Machine Learning and Medical Risk
November 9, 2020
I spotted an AAAS article called “Machine Learning Shows Similar Performance to Traditional Risk Prediction Models.” The information can be interpreted in different ways, depending upon one’s point of view. For example, machine learning misses the boat, or machine learning works about as well as humans fumbling along. One sentence warranted a blue exclamation point:
cardiovascular disease risk predictions for the same patients varied substantially between models, especially in patients with higher risks. For example, a patient with a cardiovascular disease risk of 9.5-10.5% predicted by the traditional QRISK3 model had a risk of 2.9-9.2% and 2.4-7.2% predicted by other models. Models that ignored censoring (including commonly used machine learning models) substantially underestimated risk of cardiovascular disease.
The report begs another question: “What other machine learning models underestimate risk?”
Stephen E Arnold, November 9, 2020
Comments
-
- Subscribe to Beyond Search
Feature archive
News archive
-
