Course Description
Machine Learning has achieved expert-level accomplishment over the last years: self-driving cars, face & voice recognition, Go & Chess playing, translation … Nevertheless, the tools and methods backing these breakthroughs have not shown similar momentum in Healthcare despite the considerable efforts of the scientific and industrial communities. Healthcare is an extremely challenging field for Machine Learning – thus full of scientific opportunities – as it wraps three ingredients: – Diverse mathematical applications: disease diagnosis, cell segmentation, prediction of age at disease onset, clustering of patients or symptoms, simulation, generation, etc – The incredibly large variety of data: biomarkers, imaging data, time-series, genomics, graphs, text, etc. – Case-specific constraints: small datasets, high dimensional 3D images, real-time usage, anonymization, false-positive importance, unbalanced datasets, etc. The course aims at first introducing the particularities that Machine & Deep Learning faces when dealing with medical applications. The second part of the lecture will focus on state-of-the-art models and new paradigms developed in the scientific community to tackle these challenges. In a nutshell, the lecture is suited to both neophytes interested in getting a general overview of Healthcare applications and also to advanced ML practitioners willing to expose themselves to medical challenges. Topics:- Medical data,
- Medical imaging,
- Deep Learning,
- Disease Modelling,
- Cross-sectional & longitudinal data
Course tools
- Python
Prerequisites
- Calculous & Algebras
- At ease with standard ML algorithms & objects (tensors, distributions, loss, energy, …)