Health insurers are increasingly employing artificial intelligence (AI) to further automate labor-intensive tasks and increase alignment with patients and providers. This includes identifying fraud, waste and abuse, streamlining prior authorization and adjudication of claims, improving lines of communication with patient and provider, and designing value-based care models including tools for prevention. Payor investment into AI is a consideration not only to those with a direct stake in payors and the vendors which serve them, but to those invested in provider organizations which payor impacts may touch. Target analysis in this environment requires not only an understanding of payor applications, market and both federal and state regulatory and legislative trends, but a clear grasp on billing and coding. Herein we address key areas of AI innovation by payors and the potential impacts on providers.
To receive a copy of this report, please contact us.
Payor Investment Into AI – A Source Of Savings And System Alignment
Commercial health insurance spending is a highly competitive trillion dollar market where new entrants and traditional players alike continue to look for an advantage. AI offers the ability to not only automate labor-intensive tasks, but increase alignment with patients and providers. This has been estimated to translate into a potential savings of 5-10% of US healthcare spending. It’s adoption, as Centene CEO Sarah London noted earlier this year, could play out similarly to innovations in the financial services in the late 1990’s and early 2000’s, where customers gained greater agency over their decisions.
Given the potential savings, payor investment into AI is a consideration not only to those with a direct stake in healthcare organizations and the vendors which serve them, but to those invested in provider organizations which payor impacts may touch. Herein we describe key areas of payor AI innovation and the potential impacts on providers.
Healthcare fraud costs $30B annually in the US, placing a burden not only on payors, but the patient in the form of increased premiums. Consequently, effective fraud, waste and abuse detection is an ongoing focus of payors.
AI is increasingly being used to scan large quantities of paperwork generated by claims processing. By some estimates, payors classify 7 out of 10 claims as unusual – thus requiring reinvestigation. Natural language processing (NLP) reduces the time required to identify suspicious claims. For example, BCBS of Massachusetts is using this form of AI to detect fraudulent claims. The company has developed an algorithm that looks through its claims data and flags any suspicious activity before the claim is paid.
The ramification to provider organizations is increased scrutiny into their coding and billing processes. Consequently, investment into these organizations require increased diligence into billing and coding practices, including independent audits to gauge revenue risk.
AI-driven chatbots have expanded from online commerce to healthcare, automating the majority of customer service requests that involve simple queries that previously would merit a scripted response. By some estimates, this translates into a 40% reduction in customer service costs. Beyond the scripted, chatbots are increasingly being taught to extract individual data with the plan to better respond to queries. For example, Elevance Health offers artificial intelligence-powered digital concierge care to members in an effort to provide personalized care for chronic conditions such as diabetes, Crohn’s disease and long COVID-19.
Chatbots are also being used for handling insurance claims, such as verifying claim progress. With the rise of mobile apps, plans can more effectively interact with patients to better automate claims processing.
The impact on a provider organization may include decreased days receivables outstanding, improving the organization’s cash flow. Indirectly, more rapid claims processes speed patient challenges in reimbursement which may otherwise consume provider staff time and in some cases slow patient willingness for elective provider services.
AI can analyze patient data and insurance policies to determine the likelihood of obtaining prior authorization for procedures or treatments. This can accelerate the authorization process and minimize delays in patient care. For example, Blue Shield of California employs Google Cloud technologies and associated machine learning and AI technologies to streamline its prior authorization process and claims processing. The platform converts unstructured data into structured data and can help payers meet current and proposed CMS rules around interoperability and prior authorization. Health Care Services Corporation expanded use of AI technology to speed up prior authorization. It has piloted a program in behavioral health and specialty pharmacy
Provider organizations will need to demonstrate greater alignment with payors as these prior authorization systems become less burdensome to implement by the latter party. In parallel, it will become increasingly important to conduct thorough benchmarking of a provider organization’s denial rates versus peer and industry averages in an analysis of a target.
AI has enormous potential to accelerate value-based care models just as it already plays a role in risk adjustment, quality improvement and member management programs.
AI can be put to work on the vast data challenges of abstracting, processing, and coding for risk adjustment and quality improvement programs. AI technology will play an ever growing role in mining, processing, and analyzing massive data sets; reliably applying rules and logic to digest the data quickly and accurately.
A current challenge remains data and process fragmentation. Increased ability of AI to extract unstructured data will bring siloed data together in a longitudinal record, pairing structured and unstructured data from health information exchanges, all hospital and specialist electronic health records, medications, and labs. The benefit both to physicians and payors designing value-based models is reduced burden and bias in shifting through large amounts of data to find the information they need or perhaps didn’t even know they needed.
