Abstract
Value-Based Care (VBC) continues to be a strategic priority for the Centers for Medicare and Medicaid Services (CMS). American healthcare market is on a tight schedule to achieve a blanket VBC model adoption across providers and payers. These models strive to create real value through reducing inefficiencies, enhancing human productivity, and improving patient outcomes. However, the data generated by health plans and providers remain largely unstructured and challenging to analyze. Clinical large language models (LLMs) and neural networks can refine, enrich, and process this data to provide actionable, real-time insights. Generative AI surpasses traditional rule-based models in optimizing operations and predicting risks. Although CMS has not officially endorsed clinical LLMs, its policy design encourages providers and payers to adopt generative AI solutions to maximize value and revenue potential. The timing is ideal for implementing automatic technology, as VBC becomes mainstream, cloud infrastructure is robust, and the business case for Artificial Intelligence shows clear returns on investment.
VBC’s Drive to Accelerate
The Center for Medicare and Medicaid Innovation aims to have 100% of Medicare beneficiaries in accountable-care relationships by 2030, making the VBC payments and services market a $1 trillion industry. Over the next five years, VBC patient volumes and capital investments are expected to double. CMS incentivizes health plans to adopt VBC principles through programs like HEDIS Measures, linking revenues to the effectiveness of care provided to members. The goal is to maintain population health by rewarding effective care.
Currently, VBC struggles to reach its full potential. The increasing volumes and complexity in care management necessitate continuous measurement and analysis, which only a few large national players can afford. To democratize VBC access and unlock its true potential, payers and providers need tools that can structure data, enrich inputs, and auto-generate insights with minimal human effort. Generative AI offers these capabilities to healthcare providers, payers, and biotech companies.
Health Data: The Crude Oil of Healthcare
In the U.S., clinical data and medical evidence are reported across more than 100 document types by over 10,000 entities, using diverse formats and transmission methods. Each state has its own data generation, sharing, and usage nuances, resulting in over a million possible variations for 300 million medical records. Hard-coding a solution for such a vast and diverse population is impractical. While EHRs act as partial data refineries, generative AI serves as a super-efficient machine for refining this “crude oil.” FHIR and interoperability frameworks function as the supply chain network, transporting refined data to users. Generative AI enhances this process by presenting data in user-friendly formats, such as visualizations, chatbots, FAQs, and customized dashboards.
Delivering a care model that adds actual value versus merely billing for services is challenging. Even more challenging is quantifying this value and proving its impact over time. CMS aims to reward or penalize every action taken or not taken by providers and payers, creating a connected consciousness for the healthcare system. Clinical LLMs and neural networks have the potential to be the layer that measures, enriches, analyzes, and generates actionable insights from healthcare activity.
Case Study 1: Generative AI Enhances Utilization and Reduces Care Costs Effective resource utilization includes reducing 30-day readmissions, improving medication adherence, and preventing unnecessary tests. Current predictive tools cluster high-risk patients, but significant slack remains. Continuous data input measurement from medical charts, IoT devices, and non-clinical sources (e.g., web browsing history, GPS data) is needed to assess readmission risks. Clinical LLMs, combined with non-clinical neural networks, can automate data evaluation and risk flagging, quantifying risks and suggesting mitigation approaches. A simple LLM-powered chatbot can reduce hospital revisits and improve medication adherence. Traditional models have reduced readmissions and noncompliance from 20% to 12-15%, while generative AI can potentially reduce it further to 5%, enhancing patient experience and reallocating human resources to high-impact activities. |
Case Study 2: Improving HEDIS Scores for Better Outcomes and Profitability Ninety percent of U.S. health plans use HEDIS Measures to track care and service performance. HEDIS scores influence CMS incentive payments. Extracting data from medical charts, identifying care gaps, and calculating key performance indicators is labor-intensive and expensive. For instance, a health plan with 10 million members might hire 200 temporary employees for three months to calculate its annual HEDIS score. Clinical LLMs can automate 80% of these tasks, with Artificial Intelligence further automating data visualization. Live dashboards provide actionable insights and answer queries from staff, members, clients, and auditors. This enables continuous HEDIS score measurement, improving payer scores and patient outcomes. |
Ninety percent of U.S. health plans use HEDIS Measures to track care and service performance. HEDIS scores influence CMS incentive payments. Extracting data from medical charts, identifying care gaps, and calculating key performance indicators is labor-intensive and expensive. For instance, a health plan with 10 million members might hire 200 temporary employees for three months to calculate its annual HEDIS score. Clinical LLMs can automate 80% of these tasks, with Artificial Intelligence further automating data visualization. Live dashboards provide actionable insights and answer queries from staff, members, clients, and auditors. This enables continuous HEDIS score measurement, improving payer scores and patient outcomes.
The Imperative of Timely Adoption
Generative AI has matured significantly. The post-pandemic economic shift has accelerated generative research and applications. With resources freed from the crypto boom, attention and budgets are now focused on implementing LLMs to enhance human productivity. This strategy is now a top priority for healthcare CIOs and CEOs. Providers and payers using LLM-based solutions are outperforming competitors in delivering value and realizing cost savings. Large healthcare players are consolidating their market positions through acquisitions, while smaller entities focus on scalable tech enablement. The pace of innovation necessitates swift action. VBC models must embrace generative AI to avoid missing out on its benefits. Generative AI is not the end goal but a means to achieve better value in healthcare. Clinical LLMs generate this value from data, interoperability, and intelligent insights, making “value” the new currency in healthcare.
Authors:
Dr. Rahul Garg
Rahul Garg, MD, MBA: With over 14 years in the health tech domain, Rahul offers extensive expertise vital to the healthcare venture capital industry. An executive and physician with dual MBAs from IIM Ahmedabad and Queen’s University, he excels in bridging clinical insight with business acumen. His leadership spans strategic consultancy to driving initiatives for health groups and startups, specializing in operational efficiency, digital health innovation, and portfolio strategy for insurers. Rahul is adept at guiding VC and private equity investments in health, fostering technology’s role in traditional healthcare to enhance access and outcomes. Committed to innovation and equity, Rahul frequently mentors emerging startups aiming to improve clinical outcomes, reduce the cost of care, and make care more accessible.
Ganesh Padmanabhan
Ganesh Padmanabhan: Ganesh is the CEO & founder of Autonomize AI, a technology company augmenting knowledge workflows with Trusted AI Copilots for healthcare and lifesciences. He is also the Founder & Chief Creator of Stories in AI, an edu-tainment venture. He previously co-founded and scaled Molecula Corp, a data management company, and led growth at CognitiveScale, Inc, an Enterprise AI company. Prior to that he spent a 15 year career spanning Dell Technologies and Intel Corp. Ganesh holds a bachelor’s degree in Mechanical Engineering from University of Calicut, India, and an MBA from McCombs School at the University of Texas Austin.