Business Context
A pharmaceutical company analyzes clinical trial notes. They use a two-step process: first, extract specific drug entities (Concepts), then categorize the notes based on the presence of these concepts combined with other terms (e.g., 'DrugX' AND 'Headache'). This tests the integration of an existing Concept model into the Category compilation.
Data Preparation
Creation of concept rules (Drug definitions) and category rules that reference those concepts.
Copied!
| 1 | |
| 2 | DATA mycas.concept_defs; |
| 3 | LENGTH config $200; |
| 4 | INFILE DATALINES dsd; |
| 5 | INPUT config $; |
| 6 | DATALINES; |
| 7 | "CONCEPT:DrugA,(OR, 'Aspirin', 'Acetylsalicylic')" "CONCEPT:DrugB,(OR, 'Ibuprofen', 'Advil')" ; |
| 8 | |
| 9 | RUN; |
| 10 | |
| 11 | DATA mycas.category_complex_rules; |
| 12 | LENGTH config $200; |
| 13 | INFILE DATALINES dsd; |
| 14 | INPUT config $; |
| 15 | DATALINES; |
| 16 | "CATEGORY:AdverseEvent_A,(AND, C_CONCEPT:DrugA, 'nausea')" "CATEGORY:AdverseEvent_B,(AND, C_CONCEPT:DrugB, 'dizziness')" ; |
| 17 | |
| 18 | RUN; |
| 19 |
Étapes de réalisation
1
Step 1: Compile the Concept Model first (Prerequisite).
Copied!
| 1 | |
| 2 | PROC CAS; |
| 3 | textRuleDevelop.compileConcept / TABLE={name='concept_defs'} config='config' casOut={name='drug_concepts_li', replace=true}; |
| 4 | |
| 5 | RUN; |
| 6 |
2
Step 2: Compile the Category Model using the previously created Concept Model as input.
Copied!
| 1 | |
| 2 | PROC CAS; |
| 3 | textRuleDevelop.compileCategory / TABLE={name='category_complex_rules'} config='config' concept={name='drug_concepts_li'} casOut={name='adverse_event_model', replace=true}; |
| 4 | |
| 5 | RUN; |
| 6 |
Expected Result
The action should successfully link the 'drug_concepts_li' binary model with the text rules. The final 'adverse_event_model' will contain logic that depends on the successful detection of the underlying concepts.