IIAIM generates rigorous, independent evidence on how artificial intelligence performs in clinical practice. We also identify where it falls short.
Why evidence matters
Many AI tools report excellent accuracy during development. Their performance often changes with new patients, institutions and equipment. Claims that AI matches or surpasses clinicians vary widely in methodological quality. Many rest on retrospective data, small test sets or comparisons that do not reflect clinical practice. Patients, clinicians, health systems and regulators need a trustworthy, independent synthesis of this evidence.
Research priorities
- AI versus clinician performance. Systematic reviews and meta-analyses of diagnostic and prognostic accuracy, including head-to-head and clinician-with-AI comparisons.
- Living evidence maps. Regularly updated maps of AI evidence by specialty, condition and imaging modality. The maps show where the evidence is strong and where it is missing.
- Real-world performance. External validation, performance across sites and over time, and post-deployment monitoring.
- Equity and bias. Whether AI performs equally well across populations, care settings and countries.
- Safety, regulation and reporting quality. How investigators design and report AI studies, and how regulators assess them.
- Education research. Which forms of training help clinicians use AI well.
How we work
- We register protocols before work begins, for example in PROSPERO or on the Open Science Framework.
- We follow recognized reporting guidelines, including PRISMA 2020, PRISMA-DTA, TRIPOD+AI, CONSORT-AI and DECIDE-AI.
- We assess risk of bias with established tools, such as QUADAS-2 and PROBAST, and rate the certainty of evidence with GRADE.
- We share data, code and extraction forms openly whenever possible.
- We disclose funding sources and conflicts of interest in every output.
- We assign authorship according to the ICMJE criteria.
Collaborate with us
We welcome collaboration with clinicians who contribute specialty expertise, and with methodologists, engineers and institutions. Students and trainees may join projects under supervision and gain experience in evidence synthesis.