The White House’s release of its Artificial Intelligence (AI) Action Plan last month mentions “healthcare” several times, one of the few industries specifically called out. There’s a reason: Healthcare and life sciences are about to face unprecedented AI-driven regulatory changes that will reshape everything from research and development to drug approval submissions.
The Food & Drug Association (FDA) already signaled a transition to AI-enabled infrastructure and support capabilities through its Elsa announcement earlier this year. With the addition of the AI Action Plan, it is evident that government agencies and regulatory bodies will continue to explore AI-enabled oversight. As healthcare and life sciences organizations prepare for continued adjustments to regulatory expectations, strategic planning must include internal standards in anticipation of future guidelines.
Crucial to planning for this strategy is avoiding “AI static friction,” where legacy methods of working resist changes that seem unproven or abrupt to an organization’s experience. Instead, embracing an openness to adaptation, which generates a form of “AI kinetic friction” that lowers the amount of effort and resource investment required to adopt new innovations. This openness to adaptation will be essential in aligning organizations with new regulations that replace traditional methods with updated protocol assessments, novel analyses and new systems that enable the sharing of data and results.
The Power Behind Enterprise EHR Software for Large Healthcare Systems
Enterprise EHR boosts scalability, interoperability, and governance for large healthcare systems.
Specifically, the latest AI Action Plan calls for a variety of advanced computational approaches that allows for large scale experimentation. It also proposes that future research and hypotheses consist of AI-informed experimental plans, AI-generated hypotheses and AI-assisted experiments. This marked momentum of new approach methodologies should alert organizations that evolving regulations will continue to enable AI innovation in the industry while also potentially setting a baseline expectation that future research should incorporate AI in some capacity.
Considering these proposals, healthcare and life sciences organizations should incorporate the following into their strategy to prepare for upcoming regulations.
1. Position AI as fundamental element of research and development enterprise: New technology should be vetted at the same level as medicinal chemistry or translational sciences because that data will eventually serve as the foundation for clinical decisions and new drug applications
2. Create accessible scientific and clinical data: Traditional systems have focused on micro-specialized and compartmentalized functions of clinical research. For new innovations to thrive, organizations must develop a new data access paradigm that moves beyond legacy risk and security postures.
3. Foster data partnerships with broad access, large scale and high recency: Traditional practices have guarded data as a highly controlled and protected asset. However, legacy sources and operating models usually work against AI-first strategies. When AI models are added to traditional capacities, the likelihood of bias, low reproducibility, and insufficient depth simply increase. Sharing data across multiple organizations and fostering collaboration will accelerate new areas of biomedical innovation while minimizing the risk of inaccurate results based on insufficient data.
Survey on the Healthcare Financial Landscape Offers a Roadmap for Stakeholders
Zelis highlights how employers, payers, and consumers are integrating digital tools to manage healthcare costs.
4. Incorporate AI plans into research design and documented outcomes: Publications and regulatory submissions will increasingly contain sections that include AI-augmented hypothesis, AI research plans and AI-model outcomes. The standard of clinical research and regulatory submissions will increasingly include AI-generated research. Organizations must ensure that these models complement traditional approaches while also being fully transparent when AI-enabled approaches have been used in submissions.
5. Anticipate what models know and will ask: The introduction of Elsa requires that organizations anticipate and prepare for AI-assessments of protocol, study design and endpoints. As opposed to previous regulatory processes, where committee members overlapped and new programs were compared to the previous few, AI-driven review now provides a much more extensive examination of research. With language learning models and generative AI, review cohorts will have access to a much broader view of a disease, its treatment objectives, safety requirements and patient care objectives. Plan all preparations for assessment with this broad view in mind.
6. Open access to highest-impact innovations: Through AI, new degrees of innovation are made possible, allowing for the examination of patterns and relationships in novel ways. Products resulting from this innovation may include an AI model or composite of a therapeutic plus an associated model in order to benefit the patient.
7. Early-stage companies access on-demand infrastructure and super-scale datasets: Collaborations and partnerships within the biopharma industry must change to accommodate an AI-focused research and development setting. These connections will be crucial to help facilitate early-stage companies’ access to super-scale datasets and on-demand computational infrastructure.
8. Refine approach to ai talent acquisition and development: With the evolution of AI, organizations will need to focus on two workforce groups: modest skilled individuals and highly skilled individuals. The first group consists of trained individuals who work at the “limit of their licensure” or beyond with AI augmentation. The second group will be AI-augmented and have direct control over multiple agents or super agents. Investments will need to be made not only into acquiring new personnel with these skillsets but ensuring enough members of the existing workforce can be upskilled into one of these categories.
9. Raise the bar on decision quality and speed: Comprehensive inquiries from human specialists will be incorporated into AI-assisted decision making to increase levels of confidence, clarity, and insight. Team members will need to pursue questions like “What insights did the series of deeply disease-specific LLMs and agents provide for the complex relationships surrounding this patient group’s exceptional response to this novel therapeutic?”
10. Reset time and productivity expectations: The addition of AI to scientific research and clinical decision making is not bound by time or days. It can operate up to the maximum capacity of any computing infrastructure that it is granted access to, which completely recalculates the time to decide expectations of outcomes.
The best way to prepare for and incorporate new innovations, such as AI, into your organization’s strategy is to accelerate industry partnerships to deepen and broaden access to data. Increased access to data will transform clinical development, translational science, and clinical care in order to guarantee and expedite medical advancements.
AI innovations will continue to progress, leading to more advanced and capable models along with deeper pushes to evolve the traditional healthcare and life sciences enterprise. The White House’s new AI Action Plan is an example of how AI innovation and guidelines could change or revolutionize experimental methodologies and hypothesis development. As new mandatory standards are developed, organizations that ignore these guidelines risk falling behind in innovation and violating compliance. Organizations that create their own AI Action Plan to meet these requirements will be better prepared to encounter additional AI innovations or regulations.
Photo: Jirsak, Getty Images
Jeff Elton, Ph.D., is Vice Chairman of ConcertAI, an AI SaaS solutions company providing research and patient-centric solutions for life sciences innovators and the world’s leading providers. Prior to ConcertAI, Jeff was Managing Director, Accenture Strategy/Patient Health; Global Chief Operating Officer and SVP Strategy at Novartis Institutes of BioMedical Research, Inc.; and partner at McKinsey & Company. He is also a founding board member and senior advisor to several early-stage companies. Jeff is currently a board member of the Massachusetts Biotechnology Council. He is the co-author of the widely cited book, Healthcare Disrupted (Wiley, 2016). Jeff has a Ph.D. and M.B.A. from The University of Chicago.
This post appears through the MedCity Influencers program. Anyone can publish their perspective on business and innovation in healthcare on MedCity News through MedCity Influencers. Click here to find out how.