AI can flag patients for dangerous alcohol use before surgery

AI can flag patients for dangerous alcohol use before surgery

Alcohol can cause serious surgical complications for patients who drink in the days before the procedure, but signs of serious alcohol abuse are not always evident on a patient’s chart. New analysis suggests that artificial intelligence could help shed light on such problems.

The study, published in the journal Alcohol: Clinical and Experimental Research, used a natural language processing model to evaluate the medical records of 53,811 patients who underwent surgery between 2012 and 2019.

Patient electronic medical records contain diagnostic codes, but can also include information such as notes, test results, or billing data that may indicate serious alcohol use.

To identify contextual clues, the researchers programmed a natural language processing model to identify both diagnostic codes and other indicators of hazardous alcohol use, such as drinks exceeding recommended weekly thresholds or a history of medical problems associated with alcohol misuse.

Alcohol misuse during surgery is associated with higher infection rates, longer hospital stays, and other surgical complications. Of the patients studied, 4.8% had charts that included a diagnosis code related to alcohol abuse. With the help of contextual clues, the model classified three times as many people as being at risk, a total of 14.5%.

The model also worked well with a panel of human alcohol abuse experts, matching their ratings to a subset of records 87 percent of the time.

The findings point to AI as a potential partner for doctors looking to identify patients who need postoperative intervention or support, the researchers concluded.

The analysis could “lay the foundation for efforts to identify other risks in primary care and beyond, with appropriate validation,” said Vinod Videswaran, assistant professor of health sciences learning at the University of Michigan Medical School and lead author of the study. press release. “Basically, this is a way to highlight what is already in the notes provided by other providers, without having to read the entire log.”

The researchers say they plan to eventually deploy the model, but note that it will have to be trained on medical records from individual facilities.

Automated detection of preoperative hazardous alcohol use using natural language processing

Alcohol: clinical and experimental research

You may also like...

Leave a Reply

Your email address will not be published. Required fields are marked *