Heard an excellent podcast today on MedtechTalk, featuring Douglas Kelly MD MBA, a physician and venture capitalist who served 2020-2025 as deputy Center Director for Science at CDRH.
Find the podcast online at this link (recorded March 2025):
See Dr Kelly's Linked In page here.
And see his recent re-post of Niko Hems' and Kathryn Bowsher's Linked In articles, on why so many AI-based startups crash, sometimes afer pretty massive funding.
(And for more on his theme of the value of public service by working places like FDA, see Michael Lewis's book "Who is Government? The Untold Story of Public Service.")
##
Chat GPT 5.2 summarizing the one-hour podcast:
##
In this candid and substantive Med Tech Talk episode, Justin Klein, MD (Vensana Capital) interviews Douglas Kelly, MD, former Deputy Center Director for Science at FDA’s CDRH, for a behind-the-scenes tour of how device regulation actually works—and why so many promising technologies never reach widespread patient access.
Kelly traces his path from venture investing to FDA, recounting the moment Jeff Shuren essentially challenged him to “stop complaining and come help fix it,” and uses analogies to explain how outsiders routinely misunderstand what FDA needs and how decisions get made.
He then maps the internal structure of CDRH (the offices of health technology, review staff, policy, science/technology, emergency preparedness), emphasizing that the range of devices—spanning software, AI, implantables, diagnostics, and more—require CDRH to be “intellectually promiscuous,” with staffing and communication challenges that don’t resemble standard pharma-style review.
The core of the conversation is Kelly’s strategic rationale for the TAP program (Total Product Life Cycle Advisory Program): not as an “FDA fast lane,” but as a deliberate attempt to replace the industry’s serial, disconnected failures (FDA authorization → CMS coverage → CPT code → payment → adoption) with an earlier, parallel, relationship-driven process that aligns evidence expectations across FDA, payers, and professional societies—so companies can fund the right studies once and investors can price risk realistically. He actually took people to watch events like CPT and RUC meetings to get a downstream view of the commercialization of devices.
Along the way, Kelly offers a striking metric—only about 1.5% of breakthrough-designated efforts, in his interim analysis, achieve durable commercial success or broad adoption—arguing that the system burns enormous reviewer time on products that never see daylight, and that the next era needs clearer priority-setting and better FDA–CMS alignment rather than rigid “stay in your lane” silos.
The discussion closes on today’s pressures—hiring constraints, the fragility of mission-driven expertise, and the risk that indiscriminate staffing cuts and return-to-office mandates will undercut precisely the industry-savvy talent needed to modernize review (including AI)—but ends on an optimistic call to action: Kelly urges more private-sector leaders to do public service at FDA, because the long-term fix is shared understanding, durable relationships, and evidence pathways that get genuinely valuable innovations to patients without compromising safety.