Additionally, predictive AI models utilize past data and analytics to forecast future outcomes. In value-based care, predictive AI can be used to improve clinical outcomes while creating opportunities for proactive, lower cost care at a point of lower patient acuity. For example, prescriber bias can be reduced or discouraged by presenting the most cost-effective therapy or treatment at the right time for the right patient based on their diagnostic and health profile.
Despite technological enablement, value-based models will still need to be underpinned by economic analysis. This includes awareness of disease states, treatment and reimbursement trends, whether federal, state or commercial.
Spotlight On Prevention
Health plans, most visibly Medicare Advantage plans, have expanded their benefit to improve the health of their members. This include such features as coverage for dental and hearing, as well as allocations for health-related services such as gym memberships. The value of these services will increasingly be analyzed by AI to identify trends and iterate designs which optimize patient utilization and captured benefit.
More recently, life insurance companies have begun pioneering benefit designs connected to behavior. In 2018, John Hancock introduced a new type of health insurance: an interactive policy that collects health data via smartphones and wearables. Smart devices expanding beyond phones and watches to homes, cars and appliances collect a tremendous amount of data about the customer’s health choices. AI is increasingly essential in the aggregation, processing and analysis of this data into impacts on the person’s premium. The larger impact will be that health insurance companies will be able to more efficiently conduct an effective and fair underwriting process.
Evaluation of new service models and cost-containment structures will require cognizant awareness of the impact AI will have on their acceptance and integration by payor organizations. Most notably, what they have already built internally and where they look to expand. It will also require a strong understanding of the competitive landscape of solutions and behavior of eligible lives in relation to the overarching provider landscape.
State Regulatory Considerations – An Evolving Landscape
Twelve states have enacted legislation related to AI, 13 states have proposed legislation, and 1 (New York) has both enacted and additional currently proposed legislation. Notably, the states have been active in the regulation of AI, similar to their role in the early days of the internet. However, AI may be perceived as novel in many states. Risk analysis of a potential strategy necessitates communication with regulatory and legislative stakeholders and their proxies within select states to understand the climate, stability and direction of legislation and regulation.
Federal Regulatory Considerations – An Active Space
In the five months between May 1 and October 10, 2023, the House and Senate put forward 39 proposed bills related to artificial intelligence. Based on the Federal Register as of October 30, 2023, not a single proposed bill on the subject had passed, with none progressing beyond introduction.
On a broader level, continued federal regulatory and legislative analysis is crucial in this rapidly evolving space. In June of 2023, Senate Majority Leader Chuck Schumer (D-NY) announced his SAFE Innovation Framework, policy objectives for what he calls an “all-hands-on-deck effort” to contend with AI. Asserting that the “traditional approach of committee hearings” is not sufficient to meet the moment, Sen. Schumer promised a series of AI “Insight Forums” on topics to include such topics as privacy and liability. More recently, in late October of 2023, President Biden issued an executive order in the AI space which directs new standards for AI safety and security as well as privacy, among other topics.
Federal regulatory and legislative analysis as it pertains to a target may include not only direct impacts of AI, but consideration of trends shaping the coverage and reimbursement of services, drugs, devices, and/or diagnostics.
Target analysis in this environment requires an understanding of payor market dynamics, as well as federal and state regulatory and legislative policies guiding use of AI and digital enablement. It also requires market access knowledge of patient-provider dynamics and strategic analysis on the AI landscape in healthcare applications. Finally, a hands-on knowledge of billing and coding is crucial.
Marwood brings extensive experience in each of these areas in an integrated fashion that provides a rounded view on a target or portfolio company.
 Sahni NR, Stein G, Zemmel R, Cutler D. (2023). The Potential Impact Of Artificial Intelligence On Healthcare Spending. Publications Of The NBER Economics of Artificial Intelligence Conference. https://www.nber.org/system/files/chapters/c14760/c14760.pdf accessed November 2023
 Chakrabarti S and Ghosh A. (2019). Chatbots: Riding The Next-Gen Technology Wave To Operational Success. Deloitte.
ABOUT THE AUTHOR
Mark Slomiany PhD MBA MPA is a Director of Advisory at The Marwood Group and a former faculty member of the Department of Cardiothoracic Surgery at New York University Langone Health, as well as a former research associate at the Mossavar-Rahmani Center for Business and Government at the Harvard Kennedy School of Government.