In this module, we discuss what is perhaps the most important factor that determines the success of AI algorithms: data. In particular, we will address some of the main challenges to overcome when considering the collection of, and working with, healthcare data. This module will also introduce AI techniques that comprise the supervised machine learning (ML) family. This section will walk you through the conceptual framework and fundamental math underlying these techniques. Finally, we will explore common approaches used to evaluate how well these types of models perform. At the end of this module, you should have a foundational understanding of the challenges facing healthcare data curation, mathematical approaches to formulating healthcare data science problems that involve diagnosis and prediction, and the most important fundamental approaches to ML in these problems, such as classification trees, and ensemble classifier methods.

What is artificial intelligence? Even before scientists invented the term, scholars such as Alan Turing and John von Neumann were wrestling with many of the mathematical and logical challenges that AI researchers still face today. Ever since the term was coined at a conference in the summer of 1957, scientists, philosophers, and policymakers have hotly debated how to define AI. However, because this is not a mathematics, computer science, or philosophy course, our goal in this module is to provide you with practical definitions of key terms associated with AI and provide a qualitative overview of common AI models. We do hope some of you will be inspired to take additional courses that will more rigorously explore topics covered in this module.

In this module, we will introduce the concept of generative machine learning (ML) models and explain how these models are different from the discriminative models we learned about in prior modules. If you don’t know what “discriminative models” means, keep reading and we’ll define the term before going much deeper. From this foundation, we will introduce the concept of generative AI and explore how it fits into the broader field of ML. After learning some basics about generative AI, we will dive into large language models (LLMs), learn how hallucinations are better considered errors.