Smart machines have quickly become part of everyday commerce — from iPhones to TikTok — with an increasing number of Pittsburgh-area businesses finding ways to leverage machine learning and artificial intelligence for profit.
Locally, PNC Treasury Management, a branch of banking giant PNC Financial Services Group Inc., is tapping AI to help health care providers limit the number of claims that are rejected for payment by insurers and other third-party payers. Claims are rejected for a number of reasons, including completeness, insurance coverage issues and required prior authorization for procedures, which ranks among the top reasons.
What’s more, the number of claims turned back to medical providers rose 20% in the past five years, according to Tampa, Fla.-based consultant Harmony Healthcare. Costs can exceed $100 for fixing and re-billing an insurer or other payer, so only 35% of rejected claims are submitted again.
A survey by the American Hospital Association found that 13% of claims made by Pennsylvania health care providers in 2019-2020 were rejected, which can mean delayed or lost revenue at a difficult time for doctors and other medical providers.
Results so far for the PNC Claim Predictor have been encouraging: Early pilot use of the tool found that it was 80% accurate in identifying claims that were likely to be rejected, said Doug McKinley, senior vice president and head of Treasury Management Innovation. Flagging problem claims allows providers to fix mistakes before the bills are sent off for payment, avoiding the risk of rejection.
“Claim denials are a big problem in the health care business,” he said. “We like to collaborate with clients along the way. We involve them in the process throughout.”
Hospital claims for payments may not be the first thing that springs to mind when you think of an outfit best known for retail banking and home mortgages.
But PNC has been serving the health care industry for more than 25 years with more than 500 employees and more than 51,000 health care clients in the U.S., said Brian Kelly, executive vice president and head of PNC Healthcare.
Here’s how the PNC Claim Predictor works.
The software is loaded with past claims, both paid and rejected — which prompts the machine learning part of the program — and then AI is used to predict which new claims are likely to be rejected based on past experience. Because payment criteria can vary among insurers, the PNC program is tailored to metrics used by each payer.
When an insurer tweaks its criteria for paying a doctor or hospital for care, the program “learns” the changes through constant updates. What’s more, the PNC Claims Predictor can be integrated into electronic medical records, including Epic Systems, a popular way of tracking patient care.
Sometimes, all it takes is an error in a single procedure code for the whole claim to be rejected, Mr. Kelly said.
PNC also uses machine learning and AI within the back offices of Treasury Management to improve efficiency, Mr. McKinley said.
PNC Claim Predictor grew out of conversations with people directly involved in health care. PNC Treasury Management uses an advisory committee made up of doctors, insurance administrators, health system executives and others to find solutions to the pain points in the delivery of health care.
Claim rejection was a problem identified by the committee, Mr. Kelly said.
“All of this is designed to help organizations have a better experience with their patients,” he said.
Kris B. Mamula: kmamula@post-gazette.com




Comments
Welcome to Post-Gazette subscriber-only commenting community. Please review our community guidelines.