Yujan Shrestha

Ideas on how to get foundation models and generative AI FDA cleared.

Recent advancements in radiology have led to the development of foundation models integrating vision and language capabilities, transforming medical imaging analysis

Command Line as a Medical Device (CLaMD) represents a streamlined approach to FDA clearance for medical software. By utilizing command-line interfaces instead of graphical user interfaces, developers can significantly simplify the regulatory process. This approach reduces UI hazard analysis requirements and minimizes human factors testing since interaction occurs through established systems like PACS or EHR.
CLaMD is particularly effective for AI algorithms, which can run automatically, output standardized formats, and operate server-side. This allows clinicians to continue using their familiar workstations while the AI works invisibly in the background. Numerous FDA-cleared devices have successfully implemented this pattern, demonstrating its viability in medical device submissions.

This is a recording of the talk Yujan gave at the RSNA AI Theater

Yujan discusses pragmatic postmarket surveillance to FDA advisory committee on GenAI

The discussion explores generative AI (GenAI) in medical devices, focusing on FDA's considerations for regulation, risks, and the lifecycle management of such technologies. Key points include the complexity of training data in foundation models, the risks of hallucinations in outputs, and the evolving role of post-market surveillance. Insights highlight challenges in balancing innovation with regulatory oversight and safety.

The conversation between Yujan Shrestha and J. David Giese focuses on making threat modeling more approachable and less intimidating. They share personal anecdotes to illustrate how threats can appear in daily life, then transition to discussing the distinctions between threats, vulnerabilities, and risks in cybersecurity. Broader definitions of harm and the STRIDE methodology are highlighted as essential tools for comprehensive risk management.

Dr. Elena Sizikova's presentation highlights the role of synthetic data in addressing data scarcity and privacy challenges in medical imaging AI. Synthetic data offers benefits like customizable variability, reduced risks, and improved bias control, particularly in fields like histopathology and pediatric imaging. She contrasts generative and knowledge-based models, emphasizing the potential of hybrid datasets for enhancing AI performance. The talk also stresses the importance of robust evaluation frameworks to validate synthetic data's effectiveness.

We discuss challenges in medical device cybersecurity, including balancing FDA guidelines with practical hospital and manufacturer needs. They highlight risks like network vulnerabilities and outdated dependencies, recommending best practices such as code reviews, automated testing, and prioritizing tool validation for critical security areas.

Yujan and David discussed the topic of FDA's performance in regulating AI medical devices, focusing on the ideal regulatory situation, the impact of AI on risk analysis, and the regulatory pathways for adding AI tools to existing software medical devices. They also explored the potential of continuous learning models in medical devices, the adoption of AI and ML in the US healthcare market, and the cybersecurity threats facing medical devices. The conversation ended with a discussion on the regulatory implications of continuous learning algorithms and the potential for a letter to file for a SAMD cleared by FDA for surgical planning.