It’s not easy to get depression-detecting AI through the FDA
TL;DR
California-based startup Kintsugi spent seven years building AI that detects signs of depression and anxiety from how someone speaks – not what they say, but vocal patterns. After failing to secure FDA clearance in time, the company is shutting down and open-sourcing most of its technology. Some components may find new life outside healthcare, including deepfake audio detection. Mental health diagnosis still relies largely on questionnaires and clinical interviews rather than objective lab-style tests.
Nauti's Take
Seven years of work, a genuine clinical gap identified – and it ends on regulatory timing. This isn't a technology failure; it's a structural one.
Bringing clinical AI to market requires not just good models but deep pockets and patience for FDA processes that most startups simply can't sustain. The open-source decision deserves credit – too often these efforts vanish without a trace.
The potential pivot to deepfake detection is also a reminder that solid foundational research tends to find its uses, even when the original application doesn't make it.
Briefingshow
Kintsugi is a textbook example of how hard it is to translate a promising AI concept into clinical practice. FDA clearance for psychiatric tools is exceptionally difficult to obtain, requiring rigorous validation that most early-stage startups struggle to fund and sustain. The open-source release at least ensures the underlying research isn't lost – and could provide a foundation for academic or better-funded teams to continue the work.
Sources
