Incubator for national platforms for systematic validation of AI in diagnostic imaging (startup)

Important results from the project

The project more than met its objective to establish three platforms for validating AI in diagnostic imaging. The largest pilot, VAI-B in breast radiology, built up a solid validation operation where four AI providers were evaluated with data from five regions. Similar but smaller platforms were built in pathology (VAI-P) and stroke radiology (VAI-S), and a feasibility study was conducted in prostate cancer. Another result is that the legal form of a quality register is considered applicable.

Expected long term effects

A global outlook in 2025 shows that validation is considered perhaps the most central issue for taking advantage of AI in healthcare on a large scale. The innovation environment and project efforts have thus been early in addressing these issues. We expect that the solutions and experiences from the project will have a major impact on AI implementations in Sweden and Europe, and thus in turn on the efficiency and quality of healthcare in general.

Approach and implementation

The arrangement with an overall incubator layer that connected the different initiatives has been very successful and led to good synergy and cross-talk. The arrangement with a mature, larger pilot (VAI-B) leading the way for the others has also been very efficient. The schedule has been followed well, and the most unexpected thing was that the budget efficiency led to an extra feasibility study being carried out.

External links

Samlat material om projektet och dess lärdomar, för breda målgrupper Introduktion till projektet