Machine Learning for Personalized Medicine

Scientific lecture by Felix Agakov: Machine Learning for Drug Development and Personalized Medicine

At the beginning of the -omics era, there was a great hope that developments in molecular biology will revolutionize drug development and personalized medicine. Today, many practitioners would argue that effects of the developments in genomics on producing better personalized therapies have been rather limited. We will speak about industries' view of the key bottlenecks of drug development and personalized medicine, and our vision of how machine learning can be useful for addressing some of them. We will highlight some of the segments within the personalized medicine sector that are perceived to grow at the fastest rate over the next 5 years, and practical challenges industries may face in implementing machine learning solutions as products. We will then speak in some detail about what is currently being done for addressing three important problems of analysing -omics data, with the specific focus on individual-level predictions from -omics data and clinical factors, discovery and validation of predictive and causal biomarkers, and learning structures of sparse biological networks in the presence of latent confounders. We will highlight some common challenges and mistakes in analysing -omics data, and review several other tasks which may help to understand taxonomy of disease and discover better personalized therapies.

Return to the overview of lectures ...