Искусственный интеллект в скрининге диабетической ретинопатии снижает затраты и повышает эффективность лечения

Study: Cost savings, improved access to care seen with AI-based DR screening Key takeaways: - In an analysis, annual average cost savings per patient ranged from $2.70 to $155.90. - AI was comparable to or exceeded human graders in specificity and sensitivity. MONTEREY, Calif. — Alongside improved clinical efficiency and follow-up adherence, AI-based diabetic retinopathy screening yielded positive reductions in cost for patients, according to a poster presentation. The study, presented at the Women in Ophthalmology Summer Symposium, aimed to measure the impact of AI-based screening in DR, including its impact on clinical performance and patient economics, “with a focus on standardized cost-effectiveness across healthcare systems,” Sophia Guccione, of UMass Chan Medical School, and colleagues wrote. The literature review used PubMed, Google Scholar and Scopus to compile studies that measured the diagnostic performance, patient perspectives or economic outcomes of AI-based screening between 2016 and 2025. The researchers analyzed standardized annual cost savings per patient in Australia, China, Singapore, Thailand and United Kingdom. According to the poster, AI-based screening for DR demonstrated sensitivity and specificity levels of more than 90% across multiple studies, “performing comparably to or exceeding human graders in diverse clinical settings.” Through same-visit diagnosis and real-time analysis, primary care and community settings that implemented AI-based screening also showed improved follow-up adherence and screening completion. Modeled national savings exceeded $10 million annually, with absolute savings varying across healthcare systems, according to the poster. The study concluded that annual average cost savings per patient ranged from $2.70 to $155.90. Savings were attributed to reductions in transportation needs, missed workdays and out-of-pocket costs. “This suggests that health system-only analyses may underestimate the true value of AI-based screening,” Guccione and colleagues wrote. While there are barriers to AI implementation, including inconsistent reimbursement framework and knowledge gaps among both patients and healthcare professionals regarding AI-based care, the researchers concluded that AI-based screening is a “scalable solution” for improving access, efficiency and outcomes for DR care. “High-income settings with higher labor costs generally showed larger per-patient savings, while lower-cost settings showed smaller but still positive gains,” the researchers wrote.