The American Medical Association continued its efforts to get its arms around the use of augmented intelligence this month, endorsing a new set of policy proposals under which the group advocates for a seat at the table as healthcare AI systems become more important to care delivery.
The recommendations aren’t incredibly detailed, but they do offer an idea of what the AMA worries about when it comes to the adoption of emerging health technology. In particular, they reflect the hard lessons physicians learned from participating in the EHR incentive programs rolled out by CMS.
At the AMA’s recent annual meeting, the delegates in attendance agreed to a series of recommendations on the use of AI in healthcare to boost patient care, improve population health, cut costs and support physician satisfaction.
The group said it plans to advocate for payment and coverage of healthcare AI systems:
* Which are informed by real-world workflow and human-centered design principles
* Which help physicians transition to new care delivery models
* Which foster communication and engagement among patients, physicians and the healthcare team; which integrate clinical, administrative and pop health management functions into workflow and;
* Which integrate end-user feedback into an iterative design process.
These are all worthwhile objectives, but pretty conventional stuff nonetheless.
Of greater interest (to me at least) was the extent to which the AMA delegates are looking at public policy. In the recommendations, they said want to see governments support the development of AI systems which small practices and patients can afford, rather than leaving such products solely in the hands of big medical groups and healthcare systems.
I don’t know if the group expects to see a new program in which the government funds AI as it did EHRs, but clearly, the delegates want to cover their bases in this regard.
The AMA recommendations also want to see to it that government programs don’t punish physicians who don’t adopt AI systems, at least while regulatory oversight structures, standards, clinical validation and usefulness and standards of care are still fluid. In particular, the delegates went on record as opposing mandates by payers, hospitals, health systems or government organizations requiring the use of AI as a condition of licensure, participation, payment or coverage.
Also, the AMA recommendations assert that if mandates are ever put in place to use AI, all liability related to the use of such technologies should rest with those issuing the mandate. The group said that developers of autonomous AI systems whose clinical features handle aspects of screening, diagnosis and treatment should be managing liability concerns related to system failures or misdiagnoses.
My takeaway from all of this is that AMA members can envision a time when healthcare AI technology plays as distinct and powerful a role as EHRs have. As I noted last June, when the AMA introduced its first set of AI recommendations, the group has every reason to get into the game early and make sure it doesn’t get burned this time around.