Insurance scams are everywhere. From business based scams finding patient information and misrepresenting it, to patient scams in trying to garner greater reimbursements than deserved based on the falsified information. And on either side of the equation, someone loses. The latest in artificial intelligence (AI) is trying to take care of this problem by evaluating patient charts in order to determine the exact nature of diagnosis and treatment.
Currently, insurance providers are forced to evaluate patient data manually, inputting specific codes into their systems in order to determine patient diseases. This process is painstaking and fraught with complexity, since human error can ultimately cost companies huge sums of money. The department of justice has already filed numerous lawsuits against insurance companies because they have failed to do the appropriate level of due diligence when it comes to patient medical records and coding.
Enter Apixio. The company has just released a new service called Code Compliance Auditor. The software uses AI in order to scan patient medical records and evaluate and compare the text that is mined with the appropriate medical insurance codes, effectively eliminating the need for human coders. The main target market is companies who are serving patients with Medicare Advantage plans – secondary coverage for those under Medicare. Patients with this coverage are offered large lump-sum payments from the US government. However, those payments are adjustable based on which disease the patients have, and how serious their conditions are. In order to apply and receive these payments in the proper way, companies need ways to code without substantial error.
Many companies have chosen to outsource this work to third party providers, but this solution doesn’t deal with the underlying human error. The newest software from Apixio will scan the text of patient medical records and assign codes based on the diagnoses it determines from the medical charts. Human employees can then sign off on the codes chosen, reducing the risk of long term loss.
One of the major hurdles Apixio has had to overcome is the quality of documents from medical charts. Much of their documentation comes as scans or faxes which can be difficult to read and analyze. Apixio has trained its models on those types of text in order to deal with errors within the documents. What’s more, the disorganized nature of medical chart files makes it problematic for machines to determine which files are critical. The machine models have learned to determine which documents are most important, and even which statements within them will provide information for diagnosis.
Apixio is confident that a team of just four employees can code as many as fifty thousand medical files in as little as a week – a major improvement over any legacy models, and a huge savings for insurance companies. AI continues to take the world by storm, making inroads into almost every industry sector where human controls and human error are at risk. This most recent development will certainly continue to advance the AI ecosystem.