Robot Blood Draws, GenAI & Reclassifications
In a world where medical technology is evolving at breakneck speed, the latest innovation could change the way we experience blood draws forever. This week on Trying to Keep Up, Allison Komiyama and Michael Nilo dive into the FDA’s authorization of the first robotic phlebotomy system, Vitestro’s Aletta. The device not only draws blood but also identifies veins using near-infrared light and ultrasound, inserts needles, and even disposes of them, all while putting on a bandage. Allison and Michael discuss where automated blood draws could fit into healthcare, particularly as phlebotomist shortages and access challenges continue to affect care delivery.
The conversation then turns to the FDA’s new discussion paper on generative AI-enabled medical devices. Rather than issuing formal guidance, the FDA is asking stakeholders to weigh in on some of the hardest questions surrounding GenAI: How should manufacturers evaluate systems with variable or non-deterministic outputs? What should premarket evaluation look like? How should risk and performance be assessed? And where do existing AI/ML regulatory approaches fall short?
The team also digs into the FDA’s unusually active year for device reclassification. Eight reclassifications have already occurred in 2026, compared with 51 entries in the reclassification database since 2013. They discuss why some legacy Class III devices may make more sense as Class II products today and how newer De Novo authorizations can help demonstrate that older classifications deserve another look.
Additional topics include:
Lower-extremity exoskeletons and their potential uses in healthcare and industry
FDA’s proposed Class I treatment of certain medical device accessories
What counts as an “accessory” for blood pressure and ECMO systems
Why seemingly minor device components can still carry meaningful regulatory risk
The role of public feedback in shaping FDA’s approach to generative AI
FDA’s latest commissioner nomination
What Allison will be speaking about at the MEDevice conference in Boston
Why robots drawing blood somehow seem less terrifying than robots outrunning humans