Public data, private collaborator: will machine learning relocate medical knowledge?

Machine-learning applications that assist with medical diagnostics are often developed in public-private collaborations. In these collaborations, the public medical institution generally provides the medical data needed to build the applications, while the private collaborator provides the technical expertise to create applications which it then owns.

I will evaluate the extent to which the development of these proprietary applications represents a relocation of medical knowledge from public medical institutions to private technology companies. I will also investigate whether principles of open data and open knowledge where data sets are held in public hands under public guidance, could be an effective counterbalance to this relocation, thereby also introducing medical transparency. I will use a variety of methods including a comparative case study of machine-learning projects in ophthalmology, semi-structured interviews with stakeholders and participatory workshops.

This project will result in guidance for public-private collaborations and recommendations for policy makers.