Developing Health Care AI Pilots Is Simple; Ensuring Their Success at Scale Is Difficult
As per a recent study conducted by Bessemer Venture Partners, AWS, and Bain & Company, merely 30 percent of GenAI health care pilots transition to production, with an even lower proportion achieving success. This situation arises partly because numerous health care organizations are in a “throw it at the wall and see what sticks” phase with GenAI. They are conducting various pilots without centralized coordination, planning, or metrics, hoping for ROI to emerge. Some pilots demonstrate potential and may be executed in a specific department, while others falter due to numerous challenges in the transition from pilot to scalability.
The current approach to GenAI in health care is chaotic and inefficient, failing to harness the transformative possibilities it could offer. A well-rounded strategy for developing GenAI is essential. Historically, health care has encountered difficulties moving innovations out of the proof-of-concept stage. Pilots frequently overlook the operational, financial, and cultural components crucial for sustained success.
One significant concern is misaligned objectives, as pilots cater to narrow interests instead of aligning with organizational priorities. Scalability considerations are frequently overlooked, disregarding operational and financial scalability, workflow integration, data interoperability, and regulatory adherence. Engaging stakeholders is vital; without this, adoption stagnates. Ambiguous ROI adds to the challenges of scaling. Integration issues occur when pilots function independently, leading to “pilot fatigue.”
To resolve these challenges, a platform-based model is essential. Platforms offer a cohesive infrastructure, accommodating various tools and integrating with existing systems. They emphasize interoperability, scalability, and adaptability, facilitating adoption throughout workflows. In the realm of AI, this strategy is vital, as it can boost productivity across thousands of workflows within a health system.
Indicators of excessive dependence on GenAI pilots include numerous small-scale initiatives lacking a defined route to system-wide implementation, an abundance of point solutions, minimal cross-departmental cooperation, insufficient metrics, indecisiveness from leadership, and sparse discussion surrounding infrastructure, funding, or long-term adoption.
To establish a GenAI platform, health care organizations should:
1. Leadership synchronization: Initiatives must receive executive backing and align with organizational objectives.
2. Real-world evidence: A factual basis against specified measures is essential.
3. Workflow integration: Tools should seamlessly integrate into workflows and with current technology.
4. Data governance: Ensure adherence to regulatory requirements and sustain patient trust.
5. Change management: Offer onboarding and training to mitigate resistance.
Furthermore, take into account the economic and cultural implications. End users might be doubtful, making it crucial to engage them from the outset. Assessing ROI using the Triple Aim framework can help direct AI investments: it should enhance the care experience, address population health and equity, and lower costs.
In the end, meaningful returns will arise from viewing GenAI as a strategic capability — crafted for scalability, embedded across systems, and aligned with both clinical and business objectives. By adopting this perspective, health care organizations can take the lead with GenAI, setting a standard for innovation, efficiency, and improved care.