NICHE OR COMMON PRACTICE? TRENDS IN AI-ENABLED PLATFORM USE: A CASE STUDY IN ONCOLOGY SLR PROTOCOLS PUBLISHED IN THE COCHRANE LIBRARY
Author(s)
Hoda Fotovvat, PhD1, Fani Koutentaki, MS2, Grace E. Fox, PhD1, Ella Jones, MS3, Odette Megnin-Viggars, PhD3.
1OPEN Health, New York, NY, USA, 2OPEN Health, London, United Kingdom, 3Open Health, London, United Kingdom.
1OPEN Health, New York, NY, USA, 2OPEN Health, London, United Kingdom, 3Open Health, London, United Kingdom.
OBJECTIVES: Artificial intelligence (AI)-enabled tools are increasingly promoted as transformative for evidence synthesis used to support regulatory and health technology assessment (HTA) decision making; however, published evidence describing their use in practice remains limited, including in oncology. This study evaluated reporting of use of AI-enabled platforms in protocols for oncology systematic literature reviews (SLRs) published in the Cochrane Library by examining temporal trends in reporting, reported platform use, and workflow applications.
METHODS: The Cochrane Library was searched for oncology SLR protocols published between 1998 and 2025 using terms including “cancer,” “neoplasm,” “tumor,” and “malignancy.” Protocols were reviewed for reported use of AI-enabled platforms, defined as review platforms marketed as incorporating AI features. As some platforms include both AI-enabled and non-AI features, categorization reflected reported platform use rather than confirmed AI functionality. Information extracted included platform name, year, and reported workflow tasks supported by platform use.
RESULTS: An analysis identified 232 oncology SLR protocols published in the Cochrane Library. Reported use of AI-enabled platforms increased from 1 in 5 protocols (20.0%) in 2014 to 19 in 24 protocols (79.2%) in 2025. Overall, 57 protocols (24.6%) reported platform use; none were published before 2014. The reported platforms were Covidence (51 [89.5%]), Rayyan (5 [8.8%]), and DistillerSR (1 [1.8%]). In the 57 protocols, 59 workflow applications were reported: screening (42 [71.2%]), combined screening/data extraction (8 [13.6%]), record management/duplicate removal (6 [10.2%]), and data extraction only (3 [5.1%]).
CONCLUSIONS: Reported AI-enabled platform use increased substantially over time; however, reported use remained focused on screening. More standardized, feature-level reporting may support transparent assessment of platform use and consistent evaluation by regulators, HTA bodies, and other evidence users. Although derived from oncology SLR protocols published in the Cochrane Library, these findings may have broader relevance as AI-enabled approaches become increasingly used in evidence generation.
METHODS: The Cochrane Library was searched for oncology SLR protocols published between 1998 and 2025 using terms including “cancer,” “neoplasm,” “tumor,” and “malignancy.” Protocols were reviewed for reported use of AI-enabled platforms, defined as review platforms marketed as incorporating AI features. As some platforms include both AI-enabled and non-AI features, categorization reflected reported platform use rather than confirmed AI functionality. Information extracted included platform name, year, and reported workflow tasks supported by platform use.
RESULTS: An analysis identified 232 oncology SLR protocols published in the Cochrane Library. Reported use of AI-enabled platforms increased from 1 in 5 protocols (20.0%) in 2014 to 19 in 24 protocols (79.2%) in 2025. Overall, 57 protocols (24.6%) reported platform use; none were published before 2014. The reported platforms were Covidence (51 [89.5%]), Rayyan (5 [8.8%]), and DistillerSR (1 [1.8%]). In the 57 protocols, 59 workflow applications were reported: screening (42 [71.2%]), combined screening/data extraction (8 [13.6%]), record management/duplicate removal (6 [10.2%]), and data extraction only (3 [5.1%]).
CONCLUSIONS: Reported AI-enabled platform use increased substantially over time; however, reported use remained focused on screening. More standardized, feature-level reporting may support transparent assessment of platform use and consistent evaluation by regulators, HTA bodies, and other evidence users. Although derived from oncology SLR protocols published in the Cochrane Library, these findings may have broader relevance as AI-enabled approaches become increasingly used in evidence generation.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
MSR24
Topic
Methodological & Statistical Research
Disease
Oncology