I am a pediatric cardiologist and have cared for children with heart disease for the past three decades. In addition, I have an educational background in business and finance as well as healthcare administration and global health – I gained a Masters Degree in Public Health from UCLA and taught Global Health there after I completed the program.
“The present framework is designed so that AI systems, that have been shown to be safe, effective, and where potential bias has been mitigated, and developed under an ethical framework, can be priced and reimbursed at a sustainable level, with multiple “guardrails” overseen by all stakeholders that enforce ethical principles overseen by regulators, providers, and patient organizations. The resulting reimbursement allows for sustainable, predictable financial incentives for AI creators, and continued research.”
Dr. Michael Abramoff, IDX-DR inventor
Artificial intelligence-enabled devices have undergone a Cambrian exponential increase in number and scope in the past few years. Since late 2020, the US Centers for Medicare and Medicaid Services (CMS) announced coverage provision for AI-specific CPT code and also the creation of the first New Technology Add-On Payment (NTAP) for an AI device. Following this historical development to pay for AI-enabled services, payers are starting to reimburse AI services for use of image-based AI tools.
The paper lists AI devices that are currently reimbursed by US Medicare in Table 1 (best known are the viz.ai and its Viz LVO and Viz SDH tools and the IDX-DR AI tool). The authors present five possible strategies for payment of artificial intelligence in medicine:
- Forgoing separate reimbursement of AI devices
- Incentivizing outcomes (instead of volume)
- Utilizing advance market commitments for new AI solutions
- Time-limited reimbursements for new AI applications
- Rewarding interoperability and bias mitigation.
One additional possible reimbursement strategy could be percentage of cost savings as well as a dividend for quality improvement as a combined payment since AI can potentially reduce time to diagnosis and therapy.
Of course any combination of the aforementioned strategies can be a “meta” strategy. The authors also suggest strategic design of AI and its payment to improve patient outcomes while maximizing cost-effectiveness and equity. This last statement is perhaps unfair for AI given that the entire health system is full of inefficiencies.
In other words, AI alone should not be at the forefront to yield this dividend without accompanying administrative measures. There is a possibility that CMS followed by payers will finally achieve a value-based care arrangement within a decade. Perhaps the future financial model should be based on AI as a service for entire sectors of the health system with patient outcome as the key driver rather than single-use reimbursements for every instance.
AI, therefore, can be at the collision point with a paradigm shift in healthcare: value-based care instead of fee for service. AI embedded into all sectors of healthcare can potentially be the strategy to finally achieve the Quintuple Aim, and the challenge is to devise payment schemes to allow healthcare to get to this healthcare nirvana.
Read the full paper here: https://www.nature.com/articles/s41746-022-00609-6