Editor's note: The Digital Medicine Society clarified the Duke Institute for Health Innovation is not part of the implementation playbook team.
The Digital Medicine Society (DiMe) is partnering with Google and the Mayo Clinic to create a consensus-driven artificial intelligence implementation playbook for the healthcare industry.
DiMe aims to help the industry think further afield on AI—beyond assessing the safety and effectiveness of models and toward overcoming barriers to implementation.
The free resources may include advice on how to most effectively create an AI strategy and a return on investment calculator. Google will be the technical lead on the project, while the Mayo Clinic will provide strategic guidance on the implementation of AI models in health systems.
The playbook will clarify real-world challenges, financial viability, standard language, capabilities and applications of AI. It will be devised by a broad range of stakeholders including patient advocates, health system leaders and AI developers.
Two AI collectives, the Coalition for Health AI (CHAI) and the Health AI Partnership are working on ensuring the safety and efficacy of AI models in healthcare. CHAI champions the creation of a nationwide network of assurance labs; HAIP advocates an approach to AI safety that puts more tools and resources into the hands of health systems.
Jennifer Goldsack, CEO of DiMe, who sits on the board of CHAI, said that no one is doing the work to help health systems decipher the business case for AI and how it can transform the way they provide care.
“Until key decision makers are equipped with tools and frameworks on how to implement health AI—something established consortia have not yet solved—the value health AI promises will not be realized and the adoption of these products in clinical care settings will continue to be limited,” DiMe said in a statement.
National digital healthcare leaders have publicly said that health systems are overwhelmed by the number of digital health vendors all claiming to impact care delivery.
That’s part of why groups like the Digital Medicine Society and the Peterson Health Technology Institute have cropped up in the first place: to be able to help health systems make informed decisions about new technologies. Researchers and investors like Bessemer are also putting out resources attempting to offer the industry some guidance on AI.
Goldsack said that health systems need an AI strategy, not to just implement algorithms at random. Then, they need guidance on how to best implement the models that have the potential to disrupt workflows and operations for the better.
Health systems need to identify where their biggest opportunities are where AI could help. Goldsack said that would likely include a focus on identifying health issues upstream so patients don't end up in the emergency department.
There’s a small window for AI to prove itself for investors to continue to fund new AI tools and solutions, she noted. If AI isn’t delivering big returns within a few years, investment is likely to peter out, she said.
There’s danger that AI could become like the digital therapeutics industry, which experienced the new technology hype cycle and a huge wave of investment in 2021. Afterward, payer uptake lagged, big names like Pear Therapeutics went under and investments fell off.
Overall, the portion of the budget dedicated to AI is rather small, and health systems only have so many resources to work with, Goldsack said. If their first few investments in AI don’t improve efficiency, the cost of the solution won’t be justified by the marginal efficiency gains, which could spell trouble for uptake of the technology writ large.
With the DiMe AI Playbook, Goldsack hopes a system’s first few investments garner big returns, and the organization will be able to keep reinvesting in AI. The ROI calculator will be able to take into account the resources of the health system and their unique properties.
To date, many healthcare systems have already made investments in AI. Goldsack said that she commends health systems that have already dipped their toes into AI. But she still thinks the industry is large enough and a big enough part of the national economy that DiMe’s resources are going to help more parts of the industry make industry-altering decisions about AI.