Tools & Strategies News

New AI System Assists Medication Selection for Type 2 Diabetics

Researchers used EHR data from several healthcare institutions to develop an artificial intelligence method that supported medication selection for 83 percent of type 2 diabetes patients.

Source: Getty Images

By Mark Melchionna

- Using EHR data from various healthcare institutions, Hitachi, University of Utah Health, and Regenstrief Institute collaborated to create an artificial intelligence (AI) method to enhance type 2 diabetes care.

To develop the AI method, researchers collected EHR data from Utah and Indiana hospitals and analyzed the information to create treatment plans for type 2 diabetes. Researchers used an AI method because there was a high volume of data coming from various healthcare institutions. The method was conducive to grouping and monitoring data.

The AI method works by grouping patients together based on similar disease statuses, and then analyzing their treatment and outcome patterns.

The algorithm proved successful, assisting in medication selection for over 83 percent of patients, even in cases where two or more medications were being used together.

Regarding limitations, researchers acknowledge that obtaining data from two sources, only using two demographics, using one evaluation model, and limited pre-processing, could have affected the data.

But the three organizations plan on continuously monitoring the progress of this AI algorithm, evaluating its effectiveness at assisting the selection of drug combinations given to patients.

According to the Centers for Disease Control and Prevention, 37.3 million American adults, or about 1 in 10, are diagnosed with diabetes. These organizations worked synchronously to create an AI method for care for this disease.

Externally, AI is increasingly being used to enhance clinical care frequently.  

Recently, researchers from the Yale School of Medicine developed an AI model to diagnose heart conditions. They collected extensive data from 1.5 million patients and used data from over 2 million electrocardiograms. Following the implementation of this model, researchers concluded that it was able to diagnose one in six patients with a heart rhythm disorder.

Another AI system created by researchers at Brigham and Women’s Hospital helps monitor patient reactions following a heart transplant procedure. Following this type of operation, it is typical for the immune system to naturally attack the new organ, which often goes unrecognized by the patient. The system used AI to detect any signs of rejection, alerting providers.

A group of researchers led by Mayo Clinic also created an AI algorithm, which used voice biomarkers to identify heart conditions. Previous research has identified a relationship between voice biomarkers and heart activity. The AI system used an app to study a patient’s voice sample and conduct an analysis based on various factors.