Machine learning models in production

Design and implementation of more than a hundred machine learning solutions and multi-omic integration applied to clinical decision-making, diagnosis and analytical inference.

The developments include preprocessing pipelines, ensemble models and deep networks integrated into production environments on AWS, TensorFlow, PyTorch and scikit-learn. It is data science carried through to production, with the governance required by Clinical processes and governed AI.

Among the models deployed: prediction of immune-related adverse events in immunotherapy, combining clinical, laboratory and imaging data; and the patient complexity stratification, with multimorbidity scoring algorithms and health determinants.

It also selection of biomarkers in Alzheimer's disease on genomic, proteomic and neuroimaging data; the analysis of sperm quality through computer vision, with convolutional networks applied to morphology and motility.