Impacting patient treatment by reducing coding errors
Medical coding is currently a manualintensive effort involving text analysis, inference and mapping to the correct codes from the tens of thousands of codes in the standard coding systems. The accuracy of the manual method ranges from 50-98%, directly impacting patient treatment outcomes, insurance reimbursement, and healthcare facility compliance.
Narayana Health explored AI, powered by Azure OpenAI to automate this task. What were the results? Find out.
Automating coding of medical records
To manage insurance claims, all the procedures and services availed by a patient has to be translated into universal medical codes. This medical coding process involves mapping every patient-provider interaction captured in the electronic medical record (EMR), specific data points and unstructured clinical notes. It requires months of training and yet creates significant backlogs, in addition to being fraught with potential for gaps and errors.
To solve this challenge, Medha Analytics, the AI division of Narayana Health, developed a bot powered by the Azure OpenAI GPT-4, that automates the extraction and coding of medical records, ensuring accuracy, consistency and efficiency.
The bot reduced coding errors by 40% compared to the manual coding in the POC dataset in addition to eliminating the time taken to manually code by providing instant results.
Dr. Emmanuel Rupert
Managing Director and Group CEO
Narayana Health
AI-driven clinical coding will play a pivotal role in reducing global healthcare costs and enhancing patient safety, as it is set to exponentially elevate the efficiency and accuracy of the $20+ billion medical coding industry.
40%
reduction in coding errors
Narayana Health operates a network of hospitals across the country.
More details at https://www.narayanahealth.org