AdvaMed itemizes AI imperatives for ‘the entire healthcare ecosystem’

Attention all professionals involved in the development, marketing, purchasing, implementation or oversight of medical technologies equipped with AI and machine learning: You owe it to healthcare consumers to keep certain principles top of mind in everything you do with these technologies. 

The principles, eight in number, are laid out in a white paper posted by the largest medical device association in the world. 

“The entire healthcare ecosystem should be aware of the opportunity to better serve patients while sharing in the commitment to protect safety, security and privacy through responsible development of AI,” explains the association, D.C.-based AdvaMed. “These stakeholders include but are not limited to patients, healthcare professionals, healthcare providers, IT system integrators, health IT developers, IT vendors, medical device manufacturers and regulators.” 

Here are excerpts from five of the eight must-do’s on AdvaMed’s mind. 

1. Leverage existing regulatory frameworks and promote international alignment to ensure timely patient access to innovative AI/ML-enabled medical technologies.

The current premarket and postmarket regulatory frameworks are “fully able to ensure the safety and effectiveness of AI/ML-enabled devices, whether they use locked or adaptive AI algorithms,” the authors write. More:  

‘FDA’s oversight is guided by a risk-based framework that includes a rigorous premarket review process that assesses medical device performance, reliability and safety, as well as extensive postmarket monitoring and surveillance requirements after devices are authorized for sale.’

2. Protect privacy of patient data with transparency and consent.

The data required to build AI models and deliver AI-enabled solutions “should be collected transparently with appropriate informed notice and authorization,” AdvaMed states. 

‘Technology innovators should protect patient privacy in compliance with all applicable data privacy laws and regulations and implement industry best practices, international consensus standards and organizational measures to ensure data security, integrity and confidentiality.’

3. Enable access to data and utilization for the benefit of patients.

“There should be a clear definition of which stakeholders may access patient data and for what purpose(s),” the authors argue. “Adherence to the highest ethical and trustworthy standards in management of data should be prioritized.” 

‘Healthcare stakeholders, innovator personnel and external vendors should collaborate to deliver a clear understanding about what data is collected, how it is being used and how it is being protected.’ 

4. Develop and deploy AI/ML-enabled solutions responsibly and mitigate against unwanted bias in AI/ML-enabled medical technologies.

To identify and address potential unwanted bias in AI-enabled devices, “high-quality and representative data sets of the target patient groups are essential,” the authors maintain. “Manufacturers can mitigate unwanted bias prior to product release through careful and thorough data collection, analysis and curation.” 

‘Once on the market, ongoing monitoring, evaluation and/or validation may be needed. Manufacturers are responsible for postmarket requirements after deployment, and for reporting problems to the FDA to ensure continued safety and effectiveness.’

5. Promote access to and the adoption of AI technologies to serve patients.

“The current CMS reimbursement framework is based on a statutory foundation that did not contemplate the need to capture coverage, coding or payment for these types of new diagnostics and therapeutic technologies, including algorithm-based healthcare services.”  

‘Reimbursement frameworks should be established to capture the full value of new AI-enabled technologies, including the long-term financial savings associated with better health outcomes and earlier detection of diseases and the efficiencies gained by healthcare providers.’

Also on AdvaMed’s hit list: 

• Leverage AI-enabled and digital health solutions to facilitate and promote access to health care in rural and under-served communities to improve health equity.
   
• Educate the public, patients, and clinicians on the roles and value of AI/ML-enabled health technologies while prioritizing clinician and user training in AI/ML-enabled technologies.
   
• Deliver transparency, essential to patient-centered care, that contains the appropriate level of information necessary to ensure the safe and effective use of the device.

Download the full paper.