A probabilistic framework for predicting disease dynamics: A case study of psychotic depression

Marcos L. P. Bueno, Arjen Hommersom, Peter J. F. Lucas, Joost G. E. Janzing

Abstract: Highlights•A methodology that facilitates hypothesis generation in medical domains is proposed.•Structured hidden Markov models are proposed as effective models.•Predictive models for psychotic depression treatment are built.•Association between psychosis and depression and patient recovery is demonstrated.•Patient sensitivity to intervention is uncovered.