We propose a generative statistical model to discover latent treatment patterns from EMRs.
The model unveils the latent associations between comorbidities and treatments.
A medical application is proposed in terms of treatment recommendation.
The model is verified on an unstable angina CP dataset.