TRCG Success:
Post Market Surveillance
Challenge
A large global medical device conglomerate implemented a Machine Learning based Post Market Surveillance (complaint handling and reporting) solution.
Solution
OUR ASSOCIATES PERFORMED AN ASSESSMENT TO ASSESS THE FOLLOWING:
Audit risk of algorithm development process
Effectivity of solution for performing QMS relevant tasks
Sustainability of the solution
Outcome
We were able to determine that while audit risk was moderate, small changes to the use case definition and parameters would dramatically lower residual contextual audit risk.
While very effective for several modalities; some major modalities suffered from invisible defects and would remain ineffective for the foreseeable future due to a misalignment of
the algorithm model and the clinically relevant risk factor.
The algorithm sustainability plan was ineffective, and a retraining plan would be necessary to mitigate the limitations of the algorithm training set and avoid data bias driven systemic defects with high regulatory risk beginning in as little as 6 months.