Patient safety should be the No. 1 consideration for healthcare organizations working, planning or hoping to adopt AI. Then again, nine other concerns are similarly crucial to the success of the industry-wide endeavor.
That’s according to CHIME, the College of Healthcare Information Management Executives. The association built its list after receiving input from a sizeable segment of its 5,000 members in 60 countries.
“CHIME members embrace AI as a force for good,” says the org’s board chair, Scott MacLean, chief information officer at MedStar Health, in introducing the report containing and explaining the 10 points to which, in CHIME’s view, attention must be paid. “By considering and using our AI principles, we can enhance patient care, empower clinicians and contribute to healthier communities.”
CHIME released the report this week. Here are excerpts.
1. Patient safety.
Everyone is a patient, and every patient deserves to receive care that best supports their individual healthcare needs. A one-size-fits-all policy approach to AI that is entirely sector-agnostic is unlikely to best support patient needs.
AI’s use in healthcare settings requires an added level of expertise and consideration, as patient care outcomes and lives are at stake.
2. Administrative efficiencies.
Whereas clinical tools can take years for clinicians to adopt into practice, uptake of administrative AI tools often doesn’t require clinical acceptance. This suggests a shorter period in which these tools will be widely used by healthcare delivery organizations.
It stands to reason that investments in AI tools designed to improve our sector’s efficiency could result in significant savings in time and cost.
3. Regulatory oversight.
As AI evolves in the way it is applied in healthcare, liability concerns are growing. Not least is the question of how much responsibility providers and clinicians should have to shoulder when they augment care delivery with the use of AI tools and there is an adverse patient outcome.
Regulatory oversight is needed, but it should not result in duplicative mandates or unnecessarily increasing administrative burdens on providers and clinicians.
4. Innovation and research.
AI technology is rapidly evolving, and the clinical evidence base is still emerging. Comprehensive research—shared via peer-reviewed journals and made widely available to providers—is needed.
Common definitions are needed to foster a shared understanding across diverse applications within the healthcare sector to support research and promote consistency.
5. Discrimination, bias and equity.
Ongoing efforts are needed to identify and address bias in AI algorithms and ensure that these systems are trained on diverse and representative datasets.
The implementation of AI in healthcare should not be a one-time event but rather an iterative process.
6. Affordability.
Most providers lack the resources needed to purchase and deploy cutting-edge AI tools and applications. Many remain financially stretched and are still wrestling with workforce shortages, especially for healthcare IT employees.
Small and under-resourced providers will need additional support adopting AI to prevent widening the digital divide.
7. Privacy.
When aggregated with other data, de-identified patient data can be re-identified.
AI products using de-identified data in an attempt to reduce bias have the potential to inadvertently create privacy risks such as reverse-engineering data to re-identify individuals.
8. Cybersecurity.
Third parties that store, process and/or transmit protected health information on behalf of HIPAA-covered entities are critical to the healthcare sector. However, they routinely shift millions of dollars of liability for a cybersecurity breach back to healthcare delivery organizations during contract negotiations.
If we are to make meaningful improvements in our sector, this responsibility must be equally shared and cannot be borne by providers alone.
9. High-speed broadband.
Continuing to expand nationwide high-speed broadband is crucial for harnessing AI tools and reducing healthcare disparities.
To the degree that gaps persist in broadband access, widespread AI implementation will be impeded, worsening the digital divide.
10. Education and workforce.
Positioning the healthcare sector as an AI leader requires a well-trained and upskilled workforce. However, getting there must be balanced with maintaining a strong employment rate, avoiding significant job displacement and mitigating existing inequities.
Support by large technology companies, educators and policymakers is needed to manage the use of these new tools and the changing labor demand.
Download the full report.
