Interview with Andrew Lea on how experience with a 1970s computerized diagnostic program can inform efforts to implement machine-learning prediction models in medicine.