A team of researchers at the Icahn School of Medicine at Mount Sinai has developed a new method to identify and reduce biases in datasets used to train machine-learning algorithms-addressing a critical issue that can affect diagnostic accuracy and treatment decisions.