HackAuton

Patient Safety Track Winner PatientWatch team members Michael Byrant and Kevin Deng on stage.

HackAuton held a patient safety technology track, sponsored by the Patient Safety Technology Challenge funded by the Pittsburgh Regional Health Initiative. Held on September 15-17th, HackAuton was hosted by the Auton Lab at Carnegie Mellon University, a leading research facility focused on artificial intelligence (AI) and machine learning (ML). The Patient Safety Technology Challenge was especially excited to sponsor HackAuton, as Auton Lab is an inaugural partner of the Challenge with Professor Artur Dubrawski, who leads the lab, also serving as an advisory board member.

While there was no monetary prize awarded, there were innovative ideas about how to apply AI/ML to patient safety. The top team created Prescription Evaluation using GPT4 and openFDA. The demo of a prescription checker used AI to reduce human oversight and improve effectiveness.

Team member Charlie Sun’s mother is a doctor. Through her shared experience he recognized the need to address medication errors. Charlie shared:

“One of the doctors prescribed a medication to a patient. On the prescription she wrote: ‘xxx tablet, 1/4 each time, 3 times a day.’ What she meant was the dosage should be 1/4 of a tablet each time. However, as it arrived at the hospital pharmacy, the staff who read and retrieved the medicine understood it as "1/4 mg each time, 3 times a day", where 1/4 mg is exactly the amount of one tablet. So effectively the dosage got increased 4-fold! Fortunately, that particular medicine was not too dangerous, and the new dosage would not cause too much adverse effects, but the patient still experienced a lot of discomfort. I think all these errors could be reduced and alleviated by the usage of innovative technologies, so I am eager to explore how we can apply new technologies to this important task.”

Errors like these can have long-lasting impacts on patients and Prescription Evaluation’s idea could stop them from happening in the future.

Team member Minh Tran shared that the team is excited to continue their work. “We're thinking about polishing the research paper on more datasets and publishing it. We could also improve the demo with voice2text,” said Tran.

The Patient Safety Technology Challenge is excited to see what the team does next.