By STEVEN ZECOLA
On December 19, the Department of Health and Human Services (“HHS”) issued a Request for Information seeking to leverage artificial intelligence (“AI”) to deflate healthcare costs and make America healthy again.
As described in this document, AI can be used in many dimensions to help reduce healthcare costs and improve care. However, to make significant progress with AI, HHS will need to completely revamp the regulatory approach to drug discovery and development.
Dimension #1. Incorporating AI into drug discovery
The greatest benefit of AI to the performance of the healthcare industry can be achieved through drug discovery. Taking into account the costs of failures, the average FDA approval of a drug costs society nearly $3 billion and takes decades to reach the market from its inception in the laboratory.
In contrast, AI identifies potential treatments much faster than traditional methods by processing large amounts of biological data, uncovering hidden causal relationships, and generating new actionable insights.
AI is particularly promising for complex, multifactorial conditions (such as neurodegenerative diseases, autism spectrum disorders, and multiple chronic diseases) where conventional reductionist approaches have failed.
In the short term, HHS should direct its grants toward AI-powered basic research, with a focus on hard-to-solve diseases. At the same time, the FDA should implement a new approval system for AI-initiated programs to enable innovative treatments on a compressed timeline.
Dimension #2. Incorporating AI into the drug development process
Simply relying on AI for drug discovery, while subjecting its advances to the current approval process, would undermine the use of the technology.
Rather, improvements can already be made from AI in meeting exhaustive regulatory documentation requirements, which today add up to 30% of the cost of compliance.
In the short term, AI can improve drug development by:
- Automation and validation of regulatory documentation.
- Improved trial design and participant stratification
- Near real-time safety and effectiveness monitoring
- Reduce administrative and compliance costs
For example, in the United Kingdom, the Medicines and Healthcare products Regulatory Agency reported that clinical trial approval times were twice as fast with AI and associated reforms.
To achieve much greater long-term gains, HHS should bundle all clinical work using AI into one long trial rather than discrete Phase I, II, and III trials, since AI can be used to continually update and validate documentation. This change would not require a legal change or agency rulemaking because clinical trial design is not codified in FDA rules.
As participants are added to a trial, safety results can be examined and reported in real time. Once the trial exceeds a certain number, such as 1,000 participants, with proven efficacy and meets specified safety protocols, it will be approved for implementation. The government’s role in such an approach would be that of an auditor to validate the trial result. This role would include experimental validation, mechanistic understanding, and ethical oversight.
With these changes, FDA staff would shift from episodic gatekeepers to ongoing auditors, requiring a fundamental change in organizational culture. While safety concerns would still be important, responsibility would be shared more equitably between applicants and trial participants. Furthermore, the prolonged suffering of existing patients would be taken into account in the public welfare analysis when reviewing the preliminary safety results.
Dimension #3. Improve data collection to power AI
Complete and accurate data is essential to AI success. However, this is another area where the healthcare industry has failed.
The industry has evolved and each provider, or family of providers, encourages their patients to register for a customer portal. Providers usually treat the information on these portals as their own for research purposes. However, the providers do not own the data. Each patient is the owner of their data.
To expand the scope and applicability of healthcare data, HHS should establish national standards for patient-facing data collection that:
- Use interoperable formats
- Capture both diagnostic results and relevant explanatory variables.
- Preserve patient ownership and informed consent
- Enable longitudinal tracking while protecting privacy and security
Once this format is established, HHS should set a goal of enrolling 100,000 participants within two years.
ddimension #4. Using AI to establish standards of care and maximum prices
There are no national standards of care for diseases or other health problems in the United States. Patients often do not understand the nature of their condition, the options for treating it, or the costs of various options for remedying it.
In parallel, HHS might fund basic research aimed at a particular ailment, the FDA might (or might not) approve it, Medicare might (or might not) cover it, and some insurance companies might cover the treatment and others might not.
Furthermore, the costs of various treatments can vary greatly from one center to another, without the patient knowing it.
In addition to this market dysfunction, healthcare professionals have the desire (and financial incentive) to provide the best (and probably the most expensive) service possible to their patients.
In short, there is a market failure, mainly related to the lack of actionable information.
In the short term, AI can help address these failures by aggregating and analyzing how care is delivered across the country and identifying patterns associated with better outcomes and lower costs. These insights could be used to inform evidence-based minimum standards of care and improve transparency around pricing and performance.
In the longer term, the results of these systems could be used to establish a minimum level of care for all (or most) ailments. These standards would be compulsorily covered by insurance. At the same time, the results of these standards of care could be complemented by regional maximum prices for the various practices based on a comprehensive analysis of the industry.
As experience is gained with these AI reporting systems, a future version could be programmed to automatically calculate prescribed minimum standards of care and maximum prices to mimic the workings of supply and demand curves. An algorithm could be constructed using as an equilibrium a specific level of subsidy provided by the federal government. As the federal subsidy exceeds certain preset limits, AI would be used to address the imbalance by providing policymakers with several options that would reduce the maximum price for certain conditions and/or reduce the minimum level of care.
In scenarios in which the stipulated federal subsidy was exceeded, some classes of patients would be denied payment for the best available treatment (unless they had supplemental insurance) and/or some health care providers would suffer a decrease in their profits.
Such an approach would require congressional approval, but those trade-offs are happening now, without informed options. In this dimension, AI could be used to address the industry’s huge information failure and address increasing subsidies.
Dimension #5. Incorporating AI into HHS internal processes
AI can also improve the efficiency and effectiveness of HHS’s internal operations. While the potential percentage gains would be smaller than those on the discovery and development dimensions, even modest improvements can generate significant savings given the scale of federal health care spending.
Conclusion
AI offers the opportunity to achieve significant improvements in healthcare outcomes and efficiency, but only if it is integrated into a regulatory and governance framework designed for its capabilities. Introducing AI into existing structures will mitigate its impact and increase implementation risk.
Each dimension described above requires a dedicated and independent multidisciplinary team reporting to the Office of the Under Secretary. Once the strategic direction for each dimension is established, these teams should be tasked with:
- Develop detailed implementation plans, including budget requirements.
- Identify any legal or regulatory barriers
- Establish timelines, milestones and evaluation criteria.
- Address ethical and equity considerations
Drug discovery and development represent the most impactful dimensions for AI implementation. HHS should draw on outside expertise to craft the details of an appropriate regulatory framework for these dimensions.
Detailed plans to implement AI must be approved and finalized by the end of 2026. As outlined in this document, HHS must take a proactive, forward-thinking role in leveraging AI to limit healthcare costs and improve care.
Steve Zecola sold his web application and hosting business when he was diagnosed with Parkinson’s disease twenty-three years ago. Since then, he has run a consulting practice, taught at graduate business schools, and practiced extensively.
