Learning clinical networks from medical records based on information estimates in mixed-type data

Hervé Isambert (Institut Curie)
Thursday, October 17, 2019 - 10:30
Room Aurigny
Talk abstract: 

Network reconstruction aims at disentangling direct from indirect dependences in information-rich data and has become ubiquitous to analyze the rapidly expanding resources of genomic and clinical data. However, direct and indirect interdependences in mixed-type (continuous / categorical) clinical data are notoriously difficult to assess. To this end, we developed and implemented an efficient computational approach to simultaneously compute and assess the significance of multivariate information between any combination of mixed-type variables. The method is then used to uncover direct, indirect and possibly causal relationships between mixed-type data from medical records, by extending a recent machine learning method to reconstruct graphical models beyond simple categorical datasets. The method is shown to outperform existing tools on benchmark mixed-type datasets, before being applied to analyze the medical records of eldery patients with cognitive disorders from La Pitié-Salpêtrière Hospital, Paris, and breast cancer patients from Institut Curie hospitals.