AI-aided evidence-based medical guidelines (TRAM)
Medical guidelines are crucial tools for supporting clinical decision-making, improving care quality, and reducing variation in medical practice. These guidelines depend on systematic reviews of the scientific literature, but the sheer volume of publications makes the development and maintenance of high-quality, up-to-date guidelines increasingly time-consuming and resource-intensive.
The TRAM project—Tooling for Rapid Assessment of Medical evidence—explores how artificial intelligence (AI), particularly active learning, can support and accelerate this process. By integrating AI into the literature screening phase, TRAM aims to reduce workload while preserving methodological rigor, helping guideline development organizations keep pace with the latest evidence.
Progress
TRAM is a collaboration between Utrecht University’s AI & Knowledge Discovery Lab, the Knowledge Institute of the Dutch Federation of Medical Specialists, Cochrane Netherlands, and the Julius Center. The project has focused on developing and evaluating a pipeline in which AI supports the screening of scientific literature for use in medical guidelines.
In practice, the TRAM pipeline integrates tools like ASReview to prioritize relevant studies. A proof-of-concept study showed that this approach can substantially reduce the number of abstracts that need to be screened manually while still identifying nearly all relevant studies. This work is currently being extended to explore how AI might support other phases of the guideline development process, such as the weighing and interpretation of evidence.
Funding
This project and its resulting output are being funded by ZonMw, the Stichting Kwaliteitsgelden Medisch Specialisten (SKMS) and the Dutch Research Council in the “Corona: Fast-track data” 11 (2020/SGW/00909334)
People involved
- Laura Hofstee PhD candidate and Project Lead
The Knowledge Institute of the Dutch Federation of Medical Specialists, Utrecht University - Janke de Groot Advisor
The Knowledge Institute of the Dutch Federation of Medical Specialists - Tim Christen Advisor
The Knowledge Institute of the Dutch Federation of Medical Specialists - Lotty Hooft Advisor
Cochrane Netherlands and the Julius Center - Rens van de Schoot Advisor
Utrecht University