PATH Recommendations Regarding Developing a Software Pre-certification Program: A Working Model (v0.1- April 2018)
The Partnership for Artificial Intelligence and Automation in Healthcare (PATH) is pleased to provide these comments to the Food and Drug Administration regarding a potential software pre-certification program. Improving patient outcomes is a core objective for all of PATH’s activities. Appropriate regulation of medical software is critical to provide assurances of quality, to guard against abuse and assure the safety, effectiveness, and accessibility of services.
We acknowledge and express appreciation for FDA’s extraordinary openness and transparency with this regulatory initiative. At this stage, in what we understand will be a very iterative process, our primary recommendation here is to exercise caution for the pre-certification process to be a continuous success that ensures product and patient safety while improving outcomes and productivity.
To assure such early success, we suggest that, initially, pre-certification should be focused on organizations that can demonstrate trust, notably with a record of positive compliance with FDA standards for “software as a medical device” (SaMD) and/or medical devices with a significant number of FDA approved products. It is better to look at an objective record of actual performance, rather than their “commitment.”
An organization’s mere commitment to excellence principles seems less than sufficient as a starting point. It is better for manufacturers and FDA to gradually lessen scrutiny than to have to tighten it. Further, experience with organizations who have demonstrated trust can best guide rapid setting and refining of standards for the rest of the SaMD regulatory process.
We urge consideration of a second pre-certification track for fitness/wellness and other apps with substantial consumer purposes. Another factor in expanding the eligibility for pre-certification could be consideration of the riskiness of the manufacturer’s existing product line.
Similarly, PATH recommends FDA pursue separate appraisal models for each level of pre-certification based on the manufacturers demonstrated level of trustworthiness. Since pre-certification is an accommodation by FDA, violations of trust should warrant discharge to regular order.
Finally, PATH recommends that the discussion of the five excellence principles be more specific. In particular, we recommend stating needs for software efficiency, such as clinical accuracy and number of false positives, and for privacy and security of any retained patient information. Also, for the benefit of pre-certification, there should be rigorous requirements for real world performance data monitoring.
The Partnership for Artificial Intelligence and Automation in Healthcare (PATHHealth.com) is a broad alliance of stakeholders working together to improve patient care, expand access and reduce healthcare costs through the use of automation, robotics, and artificial intelligence. These technologies have already transformed banking, aviation, and entertainment allowing each to grow, provide higher-quality products, lower costs and provide consumers greater choice. Faced with spiraling costs, increased demands and decreasing resources, healthcare desperately needs to catch up.
We look forward to working with FDA and coordinating with our members as this initiative evolves.
Jonathan D. Linkous
Chief Executive Officer, PATH