A tumor doesn't grow like a tree; it grows like a city: today's skyline is shaped by yesterday's construction and the fires of the day before. Standard tumor growth equations don't account for this: they only look at the current moment, as if the city were rebuilt from scratch every day. The authors added "memory" to the model: now the equation remembers what came before, and that changes the forecast.

Such memory is described by fractional derivatives — a mathematical tool that stretches the influence of the past into the present. Instead of an integer order (the usual derivative), a fractional order, less than one, is used. It's like viewing tumor growth through a lens that slightly blurs the past but doesn't erase it completely. The model was calibrated on data from five patients with real measurements of tumor volume.

The result: for all five patients, the fractional model matched observations more closely than the classical one. Each patient had their own optimal fractional order — some had longer memory, others shorter. This means that each person's tumor grows in its own way, and there is no universal equation.

The difference is visible in the graphs: the trajectories of the fractional and standard models diverge like two roads that run side by side at first, then one veers off. This isn't just a mathematical detail — the shape of the curve determines when a doctor notices dangerous growth and has time to intervene.

For now, this is only a first step. In the future, the model could be extended to account for treatment and uncertainty, and tested on hundreds of patients. But already it's clear: to understand how a tumor grows, you need to remember where it started.