Assertion Supported AI assessment confidence: 95% certainty 4/5 debate potential 2/5

Shah: Almost all FDA AI regulation to date targets diagnostic products

Munjal Shah · Hippocratic AI’s Munjal Shah: Building the First Safety-First LLM for Healthcare · Aug 23, 2023 · at 40:47

Munjal Shah is the co-founder and CEO of Hippocratic AI. He explains why Hippocratic AI chose not to focus on medical diagnostics.

0:00 / 0:04exact quote · 4.9s

▶ Watch the full episode on YouTube → 720p mp4 · rendered on demand · StarZero watermark

“Almost all your AI regulation to date from the FDA has been on diagnostic products.”

quote is from the automated transcript, cleaned for reading: filler sounds and stutters are removed, nothing is rephrased. names can be misheard (the analysis reads context, assessments check outside sources). how →

The claim, plainly

Munjal Shah claims that almost all FDA regulation on artificial intelligence to date has focused on diagnostic products.

a quote alone can be ambiguous, so this restates it with the references and context of the conversation around it resolved. AI-written; the tape below the player is the source.

Assessment

FDA records and academic analyses of AI/ML-enabled medical device clearances prior to August 23, 2023 confirm that the vast majority of regulated AI products were diagnostic tools or diagnostic Software as a Medical Device (SaMD). Cross-sectional studies of FDA authorizations through late 2023 show that diagnostic imaging in radiology alone accounted for approximately 76% of all cleared devices, with cardiovascular and neurology diagnostic tools making up most of the remainder (combining for over 90% total). In contrast, therapeutic or prescriptive AI devices accounted for less than 1% of authorizations.

Studies of FDA AI/ML device clearances through late 2023 show radiology alone accounted for ~76% of all authorized devices, primarily for automated image analysis, computer-aided detection, and diagnostic decision support.
Cardiovascular and neurology diagnostic devices accounted for another ~13%, bringing the total diagnostic share among top panels to over 90%.
Therapeutic AI devices comprised a negligible fraction (under 1%) of total FDA AI authorizations prior to August 2023.

Checked against

[1] medtechdive.com [2] mdpi.com [3] nih.gov [4] questglobal.com [5] nih.gov [6] intuitionlabs.ai [7] amegroups.org [8] appliedradiology.com

FDA AI ML enabled medical devices breakdown diagnostic August 2023 FDA AI ML medical devices diagnostic vs therapeutic percentage FDA public records AI ML enabled medical devices diagnostic tools percentage FDA AI-enabled medical devices diagnostic vs therapeutic percentage prior to August 2023 FDA "AI/ML-enabled medical devices" "diagnostic" software as a medical device 2023 FDA list of AI ML enabled medical devices radiology percentage "FDA" "AI/ML" "diagnostic" "therapeutic" percentage devices 2023 FDA artificial intelligence "diagnostic" "therapeutic" proportion medical devices

assessed as of 2026-07-31

This assessment is an AI reading of a public statement, made against the cited references with the confidence shown. It can be wrong. The quote, the tape and the references are all on this page; if something is off, use report an error below. full disclaimer →

Mentioned

diagnostic products FDA Munjal Shah

⚑ Spot an error? Report it

More from Munjal Shah

Prediction Not checkable as stated

Shah: AI models attempting medical diagnoses will kill patients

“I mean, they just keep going there. I'm going to do diagnoses because it's grand or it's challenging, but I mean, they're going to kill somebody.”

Munjal Shah Aug 23, 2023 ▶ 4:53 Hippocratic AI’s Munjal Shah: Building the First Safety-First LLM for Healthcare

Assertion Not checkable as stated

Shah: Patient non-adherence is healthcare's biggest problem, not diagnosis

“Our biggest issue in the U.S. And most of the developed world is the people don't follow the directions after they're diagnosed. Like, it's not an issue of diagnoses. It's an issue of adherence.”

Munjal Shah Aug 23, 2023 ▶ 5:59 Hippocratic AI’s Munjal Shah: Building the First Safety-First LLM for Healthcare

Assertion Not checkable as stated

Shah: Foundational LLMs are instruction-tuned for queries, not multi-turn conversations

“They've all instruction tuned for what I call a query, not a conversation.”

Munjal Shah Aug 23, 2023 ▶ 18:52 Hippocratic AI’s Munjal Shah: Building the First Safety-First LLM for Healthcare

Assertion Supported

Shah: Operating a voice LLM costs roughly 18 cents per hour

“A large language model speaking at a hundred words per minute will cost somewhere around 18 cents an hour. Okay. So the LLM plus what's the ASR cost, the automatic speech recognition cost, the text-to-speech cost, the TTS cost. We kind of put it all in there. …”

Munjal Shah Aug 23, 2023 ▶ 10:49 Hippocratic AI’s Munjal Shah: Building the First Safety-First LLM for Healthcare

Disclosure

Shah: Hippocratic AI is building its healthcare LLM from scratch

“That's why we're building our own model and we're building it from scratch.”

Munjal Shah Aug 23, 2023 ▶ 19:59 Hippocratic AI’s Munjal Shah: Building the First Safety-First LLM for Healthcare

Assertion Supported

Shah: Hippocratic AI outperformed GPT-4 on 105 of 114 healthcare exams

“And then we took it, and then we had GPT-IV take it, and we had all the other language models take it, and we beat them all. And we beat them on a 105 of a 114 for GPT-IV, for example.”

Munjal Shah Aug 23, 2023 ▶ 21:10 Hippocratic AI’s Munjal Shah: Building the First Safety-First LLM for Healthcare

Shah: Almost all FDA AI regulation to date targets diagnostic products

More from Munjal Shah

Turn any episode into a week of clips.