Table 4.
Logistic regression coefficients from using the model trained with notes, ICDa codes, and medications. Unexpected results are discussed in the Error Analysis section. Small coefficients for some input variables were expected and removed at the time of model deployment.
| Features | Logistic regression coefficients |
| CHFb | 2.39 |
| Congestive heart failure | 1.89 |
| With CHF | 1.34 |
| CHF father | –1.15 |
| Histori congestive heart failure | 0.94 |
| CHF mother | –0.80 |
| No CHF | –0.72 |
| I50 | 0.68 |
| Dxc CHF | 0.64 |
| Exacerb CHF | 0.61 |
| With congestive heart failure | –0.60 |
| [] CHF | –0.54 |
| Ascites | 0.54 |
| () Congestive heart failure | 0.52 |
| Diagnosis congestive heart failure | 0.49 |
aICD: International Classification of Diseases.
bCHF: congestive heart failure.
cDx: diagnosis.