Usefulness of Random Forest Algorithm in Predicting Severe Acute Pancreatitis

ConclusionsAn interpretable RF model exhibited the highest discriminatory performance in predicting SAP. Interpretation with LIME plots could be useful for individualized prediction in a clinical setting. A nomogram consisting of albumin, serum creatinine, glucose, and pleural effusion was useful for prediction of SAP.
Source: Frontiers in cellular and infection microbiology - Category: Microbiology Source Type: research