In part one, we will review a concept of synthetic healthcare data that preserves individual privacy and thus enables a much wider effort to conduct research and develop applications to improve consumers’ health and wellbeing than would be possible by using real patient data. We will discuss various approaches to generate synthetic data and present a machine learning model that is capable of generating synthetic medical claims which comprise the main form of data used in health insurance industry.
Part two will offer an introduction to AI Ethics in healthcare.
We will review some of the sources of medical ethics, vulnerable populations, and the problem of bias in training data.
We will discuss appropriate use of sensitive variables, techniques for fairly training supervised learning algorithms, and selecting the best fairness metric.