Medical record problem list generation

A non-transitory computer program product for generating a medical problem list, the computer program product comprising:

• a computer readable storage medium readable by a processing circuit and storing program instructions for execution by the processing circuit for performing a method comprising:
• receiving a plurality of disease categories;
• defining a disease category set comprising a plurality of top level disease categories selected from a disease category classification list, the disease category classification list comprising a single tree hierarchy;
• extracting a plurality of candidate training problems from an electronic patient record training set according to an ontology;
• assigning each of the candidate training problems to the plurality of top level disease categories;
• generating a plurality of disease category templates, wherein each disease category template represents one top level disease category of the plurality of top level disease categories;
• generating a unique disease category model for each disease category template from the electronic patient record training set by a machine learning technique, wherein each of the disease category models identifies, for the respective top level disease category, a set of attributes statistically determined to distinguish between a candidate problem that is a medical problem and a candidate problem that is not a medical problem;
• and
• wherein each disease category model is generated through a modified alternating decision tree (ADT) configured such that a categorical attribute performs a multiclass split that generates one prediction node for each category.