Kansas University Healthcare System Tests Neural Network to Create Medical Notes Automatically
The Kansas University Healthcare System is testing a new neural network that automatically creates notes based on the conversations that doctors have with patients. The technology, developed by startup ABRIDGE, is aimed at reducing the workload for doctors, improving patient care, and reducing the risk of doctor burnout. Shivdev RAO, cardiologist and CEO of ABRIDGE, notes that doctors often spend hours writing notes outside of working hours and that frequent note-taking leads to burnout and end of career for some.
ABRIDGE’s software automatically generates reports of medical conversations through the use of artificial intelligence algorithms and natural language processing. During initial testing, the software recorded things such as patient symptoms, prescribed drugs, and the actions that the patient must perform before the next visit. The program listens to keywords and classifies the most important information. This makes it easier for doctors to review the patient notes outside of the hospital.
The program classifies the words spoken by the doctor and patient into different categories, such as previous history, future plans, or actions that must be taken. Using a model based on a converter, ABRIDGE creates a document that combines the classified information into short sentences in different subsections.
Doctors have the ability to edit the notes created by ABRIDGE, and patients can access them through the application. RAO notes that ABRIDGE is an assistant, and the doctor must independently check and edit generated notes, if necessary.
ABRIDGE has been tested by over 2000 doctors and has assisted over 200,000 patients. The company plans to test it in hospitals with a small group of doctors.