In an article published in the journal npj | Digital Medicine, researchers investigated the use of autonomous artificial intelligence (AI) for diabetic eye disease (DED) testing in primary care settings. They aimed to determine if this technology could close care gaps across different patient groups and result in higher adherence rates to DED testing.
Study: Autonomous artificial intelligence for diabetic eye disease increases access and health equity in underserved populations. Image Credit: TSViPhoto/Shutterstock.com
Background
DED affects many people with diabetes mellitus (DM) and can lead to blindness if not detected early. Early detection and treatment are crucial, but many patients remain undiagnosed because DED often has no symptoms in its early stages. Guidelines recommend annual eye exams for people with diabetes to catch DED early. However, barriers such as lack of transportation, time, or awareness often prevent patients from getting these exams, particularly in underserved communities.
To address this issue, the FDA approved an autonomous AI system called LumineticsCore (formerly IDx-DR) in 2018. Developed by Digital Diagnostics in Coralville, IA, this system analyzes retinal images at primary care clinics and diagnoses DED without needing a specialist. While previous studies have confirmed the system’s accuracy and safety, its impact on overall health outcomes and health equity remains less understood.
About the Research
In this paper, the authors comprehensively reviewed electronic health records from Johns Hopkins Medicine, which has over 30 primary care sites. They examined data from adult diabetes patients at these sites in 2019 (before AI was used) and 2021 (after AI was implemented).
The sites were divided into two groups: “AI-switched” (sites that started using autonomous AI by 2021) and “non-AI” (sites that did not use AI). The researchers compared changes in adherence to annual DED testing between these groups and among different patient subgroups based on demographics and social factors. They used propensity score weighting to adjust for potential confounding variables.
Research Findings
The study included 17,674 patients in 2019 and 17,590 in 2021. Overall, adherence to annual diabetic eye disease (DED) testing slightly increased from 42.2% in 2019 to 44.8% in 2021. At AI-switched sites, adherence significantly rose from 46.1% to 54.5%, while at non-AI sites, it remained around 40%. The difference in adherence between AI-switched and non-AI sites was 7.6 percentage points, which was statistically significant after adjusting for confounding factors.
The authors also found that autonomous AI improved access and equity for historically disadvantaged groups. At AI-switched sites, adherence for Black or African American patients increased by 12.2 percentage points, for Native Hawaiian or Other Pacific Islander patients by 19 %, for Medicaid-insured patients by 13.7 %, and for those with high area deprivation index (ADI) scores by 11.7 %.
In contrast, adherence for these groups decreased by 0.6 percentage points at non-AI sites. AI deployment helped reduce care gaps between these groups and others based on race, insurance, and ADI. Specifically, for Medicaid-insured patients, adherence increased by 13.7 percentage points at AI-switched sites but decreased by 0.9 percentage points at non-AI sites. Additionally, the gap in adherence rates between Asian and Black/African American patients decreased from 15.6% in 2019 to 3.5% in 2021 at AI-switched sites.
Applications
The paper shows that autonomous AI for diabetic eye disease testing can increase care gap closure rates and overall adherence to annual eye exams. This can help prevent vision loss from diabetes and improve patients' quality of life. The authors suggest that autonomous AI can enhance access to retinal evaluations and promote health equity for historically disadvantaged groups who face more barriers to eye care. They provide evidence that implementing autonomous AI in primary care settings, especially in underserved areas, is feasible and effective. Additionally, AI could improve the quality and efficiency of care, as well as the performance and reimbursement of healthcare systems.
Conclusion
In summary, the study demonstrated that autonomous AI could be highly effective for managing DED in large healthcare systems. The researchers highlighted AI's potential to improve population-level outcomes and health equity for diabetes patients. Moving forward, they recommended large-scale deployment of this technology to further increase adherence rates and reduce the burden of DED. They also suggested evaluating the impact of autonomous AI on population-level metrics, as well as patient trust and acceptance of the technology. Furthermore, they proposed assessing and quantifying the follow-up rate and healthcare costs associated with autonomous AI.
Journal Reference
Huang, J.J., Channa, R., Wolf, R.M. et al. Autonomous artificial intelligence for diabetic eye disease increases access and health equity in underserved populations. npj Digit. Med. 7, 196 (2024). DOI: 10.1038/s41746-024-01197-3, https://www.nature.com/articles/s41746-024-01197-3
Disclaimer: The views expressed here are those of the author expressed in their private capacity and do not necessarily represent the views of AZoM.com Limited T/A AZoNetwork the owner and operator of this website. This disclaimer forms part of the Terms and conditions of use of this website.