Transcript Preliminary results

Coding of Selection Criteria
for Cancer Treatment Plans
Savvas Nikiforou
Part of the project:
Automated Matching of Patients
to Clinical Trials
Research Group
• Faculty
– Lawrence Hall
– Dmitry Goldgof
– Eugene Fink
• Students
–
–
–
–
Lynn Fletcher
Princeton Kokku
Savvas Nikiforou
Rochelle Harris
Motivation
• Selecting among treatment plans
• Minimizing pain and cost of the
selection process
• Reducing the physician’s effort
Expert System
• Guides the nurse through related questions
• Identifies the appropriate medical tests
Outline of the Talk
• Eligibility decisions
• Knowledge base
• Input of the knowledge
• Demonstration
Related Work
• Initial expert system
Fletcher, Hall and Goldgof, 1999
• Minimizing pain and cost
Kokku, Hall and Goldgof, 2001
Example: Selection Criteria
• Female, not older than 50
• Breast cancer, stage II
• No prior surgery
Example: Questions
Sex:
Female
Male
Age:
35
Example: Conclusion
Patient is not eligible
Example: Questions
Sex:
Female
Male
Age:
35
Example: Questions
Cancer stage:
Prior surgery? Yes
I
II
III
IV
No
Unknown
Example: Conclusion
Patient is eligible
Full Functionality
• Orders and groups the questions
• Considers multiple treatment plans
Old System
• A programmer has to code the questions
New System
• A programmer has to code the questions
• A nurse enters the questions through a
friendly interface
• Problem: Build the interface
Outline of the Talk
• Eligibility decisions
• Knowledge base
• Input of the knowledge
• Demonstration
Main Objects
• Questions
• Tests
• Eligibility criteria
Types of Questions
• Yes / No / Unknown
• Multiple choice
• Numeric
Examples of Questions
Prior surgery? Yes
No
Unknown
Examples of Questions
Prior surgery? Yes
Cancer stage:
I
II
III
IV
No
Unknown
Examples of Questions
Prior surgery? Yes
Cancer stage:
Age:
I
II
III
IV
No
Unknown
Tests
A medical test answers several questions.
It involves certain pain and cost.
Example Test: Name and Cost
Test Name:
Blood test
Cost:
42.23
Pain:
1
Example Test: Questions
• Yes / No
Question:
Histologically proven breast cancer?
Example Test: Questions
• Multiple choice
Question:
Patient’s clinical state
Options:
T-1
T-2
T-3
Example Test: Questions
• Numeric
Question:
White cell blood count
Min
0
Max
Prec
500000 0
Eligibility Criteria
• A logical expression that determines
eligibility for a specific treatment
Example: Criteria
AND
White Blood Count >100,000
No Heart Related Problems
OR
Cancer Stage I
Cancer Stage II
Outline of the Talk
• Eligibility decisions
• Knowledge base
• Input of the knowledge
• Demonstration
Input
• Questions
• Tests
• Eligibility criteria
Input: Test
• Name:
Type test name here
• Cost:
$$$$.$$
• Pain:
0-5
Input: Question Format
• Type
Yes / No
• Text
Multiple Choice
Numeric
Input: Yes / No Question
• Type
Yes / No
• Text
Is patient’s age less than 64?
Input: Multiple Choice Question
• Type
Multiple Choice
• Text
Options
Patient’s age
< 20
20 to 40
41 to 60
> 60
Input: Numeric Question
• Type
Numeric
• Text
Patient’s age
Min
Max
Prec.
0
120
0
Input: Eligibility
• Logical structure AND
Sex
• Questions
Age
OR
• Eligibility answers
Female
Male
40
From
65
To
Yes No
Tumors?
Yes No
Lesions?
Outline of the Talk
• Eligibility decisions
• Knowledge base
• Input of the knowledge
• Demonstration
Current System
Online Demo
Main results
• Formal model of selection criteria
and related optimization problems
• Visual representation of these criteria
• Friendly input interface
Coming soon
• Completion of the interface
• Converting the input knowledge
into internal logical structures