Regulatory Challenges for a Machine Learning Solution

Regulatory Challenges for a Machine Learning Solution in the GxP Space

The presentation will show the implemented approach for the initial validation upon the Life cycle implemented for the development, release and maintenance of the Machine learning embedded in the SW Solution, focusing upon the determination of model performance (e.g., prediction accuracy and model sensitivity) and the adequate sizing of the dataset for the associated evaluation. In addition, the presentation will describe the established mechanisms by which the performance of the model is monitored and the criteria which may trigger a model update, in case data drifts are observed. The implemented Life Cycle allowed to create and maintain the required qualification documentation of Machine Learning to be embedded in the Validation documentation, which allows to meet the current regulatory requirements ensuring the Accuracy of the outputs generated by the Machine Learning solution and ultimately the compliance against the ALCOA+ expectations for the entire ecosystem.