11

(or…or...) XOR(or…or...) XOR 00 11 00

• EquivalenceEquivalence q q pp 11 00

(p (p q) q) 11 11 00

(if and only if)(if and only if) 00 00 11

Page 8: “Victor Babes” UNIVERSITY OF MEDICINE AND PHARMACY TIMISOARA DEPARTMENT OF MEDICAL INFORMATICS AND BIOPHYSICS Medical Informatics Division

4.3. LOGICAL INFERENCE4.3. LOGICAL INFERENCE• Structure: premises (2 sen)Structure: premises (2 sen) conclusionconclusion

• ModusModus (p (p q) q) qq ponensponens p p

• Modus Modus (p (p q) q) ~ p~ p tollenstollens ~ q ~ q

• SillogismSillogism (p (p q) q) (p (p r) r)(q (q r) r)

Page 9: “Victor Babes” UNIVERSITY OF MEDICINE AND PHARMACY TIMISOARA DEPARTMENT OF MEDICAL INFORMATICS AND BIOPHYSICS Medical Informatics Division

4.4. Ex: PROLOG LANGUAGE4.4. Ex: PROLOG LANGUAGEdomains

person, activity = symbol

predicates

likes (person, activity)

clauses

likes (ellen, tennis)

likes (tom, baseball)

likes (bill, X) if likes (tom, X)

RUN: goal: likes (bill, baseball)

TRUETRUE

Page 10: “Victor Babes” UNIVERSITY OF MEDICINE AND PHARMACY TIMISOARA DEPARTMENT OF MEDICAL INFORMATICS AND BIOPHYSICS Medical Informatics Division

5. HEURISTIC METHODS5. HEURISTIC METHODS:: EXPERT EXPERT SYSTEMSSYSTEMS

5.1. SCHEME5.1. SCHEME

Page 11: “Victor Babes” UNIVERSITY OF MEDICINE AND PHARMACY TIMISOARA DEPARTMENT OF MEDICAL INFORMATICS AND BIOPHYSICS Medical Informatics Division

5.2. COMPONENTS5.2. COMPONENTS

• a) COGNITIVE SYSTEMa) COGNITIVE SYSTEM

–KNOWLEDGE BASE: 3 levelsKNOWLEDGE BASE: 3 levels• FACTUALFACTUAL

• CONCEPTUALCONCEPTUAL

• (meta-knowledge)(meta-knowledge)

Page 12: “Victor Babes” UNIVERSITY OF MEDICINE AND PHARMACY TIMISOARA DEPARTMENT OF MEDICAL INFORMATICS AND BIOPHYSICS Medical Informatics Division

• b) REASONING SYSTEM b) REASONING SYSTEM (inference machine)(inference machine)

• c) COMMUNICATION SYSTEM c) COMMUNICATION SYSTEM (user interface)(user interface)

• d) EXPLANATORY SYSTEMd) EXPLANATORY SYSTEM

• [ e) Meta-resolutive system[ e) Meta-resolutive system– to check if results are valid and to check if results are valid and

reasoning is adequate) ]reasoning is adequate) ]

Page 13: “Victor Babes” UNIVERSITY OF MEDICINE AND PHARMACY TIMISOARA DEPARTMENT OF MEDICAL INFORMATICS AND BIOPHYSICS Medical Informatics Division

5.3. COGNITIVE SYSTEM AND 5.3. COGNITIVE SYSTEM AND KNOWLEDGE BASEKNOWLEDGE BASE

• Medical knowledge extraction:Medical knowledge extraction:– general knowledgegeneral knowledge– knowledge from clinical experienceknowledge from clinical experience

• Medical knowledge formalizationMedical knowledge formalization• KB may be extendedKB may be extended• Self-completion (PROLOG language – Self-completion (PROLOG language –

appropriate for “Artificial Intelligence)appropriate for “Artificial Intelligence)• May be used either for decision support or for May be used either for decision support or for

educational purposes)educational purposes)

Page 14: “Victor Babes” UNIVERSITY OF MEDICINE AND PHARMACY TIMISOARA DEPARTMENT OF MEDICAL INFORMATICS AND BIOPHYSICS Medical Informatics Division

• 5.4. REASONING SYSTEM5.4. REASONING SYSTEM– leading position - drives the dialogueleading position - drives the dialogue

– may accept statistical proceduresmay accept statistical procedures

• 5.5. EXPLANATORY SYSTEM 5.5. EXPLANATORY SYSTEM

• shows the reasoning ‘trace’shows the reasoning ‘trace’– educational purposeseducational purposes

• 5.6. COMMUNICATION SYSTEM5.6. COMMUNICATION SYSTEM– natural language usenatural language use

Page 15: “Victor Babes” UNIVERSITY OF MEDICINE AND PHARMACY TIMISOARA DEPARTMENT OF MEDICAL INFORMATICS AND BIOPHYSICS Medical Informatics Division

5.7. MEDICAL Expert Systems5.7. MEDICAL Expert Systems

• MYCIN - bacterial infections

• PUFF - pulmonary deseases

• HEADMED - neuro-psychiatry

• CASNET - ophtalmology

• VM -’ventilator monitor’

• INTERNIST - internal medicine

• TROPICAID - tropical diseases

• Domain-independent E.S.

