Machine learning has amazing potential in the medical field. Automated analysis of large volumes of medical data could save lives and millions of dollars. However, blackmail and other threats creates an imperative to keep these records private. Data theft, model theft, and a host of other vulnerabilities may prevent machine learning from being employed for healthcare. Privacy-preserving ML using federated learning and homomorphic encryption could be the solution. The technology is still emerging but the initial results are promising and may lead to a new era in healthcare data management.