Startup Diagnoss has developed an artificial intelligence-based coding assistant to help automate the painstaking process of medical coding and billing.
The Diagnoss AI medical coding engine acts as a “sidebar” to electronic health records (EHRs) and uses machine learning to improve a clinician’s accuracy. The tool provides real-time feedback to medical practices during the administrative process and helps to reduce coding errors on claims.
Abboud Chaballout, founder and CEO of Berkeley, California-based Diagnoss, compares the AI tool to an assistant whispering in a doctor’s ear.
The AI tool works similarly to the Grammarly AI grammar-checking tool. It uses natural language processing to evaluate doctors’ notes while they are being typed or when they are uploaded to an EHR. It then suggests to the user the correct codes to use, according to the company.
Grammarly “reads what you're writing and gives you actionable information around grammar,” Chaballout said. “We're doing the same thing with codes.”
Founded in 2018, Diagnoss is backed by early-stage venture capital fund The House Fund. The startup joins a growing list of companies leveraging AI to help with tedious, time-consuming administrative tasks like coding. EHR vendor Cerner has developed AI tools to help with clinical documentation. Startup Olive uses AI to help healthcare workers carry out critical activities such as billing and patient verification. San Francisco-based Fathom also is developing deep learning tools to automate the painstaking medical coding process while increasing accuracy.
Diagnoss is working with mobile EHR vendor DrChrono so that doctors that are using that EHR platform can leverage the AI engine to improve the accuracy of their medical codes. In a study of more than 30,000 de-identified EHR charts, Diagnoss found that machine-coded charts were at least 50% more accurate than those that humans coded.
The medical coding engine incorporates an E/M Level Meter, which provides real-time feedback about whether the evaluation and management code is relevant to a patient evaluation. Based on evaluating a doctor’s documentation and charting in real time, the AI tool gives feedback on whether the medical practice can bill the codes. Machine learning allows the tool to learn and adapt so it can improve code accuracy, the company said.
It also makes predictions on ICD-10 diagnosis codes and CPT procedure codes.
Easing the strain of coding
Coding is a chore that healthcare providers, front desk workers or revenue cycle specialists such as billers or coders usually carry out. The task can bring financial stress for medical practices when coding errors occur.
Chaballout explained that physicians are often not as adept with coding as the trained medical coders. As medical coders try to decipher what the physician documented during the patient visit, they have a limited understanding of the documentation. An AI medical coding engine can help providers improve their documentation and codes from the start.
“We want to give providers an idea of what codes are being generated by their documentation and to give them an opportunity to fix their notes if they are not getting something that they would have anticipated being applied to that patient encounter,” Chaballout said.
Sometimes a coder will down-code what the doctor had selected because the coder lacked the documentation to understand which code is correct, he added.
RELATED: On heels of $100M funding round, Olive expands healthcare AI capabilities
Coding during the COVID-19 pandemic
During the COVID-19 pandemic, stress and strain can lead to medical coding errors. A Luma Health survey revealed that one-third of medical providers are already operating below 60% capacity, and costly medical coding errors leading to revenue loss could put practices out of business.
“Anytime you have stress and you have strain, you're going to have mistakes,” Chaballout said. “The mistakes come from flying through the process without giving it the time it needs to do it correctly.”
He noted the direct relationship between proper coding and payment.
“If you don't get paid, there's definitely a domino effect in terms of the viability of the medical establishment,” Chaballout said.
Getting doctors to choose better codes will streamline the process of getting claims approved and reduce costs.
“If we can help providers select better codes from the onset of the coding process, it will reduce a lot of costs downstream in terms of costs due to denials, and the appeals and the collections that have resulted from a denial,” Chaballout said. “It will reduce costs in terms of the amount of time that it's going to take coders to actually clean up what the doctors have done.”