Page 16: “Victor Babes” UNIVERSITY OF MEDICINE AND PHARMACY TIMISOARA DEPARTMENT OF MEDICAL INFORMATICS AND BIOPHYSICS Medical Informatics Division

a) Classification table Classifier T+ T-

D+ N11 N12 R1

D- N21 N22 R2

C1 C2 N

Real

6. CLASSIFICATION6. CLASSIFICATION Q QUALITYUALITYASSESSMENTASSESSMENT

Page 17: “Victor Babes” UNIVERSITY OF MEDICINE AND PHARMACY TIMISOARA DEPARTMENT OF MEDICAL INFORMATICS AND BIOPHYSICS Medical Informatics Division

b) PARAMETERSb) PARAMETERS• true positive true positive = TP = N= TP = N11 11

• false negativefalse negative = FN = N = FN = N12 12

• false positivefalse positive = FP = N = FP = N2121

• true negative true negative = TN = N= TN = N2222

• sensitivitysensitivity = p(T+/D+) = N = p(T+/D+) = N11 11 / R/ R11

– (capacity to detect the diseased)(capacity to detect the diseased)

• specificityspecificity = p(T-/D-) = N = p(T-/D-) = N2222 / R / R2 2

– (capacity to reject the non-diseased)(capacity to reject the non-diseased)

Page 18: “Victor Babes” UNIVERSITY OF MEDICINE AND PHARMACY TIMISOARA DEPARTMENT OF MEDICAL INFORMATICS AND BIOPHYSICS Medical Informatics Division

• positive predicitve valuepositive predicitve value

ppv = p(D+/T+) = ppv = p(D+/T+) = NN11 11 / / CC11

• negative predictive valuenegative predictive value

npv = p(D-/T-) = npv = p(D-/T-) = NN2222 / / CC22

Global indicatorsGlobal indicators• accuracyaccuracy = (N = (N11 11 + N+ N2222) / N) / N

• classification classification error rateerror rate = (N = (N1212 + N + N2121) / N) / N

Page 19: “Victor Babes” UNIVERSITY OF MEDICINE AND PHARMACY TIMISOARA DEPARTMENT OF MEDICAL INFORMATICS AND BIOPHYSICS Medical Informatics Division

Threshold valueThreshold value

Page 20: “Victor Babes” UNIVERSITY OF MEDICINE AND PHARMACY TIMISOARA DEPARTMENT OF MEDICAL INFORMATICS AND BIOPHYSICS Medical Informatics Division
Page 21: “Victor Babes” UNIVERSITY OF MEDICINE AND PHARMACY TIMISOARA DEPARTMENT OF MEDICAL INFORMATICS AND BIOPHYSICS Medical Informatics Division
Page 22: “Victor Babes” UNIVERSITY OF MEDICINE AND PHARMACY TIMISOARA DEPARTMENT OF MEDICAL INFORMATICS AND BIOPHYSICS Medical Informatics Division

d) ROC Curved) ROC Curve SN = f(1 – SP)SN = f(1 – SP) Receiver – Operator CharacteristicReceiver – Operator CharacteristicSN si SP vary inverslySN si SP vary inversly p proportionalroportional

Page 23: “Victor Babes” UNIVERSITY OF MEDICINE AND PHARMACY TIMISOARA DEPARTMENT OF MEDICAL INFORMATICS AND BIOPHYSICS Medical Informatics Division

cc) ) ExampleExample

• A study was carried on a population of 3000 A study was carried on a population of 3000 individuals. 500 of them had virosis during individuals. 500 of them had virosis during last year. Our computer program was able to last year. Our computer program was able to detect 480 of them, but gave the same detect 480 of them, but gave the same diagnosis to another 50 persons. Compute: diagnosis to another 50 persons. Compute:

• number of number of false positivefalse positivess and false negativesand false negatives• sensitivity sensitivity and and specificity specificity • accuracy accuracy and and classification error rateclassification error rate• positive and negative predictive ratepositive and negative predictive rate

Page 24: “Victor Babes” UNIVERSITY OF MEDICINE AND PHARMACY TIMISOARA DEPARTMENT OF MEDICAL INFORMATICS AND BIOPHYSICS Medical Informatics Division

EndEnd