More than 40% of submitted claims aren’t paid electronically or automatically upon their first submission. The AI applications in this area include more accurate coding from processing provider documentation and identifying the relevant billable medical codes. There are now algorithms which can tap into historical data and use it to predict whether a claim will be approved or denied. The medical insurance industry is increasingly recognizing the potential of AI for improving billing and insurance tasks. There are three main ways that AI is being used to revolutionize healthcare billing and insurance: NLP for automated, computer-assisted provider documentation (CAPD) and computer-assisted coding (CAC), AI for the prioritization of clinical documentation improvement (CDI) and electronic prior authorization (ePA), and AI-powered RPA (robotic process automation) to help with claims management.
Generative AI is showing great promise in this area and the large language models (LLMs) at the basis of generative AI will undoubtedly become the basis of AIs solutions that will anticipate the words and thoughts of clinicians so that they can further speed access to information and reduce administrative tasks. Today’s AI-powered RPA startups are on a mission to automate the repetitive, manual tasks like claims submission and denial which occur during the end-to-end revenue cycle. In medical coding, players such as Google-backed Nym use NLP to automate the labor-intensive process of translating EHR notes into billable code. Infinitus Systems is developing a “voice RPA” chatbot that asks questions about payment authorization and other procedures and records answers in the relevant fields.
LLMs from Google (Bard) and OpenAI (ChatGPT) have already made impressive progress when it comes to interpreting medical documentation. Healthcare-based NLP tools have also been released on open-source licenses by the National Library of Medicine (part of the National Institutes of Health), Amazon Comprehend Medical and Google via its Healthcare Natural Language API. Healthcare AI tools for administrative automation will mostly depend on EHR data. There are tools like NYM which tap into EHR notes to translate health services into billable codes, while RPA platforms are using AI to extract data from EHRs to populate insurance claims forms. This is also known as “charge capture”, the process by which doctors translate patient visits and diagnoses into medical codes that can be billed to an insurer. Historically, this has been done by hiring people who read the doctor’s notes and update the coding for that visit.
Companies such as Apixio and 3M have developed AI-powered risk adjustment and hierarchical condition category (HCC) auditors to help practices to optimize their coding for quality payment programs. We already know that AI does a better job of spotting patterns than human beings. Traditionally, rules-based engines needed to be constantly updated to reflect changes and to stay accurate over time, but AIs can automatically become better at what they do, especially when it comes to pattern recognition.
One of the companies working with CAPD technology is HITEKS, which provides embedded EHR workflows within Epic to provide timely advice on documentation to support ICD-10 claims coding. Their CEO, Dr. Petratos, is a medical informatics-trained physician who talked to me about some of the issues that medical centers are facing and how their NLP technology is solving some of these issues. According to Dr. Petratos, the sophisticated hospital systems in the U.S. are now seeking to notify their providers of clinical documentation improvement (CDI) opportunities in near real-time before they sign their notes, and within the workflows of the EHR. Legacy processes, which include a nurse, physician or coder manually reviewing each chart after provider notes are written (reactive CDI) and sending a note to the provider’s inbox, are inefficient.
“Reactive” CDI takes three times as long and can leave an organization vulnerable to denials due to changed documentation flags by insurance. CAPD software which includes advances in accuracy and timeliness can reduce what’s commonly known as “query fatigue”, when providers see too many alerts in the EHR and start to tune out. HITEKS developed an approach where they combined certain rules, which are known to be clinically relevant and are required for compliance reasons, with machine learning to suppress inappropriate alerts. “Proactive” CDI is the result of these AI advancements and has been measured to improve the bottom line with a 3% increase in revenues from inpatient accounts, 17% reduction in provider administrative time, and 25% reduction in CDI resources to oversee the CAPD processes.
