Stroke Risk Algorithms Less Accurate for Black Patients, Study Finds

Stroke-specific machine learning algorithms were no better at predicting stroke risk than self-reported risk factors and were worse in predicting stroke risk in Black patients than in White patients, according to a new study published in JAMA Network.

The research, led by Duke University researchers, included nearly 63,000 middle-aged and senior stroke-free participants to evaluate the accuracy of various existing algorithms and two artificial intelligence assessment methods to predict a person’s stroke risk within the next ten years. The study also looked at risk ordering, which provides perspective on how likely someone is to experience stroke compared to others.

The study found that a simple method using patients’ answers to questions was the most accurate technique. This type of data collection includes social factors impacting stroke outcomes that can be left out of artificial intelligence algorithms, which can lead to inequity in risk assessment, the researchers said.