Designing and Creating Testing Datasets for Medical AI/ML End to End Testing
Testing datasets are important for AI/ML end to end testing in a real production environment, which is different than testing datasets for model testing. This paper talks about how I designed and created testing datasets for AI/ML end to end testing (for Mayo Clinic) using a real example from one of my projects. I also share how to protect Patient Heath Information in testing datasets, the issues/difficulties I had during dataset creation and the solutions for those issues and problems.
Nancy McCormack
Experienced SQA professional with over 1 year AI/ML verification and validation in Mayo Clinic, 8 years of management experience and 14 years of testing experience working in semiconductor, networking and IT industries. Accomplished at managing/leading diverse quality assurance teams using waterfall and agile methodologies and an expert collaborator with development, support and customers. Adept in all aspects of the SDLC but excel in test strategy, planning, process, standards, metrics etc.