A new analysis suggests that artificial intelligence could help bring risky alcohol use to light in patients’ medical records. The study, published in Alcohol: Clinical & Experimental Research, used a natural language processing model to identify patients at risk for surgical complications due to alcohol use. The model examined medical records of over 53,000 surgical patients and identified both diagnostic codes and contextual clues of risky alcohol use. It found that 4.8 percent of patients had diagnosis codes related to alcohol use, but with the help of contextual clues, the model classified 14.5 percent as being at risk. The findings suggest that AI could be a potential partner for clinicians in identifying patients who may need intervention or postoperative supports. The researchers also noted that the model could potentially be used to identify other risks in primary care and beyond. The lead author of the study, V.G. Vinod Vydiswaran, expressed that the model could assist healthcare providers in highlighting information contained in the patient’s notes without having to read the entire record. The researchers plan to eventually make the model public, but it will have to be trained on medical records from individual facilities. This research highlights the potential of AI in healthcare for improving patient care and outcomes.