SUMMARY: This review looks at the various ways people and companies have tried to use machine learning in medicine, specifically to predict cardiovascular issues and cancer. There are two types of machine learning: supervised and unsupervised. Supervised learning uses algorithms to look at data and classify the information. Unsupervised learning takes a set of data and looks for patterns. The problem with using machine learning in medicine is that there isn’t enough data to ensure a truly accurate prediction. Some of the successes have been on a smaller scale with confusing information being omitted. Questions such as liability and insurance arises as well. Who is responsible for a misdiagnosis? The doctor? Or the company who built the machine?
LESSON COMMENTS: The sections describing supervised versus unsupervised learning is easiest to read and introduces the two different types of machine learning. The authors note the pros and cons of both. This article can be used in a computer science class or any class learning about or using machine learning open source code. The answers raised by this review can also be used in biology classes to discuss the role technology plays in medicine. Math and statistics teachers can also use this paper, though since I’m not a math teacher, I will refrain from giving too much advice on this topic. I would think that the statistical analysis and graphs would be particularly useful, but that’s the extent of my math-teaching expertise.
Deo R. C. (2015). Machine Learning in Medicine. Circulation, 132(20), 1920–1930. doi:10.1161/CIRCULATIONAHA.115.001593