Healthcare AI Summit

AI advancements are transforming diagnostic tools to provide clinicians with more detailed information which is significantly enhancing their diagnostic capabilities. This introductory session will lay the foundation for the essential data science principles that healthcare providers need to understand about vital AI and ML concepts.

• Defining artificial intelligence (AI) and machine learning (ML) 
• Differentiating between traditional, generative and agentic AI 
• Understanding core concepts including natural language processing, neural networks, and deep learning
• Exploring how these concepts are weaved together to expand medical device and diagnostic capabilities
• Delving deeper into the AI lifecycle from data collection and model training to testing and validation

This session will provide a comprehensive overview of current and evolving regulatory pathways for AI enabled medical technology, while also addressing grey areas concerning digital health. Each day more and more healthcare organizations and medtech manufacturers are forming collaborations for the development of AI tools and AI enhanced diagnostics.

Panelists will provide MedTech companies and innovative healthcare providers with practical insights on how to navigate product submissions and post-market surveillance.

• Using the 510(k) and De Novo pathways to evaluate the safety and effectiveness of AI enabled medical devices predicated on an existing device 
  • Demonstrating testing and risk mitigation steps were taken in accordance with FDA pre-market guidance
• Assessing whether the 510(k) channel is appropriate for any medical device that includes AI 
• Clarifying when software as a medical device (SaMD) becomes subject to FDA regulation 
  • Examining current SaMD carveouts for clinical support tools (CDS)
  • Wellness vs. Medical device
• Exploring the FDA’s total product lifecycle approach to AI/ML devices that are adapting to real world data 
  • Predeterminate change control plans (PCCP)
  • Algorithm change protocol (ACP)
• Regulatory exemptions for models being developed and used by medical practitioners internally

The FDA recently launched a generative AI tool called Elsa to help agency staff work more efficiently. Elsa will be used to accelerate time consuming tasks, including product safety and scientific evaluations. This session will examine the FDA’s strategy to regulate in real time, safeguards to ensure responsible use and insights on how industry should prepare for AI augmented–and eventually–generated reviews.

• Gain insights on what Elsa is and how it is currently being used by the FDA to streamline product reviews
• Analyzing FDA guardrails to protect industry data, prevent cyber-attacks and avoid hallucinations  
  • Takeaways from the agency approach to AI development and deployment
  • Assessing how industry can assist with flagging accuracy and quality issues with Elsa outputs
• Preparing to tailor submissions for AI interpretation

The use of AI devices and technology for patient care has given rise to new concerns about how protected health information is being used. The flow of PHI between hospitals and business associates to train, develop and ultimately use AI in clinical practice is often murky. This session will provide insights on how to remain HIPAA compliant and adhere to new and evolving state regulations that seek to keep up with innovation and reinforce patient protections.

• Overview of HIPAA and HIPAA covered entities
• Identifying privacy risks presented by AI-enabled devices and health tools
• Examining scenarios where there could be inappropriate access or misuse of PHI and how to protect against it when working with manufacturers and business associates
  • Developing data-use agreements to address those risks
• Sourcing and aggregating data to train AI devices in a HIPAA compliant manner
• Mitigating the risk of reidentification
• Examining how HIPAA applies to ‘payviders’ who are using AI
• Navigating the grey area of third-party digital health applications not covered by HIPAA 
• Round-up of applicable state privacy laws 

• Exploring the potential for AI to help facilitate more personalized care
• Emphasizing the importance of usability, and human-centered design of these products
• Understanding the limitations of AI for clinical care
• Evaluating current and future tools: diagnostic, predictive and generative AI  
  • Examining the potential for different prompts to affect care recommendations
• Exploring the role of the physician in monitoring: identifying bias, hallucinations, or other inaccuracies within the AI algorithm
• Navigating clinical decision making the potential liability when you use it and when you don’t but could have
• Navigating patient consent to the use of AI for various purposes ranging from documentation or to clinical decision support 
  • Establishing what amounts to meaningful patient consent

• Examining CMS pathways to reimbursement for the use of AI enabled tools
• Assessing current provider uptake of AI medical devices and how private payers are viewing them 
  • Tracking payer trends around prior authorization and step therapy
• Demonstrating of the economic value of AI medical devices 
  • Using RWE (real world evidence) to continue to validate diagnostic and predictive capabilities

For MedTech companies and physician innovators, AI presents an incredible opportunity to bring groundbreaking insights to clinical practice. Securing IP for new medical technology is a critical part of the commercialization journey. However, the addition of AI brings with it new questions concerning patentability, inventorship and data rights. This session will examine the components of AI devices that are patentable; while also exploring how different types of collaborations between MedTech and providers can affect IP rights.

• Identifying what components of the invention are patentable
  • Determining whether algorithms, and AI prompts can be patented
  • Navigating explainability challenges for ‘black box’ autonomous AI tools
• Determining whether the medical technology company, the institution or doctor holds IP rights on an AI powered device
  • Assessing the type of agreements between parties and policies that could influence IP ownership
• Mitigating infringement risks and the difficulty with identifying infringement due to the evolution of models
• Protecting AI model training as a trade secret

A robust cyber security strategy involves a complex web of responsibilities among healthcare stakeholders. The manufacturer, healthcare provider and even the patient have a role to play in ensuring the device is updated and secure with the proper controls in place to protect data and the integrity of the AI model. This session will examine the scope of responsibility for various stakeholders and strategies for exposing hidden vulnerabilities. Presenters will also share insights on how to develop a cybersecurity/compliance plan that encompasses the full lifecycle of a device from design through to decommissioning.

• Understanding cyber risks for AI systems and tools being used in healthcare
• Determining how the AI medical tool/system interacts with your organization and third parties
• Developing a cybersecurity strategy that covers the whole lifecycle of an AI medical device
• Examining current and developing NIST standards on AI and cybersecurity to help guide your approach
• Red teaming AI healthcare tools to identify data leakage, bias, and security vulnerabilities