Machine Learning for Precision Medicine and Humanized AI for Future Healthcare

Machine learning (ML) and artificial intelligence (AI) technology have been widely used in drug discovery for over 25 years, while applications in the pharmaceutical industry beyond drug discovery are developed much later, and the scope is still limited, especially in clinical trials. ML/AI is recognized as a major innovative approach after the wave of adaptive clinical trial designs. We will provide an overview of AI/ML in the pharmaceutical industry and healthcare; discuss the challenges in classical statistics, including why all one-sided rank-based test methods are practically invalid, and the necessity of a paradigm shift to AI; and provide an example of similarity-based machine learning in clinical trials for the purpose of precision medicine. In healthcare, humanized AI (HAI), as the ultimate version of artificial general intelligence (AGI), will become a viable solution; HAI looks, thinks, and behaves like a human. We will discuss the foundation, architecture, and prototyping of HAI, including dynamic knowledge representation, attention, learning, and response mechanisms.