Predictive AI Adoption Rises in Hospitals, But Access Remains Uneven
The use of predictive artificial intelligence (AI) within hospitals is growing, but significant disparities in adoption exist, particularly between rural and urban facilities, and larger versus critical access hospitals.While predictive AI – which leverages machine learning to forecast outcomes like patient readmission risk - has been present in healthcare for years, its integration into hospital workflows has accelerated over the past decade.
Recent analysis of survey data from the american Hospital Association reveals that 71% of non-federal acute care hospitals reported utilizing predictive AI integrated with their electronic health records in 2024, an increase from 66% in 2023.
The most significant year-over-year gains were observed in three key areas: streamlining and automating billing processes, optimizing appointment scheduling, and identifying high-risk outpatients who would benefit from follow-up care. Though, applications involving direct health monitoring and treatment recommendations remain less prevalent, likely due to concerns surrounding potential errors. Researchers suggest adoption in these areas may increase as hospitals gain confidence with the technology in less critical applications.
Despite the overall upward trend, access to predictive AI isn’t uniform. Only half of critical access hospitals – small facilities located at least 35 miles from another hospital – reported using the technology last year, compared to 80% of non-critical access hospitals. A similar gap exists between rural and urban hospitals, with adoption rates of 56% and 81% respectively.
Implementing and maintaining these AI tools presents challenges for healthcare providers. Triumphant adoption requires establishing robust governance structures and continuous monitoring to ensure ongoing accuracy and performance.
Recognizing these challenges, a large majority of hospitals are actively evaluating their predictive AI tools. The ASTP report indicates that in 2024, 82% assessed accuracy, 74% checked for bias, and 79% conducted post-implementation evaluation or monitoring. These evaluations are typically overseen by multiple entities, with nearly three-quarters of hospitals assigning accountability to several task forces, committees, or department leaders.
This increasing adoption of predictive AI contrasts with the current state of other AI applications, such as Generative AI. while interest in Generative AI is growing,implementation remains limited,with few pilot programs progressing to full-scale deployment.
