SUDEP stands for Sudden Unexpected Death in Epilepsy, and 70% of cases occur during sleep.
1. cc: Life Science Podcast centers this episode on Soteria’s Chorus bed and its work on SUDEP prevention.
2. Norman Wen is the CEO of Soteria, and he explains why the company built a self-positioning bed with AI-driven software.
3. The episode asks how a bed can detect sleep posture, reposition a sleeper, and support both epilepsy care and broader sleep wellness.
4. Wen defines SUDEP as Sudden Unexpected Death in Epilepsy and says it is a catch-all term for epilepsy-related deaths without another specific cause.
5. He says 3,000 people die of SUDEP each year in the US, that the number is probably under-counted, and that 70% of cases occur during sleep.
6. Wen says 90% of the time people are found face down in the prone position, and he links that detail to the technology behind Chorus.
7. Current care, as he describes it, focuses mainly on seizure management through medication and on repositioning a person to their side in hospitals.
8. The bed uses pressure sensors embedded in the sleep surface to create a pressure map, which the system uses to infer sleep posture.
9. Wen contrasts that with camera-based tracking and says the pressure-sensor approach avoids constant video recording.
10. He says the machine learning challenge is harder because there are not many public datasets for sleep posture, unlike image datasets used for other AI systems.
11. Wen also says the bed is not dependent on the cloud, because epilepsy intervention cannot rely on the internet staying online.
12. The autonomous version of the system uses pneumatically driven cells that can move the sleep surface up and down in a modular architecture.
13. He says accurate position recognition matters because the system has to know where a person’s body, limbs, and angle are on the bed.
14. He says the bed must also detect whether someone is actually asleep, because repositioning a waking person would not work.
15. Wen adds that the actuation has to be quiet enough to avoid disturbing sleep while the bed is moving someone into a safer position.
16. He describes the team as spanning robotics, mechanical engineering, embedded systems, sensors, machine learning, and software engineering.
17. The interview style is technical and conversational, with the host repeatedly checking details about sensors, data, and sleep-position use cases.
18. The tone stays practical and product-focused, moving from epilepsy risk to consumer sleep health and market entry.
19. People interested in epilepsy care, sleep technology, and AI hardware would likely follow this closely.
20. People seeking entertainment-first conversation or light medical discussion may skip it.
