Researchers found that facial infrared thermography can predict coronary artery disease more accurately than traditional pretest probability tools, using deep-learning algorithms to analyze temperature patterns.