We've been using "lite" ML for phenotype adjudication in electronic health records with mild success. Random forests and support vector machines will outperform

We've been using "lite" ML for phenotype adjudication in electronic health records with mild success. Random forests and support vector machines will outperform simple linear regression when disease symptoms/progression don't neatly map to hospital billing codes.