Artificial Intelligence Identifies Risky Alcohol Use Before Surgery, Study Finds

Alcohol consumption prior to surgery can lead to risky complications, but identifying signs of dangerous alcohol use is not always straightforward. However, a new analysis suggests that artificial intelligence (AI) could help shed light on this issue. According to a study published in the journal Alcohol: Clinical & Experimental Research, researchers used a natural language processing model to examine the medical records of 53,811 patients who underwent surgery between 2012 and 2019.

Electronic medical records contain diagnostic codes, but they also provide additional information such as notes, test results, and billing data that could indicate risky alcohol use. To identify contextual clues, the researchers programmed a natural language processing model to detect both diagnostic codes and other indicators of risky alcohol use, such as exceeding recommended drink thresholds per week or a history of alcohol-related medical issues.

The study found that misusing alcohol prior to surgery is associated with higher infection rates, longer hospital stays, and other surgical complications. Among the patients analyzed, 4.8% had diagnosis codes linked to alcohol use. However, with the help of contextual clues, the model identified three times as many patients at risk, bringing the total to 14.5%.

Interestingly, the model’s accuracy was comparable to that of a panel of human alcohol-use experts. The model matched the experts’ classifications for a subset of records 87% of the time. These findings have led researchers to suggest that AI could be a valuable tool for clinicians seeking to identify patients who may require intervention or postoperative support.

V.G. Vinod Vydiswaran, the lead author of the study and an associate professor of learning health sciences at the University of Michigan Medical School, highlights the potential of AI in primary care and beyond. He believes that this technology could help providers see relevant information contained in the notes made by other healthcare professionals, without the need to read the entire medical record. Vydiswaran suggests that this analysis might lay the groundwork for future efforts to identify other risks in healthcare with proper validation.

While the researchers plan to eventually make the model public, they note that it will need to be trained on medical records from individual facilities. The implementation of AI could improve patient outcomes by enabling healthcare providers to identify risky alcohol use before surgery and provide appropriate interventions as necessary.

In conclusion, AI has the potential to assist clinicians in detecting risky alcohol use in patients before surgery. By analyzing medical records and identifying contextual clues, AI can help identify individuals at risk of complications associated with alcohol consumption. This analysis opens doors for future advancements in healthcare risk identification and underscores the value of AI as a tool for healthcare providers.