Skip to main content
. 2024 Nov 27;24:360. doi: 10.1186/s12911-024-02776-w

Table 3.

Top ten features with the highest important scores of the best performance XGBoost case definition

Feature Code Description Data type Importance score
One occurrence of text “rheumatoid arthritis” Billing, encounter diagnosis or problem list 0.2395
One occurrence of text “rheumatoid” Billing, encounter diagnosis or problem list 0.1461
Two occurrences of text “rheumatoid” Billing, encounter diagnosis or problem list 0.0562
Two occurrences of text “arthritis” Billing, encounter diagnosis or problem list 0.0297
Two occurrences of text “rheumatoid” in 24 months Billing, encounter diagnosis or problem list 0.0295
Two occurrences of text “other inflammatory” Billing, encounter diagnosis or problem list 0.0228
One ATC code P01BA02 Hydroxychloroquine Medication 0.0225
Two occurrences of text “arthritis” Problem list 0.0224
One ATC code B03BB01 Folic acid Medication 0.0197
Two occurrences of text “rheumatoid arthritis” in 24 months Billing, encounter diagnosis or problem list 0.0189

“Or any sub-codes” means all codes starting with the same characters (e.g., B03 (or any sub-codes) includes B03A, B03B, B03BB, etc.)

ATC Anatomical Therapeutic Chemical (ATC) Classification, ICD-9 International Classification of Diseases, Ninth Revision