A car's speedometer shows the speed right now, but to know whether the car will reach its destination in a year, a single glance at the gauge isn't enough. That's roughly how conventional formulas for kidney failure risk work: they rely on one blood test and miss how values have been shifting over months and years. The new CLARK model (longitudinal assessment of kidney failure risk) reads the entire "medical history" in motion, like watching the road rather than a single point on the dashboard.

This approach markedly improves long-term prediction accuracy. Conventional statistical models achieved an accuracy of 0.516, while CLARK reached 0.541—a seemingly modest gain, but enough to more confidently identify people who will soon need help. The system analyzes routine lab tests, such as the glomerular filtration rate (the kidneys' ability to filter blood)—and spots worrying trends well before the critical point.

Thus, the history of lab results turns from a mere archive into a navigation system that signals when it's time to act, and helps delay the moment when the speedometer needle drops to zero.