Skip to main content

Table 2 Feature importance for both logistic regression and random forest. GP = general practitioner, NS = non-steroid

From: Detection of primary Sjögren’s syndrome in primary care: developing a classification model with the use of routine healthcare data and machine learning

  Logistic Regression Random Forest
1. S01X (other ophthalmologicals) Age
2. Age S01X (other ophthalmologicals)
3. Gender Number of GP consults < 20 min
4. S01A (anti-infectives for ophthalmological use) Number of GP consults > 20 min
5. Number of GP consults > 20 min Number of GP consults by phone
6. S01G (decongestants and anti-allergics for ophthalmological use) A02B (drugs for peptic ulcer and gastro-oesopheagel reflux disease)
7. Number of GP visitations at home < 20 min Gender
8. S01C (anti-inflammatory agents and anti-infectives in combination) Repeat prescription
9. J01F (macrolides, lincosamides and streptogramins) N02A (opioids)
10. A99 (other generalized/non-specified diseases) M01A (Anti-inflammatory and anti-rheumatic products, NS)