Machine learning techniques for predicting complications and evaluating drugs efficacy | Machine Learning for Personalized Medicine

Research Talk by Michal Rosen-Zvi: Machine learning techniques for predicting complications and evaluating drugs efficacy

Watch the video at videolectures.net:
Machine learning techniques for predicting complications and evaluating drugs efficacy

In this talk we focus on potential contribution of machine learning methods to healthcare and focus on the somewhat new trend called real world evidence or post launch monitoring. We review the machine learning paradigm of supervised learning. The value of ensemble methods and generative and discriminative prediction algorithms is discussed. Examples of how these methods can be utilized for making better informed medical decision will be reviewed. Finally, the talk includes results of studies performed on longitudinal patients' data of patients from three different disease areas. 1. Studies of data of 50K European HIV patients. 2. Studies of American diabetic patients and 3. An analysis of more than 1M epilepsy patients.