PULSE CHECK: COMPARISON OF REPORTED AI-ASSISTED METHODS IN ONCOLOGY AND CARDIOVASCULAR DISEASE PROSPERO-REGISTERED SLR PROTOCOLS
Author(s)
Hoda Fotovvat, PhD1, Odette Megnin-Viggars, PhD2, Grace E. Fox, PhD3, Ella Jones, MS4, Fani Koutentaki, MS5.
1OPEN Health, Bethesda, MD, USA, 2Open HEALTH, London, United Kingdom, 3OPEN Health, New York, NY, USA, 4Open Health, New York, United Kingdom, 5OPEN Health, New York, United Kingdom.
1OPEN Health, Bethesda, MD, USA, 2Open HEALTH, London, United Kingdom, 3OPEN Health, New York, NY, USA, 4Open Health, New York, United Kingdom, 5OPEN Health, New York, United Kingdom.
OBJECTIVES: Artificial intelligence (AI)-assisted methods are widely discussed for their potential to support evidence synthesis; however, their reported use in practice remains unclear. This study compared reported use of AI-assisted methods in systematic literature review (SLR) protocols registered in PROSPERO focused on oncology and cardiovascular disease (CVD) by examining temporal trends, reported platforms, and workflow tasks to identify therapeutic-area differences in reporting and use.
METHODS: PROSPERO was searched for oncology or CVD SLR protocols registered from 2020 to 2025 using disease-relevant terms. Ongoing and completed protocols were screened for mentions of AI, named AI-enabled tools, or review-management platforms with AI-enabled features. For protocols referencing AI, reported platforms and workflow stages were extracted. Reporting of AI use, platform use, and workflow applications were compared over time and between therapeutic areas.
RESULTS: Over the timeframe of interest, fewer PROSPERO SLR protocols were related to CVD (4,287) than oncology (53,005). However, AI-assisted methods were reported more than twice as frequently in CVD than oncology protocols (1.64% vs. 0.71%). Reporting increased over time in both areas, from 0.23% in 2020 to 1.08% in 2025 for oncology and from 0% in 2020 to 2.29% in 2025 for CVD. Rayyan and Covidence were the most frequently reported platforms in oncology (49.0% and 37.6%) and CVD (47.1% and 32.9%). Reported use of AI focused on screening in both therapeutic areas, with limited reporting for data extraction or analysis.
CONCLUSIONS: AI-assisted methods were reported more frequently in CVD than oncology PROSPERO-registered SLR protocols; however, this difference may reflect variation in adoption of AI-assisted methods, reporting practices, protocol detail, or therapeutic-area-specific factors. Across both areas, reported use was concentrated in screening, and similar platforms were commonly reported. More standardized reporting guidance is needed to clarify how AI-assisted methods are applied in SLR protocols and to support consistent evaluation of AI-assisted evidence synthesis practices.
METHODS: PROSPERO was searched for oncology or CVD SLR protocols registered from 2020 to 2025 using disease-relevant terms. Ongoing and completed protocols were screened for mentions of AI, named AI-enabled tools, or review-management platforms with AI-enabled features. For protocols referencing AI, reported platforms and workflow stages were extracted. Reporting of AI use, platform use, and workflow applications were compared over time and between therapeutic areas.
RESULTS: Over the timeframe of interest, fewer PROSPERO SLR protocols were related to CVD (4,287) than oncology (53,005). However, AI-assisted methods were reported more than twice as frequently in CVD than oncology protocols (1.64% vs. 0.71%). Reporting increased over time in both areas, from 0.23% in 2020 to 1.08% in 2025 for oncology and from 0% in 2020 to 2.29% in 2025 for CVD. Rayyan and Covidence were the most frequently reported platforms in oncology (49.0% and 37.6%) and CVD (47.1% and 32.9%). Reported use of AI focused on screening in both therapeutic areas, with limited reporting for data extraction or analysis.
CONCLUSIONS: AI-assisted methods were reported more frequently in CVD than oncology PROSPERO-registered SLR protocols; however, this difference may reflect variation in adoption of AI-assisted methods, reporting practices, protocol detail, or therapeutic-area-specific factors. Across both areas, reported use was concentrated in screening, and similar platforms were commonly reported. More standardized reporting guidance is needed to clarify how AI-assisted methods are applied in SLR protocols and to support consistent evaluation of AI-assisted evidence synthesis practices.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
MSR255
Topic
Methodological & Statistical Research, Real World Data & Information Systems, Study Approaches
Disease
Cardiovascular Disorders (including MI, Stroke, Circulatory), Oncology