Shaalan Beg MD MBA FASCO’s Post

View profile for Shaalan Beg MD MBA FASCO, graphic

Modernizing cancer clinical trials. Medical Oncology. Digital Health Care. Drug Development

'SOUP' can ruin your AI-Enabled medical device development? "The process of fine tuning foundation models for more specific medical purposes may introduce significant quantities of Software of Unknown Provenance (SOUP), as the manufacturer performing the fine tuning may only have limited documentation for the base foundation model.' The "Good Machine Learning Practice for Medical Device Development: Guiding Principles" document outlines essential guidelines for developing safe AI-enabled medical devices. Guiding Principles: 1- Understanding Device Purpose and Multidisciplinary Expertise An understanding of a device's intended use within clinical workflows and leveraging multidisciplinary expertise throughout the product lifecycle is crucial.(AKA involve your clinicians!) 2-Good Software Engineering and Security Practices: This includes robust software engineering, usability, data quality assurance, data management, and cybersecurity. 3- Representative Clinical Study Participants and Datasets: Prevent bias and promote generalizability by ensuring datasets and study participants reflect the intended patient population (eg demographics and clinical conditions). 4- Independence of Training and Test Datasets: To improve validity, the 'training' and 'test' datasets should be independent.

I’m excited to share that we at FDA with our colleagues at Health Canada | Santé Canada and the United Kingdom’s Medicines and Healthcare products Regulatory Agency (MHRA) are jointly issuing the “Transparency for Machine Learning-Enabled Medical Devices: Guiding Principles.” For AI/ML-enabled medical devices, effective transparency using a human-centered design approach ensures that right information related to risks and patient outcomes is communicated to all the people who could be interacting with the device to help make informed decisions. #digitalhealth #ai #patientcentered

Transparency for Machine Learning-Enabled Medical Devices

Transparency for Machine Learning-Enabled Medical Devices

fda.gov

To view or add a comment, sign in

Explore topics