The challenges to health care technology adoption include:

• Investment requirements: For all actors, implementing novel technologies requires a baseline investment in digital infrastructure, significant computing power, access to large datasets and talented professionals to install and operate complex systems. Securing investment dollars to pay for these technologies is difficult for many health care organizations already operating on slim margins. While administrative tools can be tied to productivity gains or ideal direct cost savings, clinical technology innovations often lack direct reimbursement.³ Providers may struggle to develop a clear business case before implementing these tools — or be forced to pass cost increases on to payers, patients, employers and other purchasers including the government. This can limit adoption speed and value for the system at large.
  
  
• Trust: Clinician trust in artificial intelligence (AI) technology and concerns over data quality are significant barriers to adoption, as clinicians remain the primary arbiters of care decisions. Beyond the challenge of collecting the fragmented, unstructured and highly sensitive data necessary, significant concerns about accuracy and bias limit practical use. Many clinicians today have not been trained to use new tools and may not understand the flaws and biases inherent to these models. Working to ensure that AI systems are reliable, unbiased and integrated into clinical workflows is essential for widespread acceptance.
  
  
• Regulation and liability: Novel technologies supporting care delivery lack a clear regulatory framework to define appropriate testing procedures, use cases and human intervention. Stakeholders and patient advocates are calling for more precise regulations to clarify the safety and liability of AI tools in support of adoption decisions. If AI applications are tied to patient care, they are subject to U.S. Food and Drug Administration (FDA) approval. As of mid-2022, the FDA had approved roughly 350 applications for “software-based medical devices.” Without greater oversight and inter-agency coordination, few organizations will take the risk to rapidly embrace these technologies.

Solutions to health care technology adoption challenges:

• Physician training and engagement: As more products are launched and their impact is studied, stakeholders must share knowledge of new tools’ limits and biases. Medical education and graduate medical training programs must begin to integrate novel care delivery modalities and train providers to evaluate the accuracy of AI-generated outputs. In the long term, training systems must adapt their core curriculum and testing to reflect the rapid advance of technology-enabled health care guidelines. To improve ease of adoption, product developers and practicing clinicians can collaborate to enhance and deploy the most useful and trusted tools, while building a business case based on increased clinician capacity and workforce retention.
  
  
• Create products and tools: To generate additional returns on new health care technology innovation investment, organizations developing and implementing them must seek opportunities to collaborate with technology developers to reduce costs and create marketable tools and proprietary protocols. Partnering with established technology vendors and pursuing downstream product launches can help secure investment with a clear strategy to diversify revenue. Similarly, a goal to focus on products can help reprioritize the digital transformation necessary to build infrastructure and develop, launch and scale new tools within and across organizations.

• Regulatory frameworks: The federal government will play a key role in supporting clinical health care technology adoption by developing appropriate governance and liability frameworks. Current regulations can be advanced and adapted to meet the needs of AI. For example, molecules discovered by AI would still require thorough clinical trials before entering the market. For clinical decision support, AI models could leverage current medical licensing and credentialing procedures to determine the appropriate level of decision-making allowable.

Example: One major health system, seeking to innovate carefully, chose to partner with an external firm accomplished in digital technology, cloud services and cybersecurity. The goal: to dive deeper into data, in a secure way, to tackle the most complex medical problems, provide care wherever the patient might be, drive cures and develop products to diversify revenue. The health system’s researchers, doctors and IT staff will use AI and cloud computing on the external data, permissioned to the external party.