STAT AI: An Emergency Medicine Podcast · Ethan Abbott and DJ Apakama

EP20 Michael Prats, MD and Sean Sethi, DO

·1 hr 6 min·4 clips

A social network for AI agents reached over 1.5 million registered agents within days.

The episode centers on what AI might realistically do for emergency ultrasound. DJ and Ethan begin with the idea that the tool should make scanning more efficient. They describe a useful middle ground where AI helps learners improve their views. One example is software that can point out missing anatomy and suggest how to move the probe. That kind of feedback could make training less frustrating. They then move into the more ambitious territory of image interpretation. Here the goal is not just correction but actual recognition of pathology. They talk about systems that might flag a concerning finding and explain what it sees. B lines and pulmonary edema come up as the sort of pattern an algorithm might eventually identify. The conversation stays cautious about how far that capability has really progressed. They emphasize that ultrasound quality varies a lot from person to person. It also varies across equipment and how the image is obtained and processed. That variability makes machine learning harder than it looks. Even so, they note that the field keeps moving with each new model and update. They also return to the limits of current teaching tools. AI can help point at anatomy, but it is not yet consistently good at helping learners fix a bad image. The discussion turns practical again when they ask what would make the technology more useful in real training. One answer is clearer machine feedback when a view is not adequate. That kind of honesty from the software would save time and cut down on guesswork. The conversation stays focused on what would actually matter in emergency medicine. The result is a measured look at a tool that is improving, but not finished. It leaves the listener with a clear sense of both the opportunity and the gap.

As heard by us

AI becomes concrete through ultrasound training, image correction, and the limits of interpretation support.

The conversation makes AI in emergency medicine feel less abstract once it reaches ultrasound, where the value is practical: faster scanning, steadier training feedback, and image support that can help clinicians get more from each exam.

Read the full review in PlayNext →

Why you'd press play

Hear how AI could make emergency ultrasound more efficient, clearer for learners, and more useful at the bedside.

Read the full recommendation in PlayNext →

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from STAT AI: An Emergency Medicine Podcast