SEAL Team ran seven seasons on CBS and Paramount+ before landing on Netflix this August — and the wider audience found two storylines the show treats with unusual seriousness. The team's leader develops a brain injury he hides from the men who trust him with their lives. His closest operator loses a leg and comes back to a body that argues with him. Neither storyline is played for shock. Both are played for exactly how disorienting, slow, and unglamorous these injuries actually are — which is, unusually for network television, close to the truth.
We build diagnostic AI research for exactly these two conditions. So we're using the show's audience to do something we think is more useful than a normal recap: tell you what's real, what's fiction, and what our own research — Project 26 and Project 26B — actually is and isn't, right now, honestly.
The Invisible Wound
Across Season 5, Jason Hayes's character starts missing things: a name, a step in a plan, the thread of his own sentence. Headaches that don't quit. Moments where his judgment — the one thing a team leader can't afford to lose — becomes the thing his team quietly starts worrying about. He hides it as long as he can, because admitting it means the job might end.
The show never puts a diagnosis on screen. It doesn't need to. What it's depicting — a brain injury whose symptoms show up as behavior before they show up as anything a scan can point to — is one of the most common and most under-reported injuries in medicine.
This is the real question our Project 26 research is built around: can an AI system flag likely intracranial hemorrhage from a CT scan fast enough to change how quickly a patient gets seen — without ever taking the decision away from the clinician reading the scan. The AEGIS AI ensemble scores each case; the clinician sees the result and decides. Benchmarked on the RSNA public dataset — 21,744 real CT studies — with strong, consistent accuracy across testing.
The demo above runs on our own validation cases. If you have your own head CT and want the AEGIS team to actually look at it, send the whole DICOM study as a single .zip file (not a single image or a screenshot) — export the full study folder from your PACS/viewer and zip it before uploading. This is not automated — it goes to a real person.
1. Not a medical device and not a diagnosis. AEGIS NeuroSentinel is a research-stage tool. Nothing it outputs is medical advice, a diagnosis, or a substitute for a qualified radiologist or clinician. It does not diagnose, treat, cure, or prevent any condition.
2. De-identified data only, as a DICOM study .zip. Remove all patient-identifying information (name, MRN, date of birth, institution) before submitting. Submit the full DICOM study (all slices, zipped) rather than a single exported image or screenshot, and only your own study or one you have the right and consent to share.
3. Liability release. By submitting, you agree to indemnify and hold harmless AEGIS, Dr Loh Kah Meng, and associated parties from any claims arising from use of or reliance on AEGIS outputs.
4. Not for emergencies. This is not an emergency service. If you have symptoms of concern, seek immediate medical care — do not wait for a response here.
5. PDPA compliance. All data handled under strict PDPA and applicable governance frameworks.
The Wound That Doesn't End at the Scar
Clay Spenser's character is hit by an RPG near the end of Season 5. The wound turns septic. Season 6 opens with the leg gone, and with it, an identity Clay had spent his whole adult life building around being an operator. The show spends real time on the parts most war stories skip: the phase after the surgery, when the body doesn't match the person's sense of themselves anymore.
What the show doesn't dwell on — and what almost every real amputee will tell you unprompted — is that the missing limb doesn't always feel gone. It itches. It cramps. It hurts, sometimes badly, in a place that a doctor can't touch because it isn't there. That's phantom limb pain, and it's far closer to universal than most people, including many newly injured patients, expect.
This is where we'd rather undersell than oversell. Project 26B is a pilot feasibility study built on a small published clinical dataset (15 amputee participants, gaze/EMG/movement recordings) exploring whether objective physiological signals correlate with phantom limb pain. At this sample size, our own internal review found the population too small to legitimately set diagnostic thresholds — so we haven't set any. No app. Just an honest, early signal worth continuing to study.
If you're living with phantom limb pain, or love someone who is, we're not going to pretend this research can help you today. But real lived experience is exactly what makes research like this better — and we'd genuinely like to hear from you.
Real EMG from five different transradial/upper-limb amputees in this dataset, played back on a loop — pick a subject below. The glow brightens and dims exactly as their forearm muscles did — brighter means more electrical activity. The right arm is drawn in silhouette because, physically, it isn't there anymore; the glow is what's left of it responding anyway.
