Implementing a new VBC program in healthcare requires cross-functional support and overcoming numerous challenges. Simplification opportunities exist to address pain points for program administrators such as rigorous research, ROI assessment, and stakeholder engagement. Manual processes, including participant recruitment, financial modeling, program integrity management, and technical assistance can benefit from technology to streamline and automate tasks, allowing skilled resources to focus on higher-order activities.
Hurdles also exist for providers seeking to adopt VBC beyond a change in reimbursement cash flow. Successful adoption requires dedicated resources to educate and guide implementation of behavior change among staff and physicians. It also requires a considerable amount of data and analytic resources to make sense of that data. Some organizations, such as small, independent practices, lack the sophistication or capital to make it happen. Others that have the resources still find the investment hefty and the process burdensome. The American Society for Clinical Oncology (ASCO) reports that oncology practices struggle with government and private payer VBC models. Key challenges include understanding VBC terms, leveraging data for improvement, tracking care costs, data sharing and compliance, and lacking integrated technology. Practices see significant opportunities in interoperability, AI, and machine learning integration. Additionally, varying program requirements lead to provider fatigue and duplicate efforts.
Use of Generative and Agentic AI has the potential to address challenges within VBC and improve efficiency of operational processes for clinical and non-clinical healthcare professionals.
Model Design & Implementation Powered by AI
Processes to design a new model can take months or years, influenced by existing infrastructure and the model’s financial and quality mechanics. Today, program staff can use AI-powered natural language processing and machine learning to efficiently conduct environmental scans and literature reviews to identify best practices and lessons learned.
In addition, AI can enhance actuarial modeling by incorporating a wider range of data sources, including social determinants of health, to more accurately predict healthcare costs and utilization. This can support payers in designing and managing VBC programs.
Incorporating risk adjustment in a program’s financial methodology is essential for program administrators in ensuring targets are accurate and achievable. Getting the math right, however, can be incredibly burdensome and this is a cyclical task that requires ongoing updates. Use of Agents to group claims and stratify patients can accelerate these tasks.
For instance, commercially available Agents already exist to group claims into relevant categories, such as attributing them to providers participating in the alternative payment model and filtering out ineligible beneficiaries based on defined algorithms. Others exist that can segment patient populations based on risk profiles, disease states, and social determinants of health, allowing payers to tailor care management and reimbursement strategies. Similar approaches can help program staff deconflict overlap in claims and avert potential “double-dipping” among financial incentives.
AI-powered systems can automate financial reconciliation and report generation, allowing payers to efficiently manage the financial aspects of VBC programs. This automation reduces the time required for producing large-scale reports for large programs from weeks to just a few hours. Tools such as Databricks and Snowflake can support this by providing scalable data processing and storage capabilities to handle large volumes of provider data.
Agentic AI for Contracting and Provider Support
Agreements can be incredibly complex and the process of negotiating terms can take months or in worst case, years. Most sophisticated organizations use a rubric of terms to indicate what provisions are preferred, acceptable, discouraged, or unacceptable (PADU). Generative AI has the potential to analyze proposed terms against these rubrics and generate best-case scenarios and options for negotiators to reach win-win agreements with providers.
Most organizations underestimate the time and effort required for provider performance support. Complicating the matter is that there can be no “one size fits all” approach – providers vary widely in resourcing and level of sophistication with VBC. Integrating conversational Agents into provider dashboards can help meet providers where they are through tailored data-sharing and reporting and integrating interoperable data sources. Chatbots and virtual assistants powered by AI can provide 24/7 support to providers, answering questions about the APM, helping them access and interpret their performance data, and directing them to relevant learning resources.
Conclusion
In conclusion, the adoption of VBC in healthcare presents significant challenges, from the need for extensive data analytics and risk adjustment to the integration of advanced technologies like AI. These challenges are particularly pronounced for smaller practices, but even larger organizations find the process burdensome. AI offers substantial opportunities to streamline and automate complex tasks such as actuarial modeling, patient stratification, and financial reconciliation, significantly reducing the time and effort required. By leveraging AI and other advanced technologies, healthcare providers and payers can more effectively manage VBC models, ultimately leading to improved care quality and cost management.
About Leslie Vasquez
Leslie Vasquez is a value-based care and performance improvement evangelist with 20 years of experience in healthcare. She has designed, launched, and led private and Federal programs to improve access, efficiency, and quality in a variety of care settings. Ms. Vasquez has worked in the payer, provider, and health IT sectors across her career. She presently serves as Federal Value-Based Care Lead at Accenture, a global consulting firm.
