Machine Learning in Medicine

Motivated by advances in processing power, memory, storage, and an unprecedented wealth of data, this review introduces basic machine learning concepts for a medical audience. It notes that computers have recently mastered tasks once thought impossible, such as a popular variant of poker and video games, and that many companies now apply complex data analysis across industries, including healthcare. Using examples from the literature, the author explores which problems in medicine might benefit from such learning approaches.

A central observation is that seemingly adequate medical datasets and learning algorithms have been available for decades, yet despite thousands of papers applying machine learning to medical data, very few have contributed meaningfully to clinical care. This lack of impact contrasts sharply with machine learning's relevance to other industries, so the author works to identify the obstacles to changing medical practice through statistical learning and to discuss how these obstacles might be overcome.