The Fall at Kilometer 21: The Medical-Data Gap in the Amateur Running Boom

At 8:49 a.m. on October 11, on a race course in Seoul, fewer than 500 meters remained in the half marathon. Yoo Ji-won collapsed on the road itself, in the very stretch every distance runner knows to be the harshest — the final 5 percent, when heart rate, core temperature, and nervous-system arousal all peak together.

On-site medics were there. CPR. An ambulance. A transfer to hospital. She did not survive. Several outlets called it "cardiac arrest." No official cause has been released.

Yoo Ji-won, 25, born in 2026, was a graduate researcher in AI and semiconductors at KAIST. She began running in 2026 but only moved into serious performance training in the summer of 2026. Yet within roughly eight to twelve months she reached the "sub3" milestone — 2:56:32 at the Seoul Marathon in March 2026. That same month she ran a half marathon in 1:22:01. Her personal best over 10 km was around 36 minutes.

The race she entered was Seoul Race 2026, the 24th edition, drawing about 12,000 athletes across two distances, the half marathon and 11 km. This is a mass-participation race, not part of World Athletics' road-race label system.

I have spent years documenting races in East Africa, where a weekend long-distance event can double as a national selection trial. And what I have learned is this: in distance running, performance data is always published, while medical-safety data is almost always left blank. That is the largest gap in this entire story.

Start with the performance data, because it tells you her true position within the amateur running boom.

A marathon of 2:56:32, against a women's world record of roughly 2:09–2:10, is about 36 percent slower. A half marathon of 1:22:01, a pace of 3:53 per kilometer, against a world record of roughly 1:02–1:04, is about 28–30 percent slower. A 10 km of about 36 minutes is roughly 24 percent slower than the road world record.

Set against the Boston Marathon standard of 3:00:00 for women aged 18–34, her marathon sits comfortably inside qualifying range. She belonged to the strong-recreational tier, just below the professional band — not an average jogger. For someone with only about a year of focused training, that rate of improvement is fast.

This is where I want to pause. Fast progress is not, in itself, a fault. But it is a signal about training load. From someone running for health in 2026 to someone running a sub-3 marathon in 2026, the training volume has to climb a steep slope, not a staircase. And all the while, she was a full-time graduate researcher.

Her daily rhythm, as it has been recounted, was this: in the lab from 9 a.m., a roughly two-hour run at 8 p.m., sometimes returning to the lab after training. A day that burns both cognition and physical capacity. Recovery — sleep, nutrition, rest — gets compressed. That is the pattern I call "recovery compression," and it is common among amateur athletes who hold down knowledge-work jobs while training hard.

The key point: the collapse occurred in the final 500 meters, precisely the zone where, by epidemiology, exertional events cluster most densely. This says nothing about any specific individual's cause. It says only that the location and timing of her collapse match the general pattern of exertional cardiac events — nothing more.

And here is the counterintuitive part.

After a death, what the press usually does is count achievements: how many kilometers, how many races, how many medals, "the pretty girl who runs." I have seen this before, in another instance: Nigerians were watched, not seen. Here, she is watched as an internet phenomenon, not read as an athlete.

The right question for a 12,000-person race is not how fast she ran. The right question is: how many defibrillators did that race have, where were they placed, how many medical stations, how long from collapse to defibrillation, and did the organizer require cardiovascular screening before the start?

Not one of those facts has been released. The organizer said only that it would "review related details and take necessary measures." That is a procedural signal, not yet a safety commitment.

The problem is not one race. The problem is speed. The amateur running boom is expanding faster than the health-screening infrastructure that serves it. A mass-participation race in Asia can now start tens of thousands of people, many of whom have never had an ECG, never known whether their heart carries an abnormality.

The people who pioneered this running boom did their part well: they put tens of thousands onto the roads. What remains is to build the medical infrastructure to keep up. East Africa does not lack running talent; neither does the amateur wave — it does not lack runners, it lacks people who record the medical data.

At 25, she was in the ascending phase of a running career, not the descending one. Her performance ceiling was still rising. That is why this death is more shocking — and why youth cannot be used as a safety filter.

In every running analysis there is a blank column no one wants to fill: defibrillator density, response time, cardiac-screening rate. Those numbers do not climb leaderboards and do not generate views. But they are the only thing that can save a person at kilometer 21.

I am not looking for a level playing field. I draw the lines of the field myself. And this time, the line I want to draw sits in exactly the data column that every mass-race organizer is leaving blank.