A CE mark and a vendor's study show that an imaging algorithm can work somewhere. Only a hospital's own scanners, patients and habits show whether it works there, and the cost of finding out belongs in the purchase rather than in the data team's spare time.
A CE mark and a vendor's study show that an imaging algorithm can work somewhere. Only a hospital's own scanners, patients and habits show whether it works there, and the cost of finding out belongs in the purchase rather than in the data team's spare time.
Hospitals have long shared research data by sending pseudonymised extracts to their partners. I argue that the extract should become the exception: partners bring their analysis to a controlled environment, and only checked results leave it.
Sponsors increasingly ask hospitals to describe their data as research-grade. No such general label is honest. Fitness depends on the question, and a hospital should answer each request with a short account of what its records can and cannot show for that purpose.