Gaussian Process–Based Bayesian Optimization of Lower-Limb PNS–TMS Interstimulus Intervals in Adults and Children: A Proof-of-Concept Toward a Personalized Framework

The human brain is a master of timing. It does not simply react to the world; it anticipates it, weaving together sensory signals from the body with commands from the mind in a precise, split-second dance. When a nerve in the leg sends a signal upward, and a magnetic pulse hits the brain at just the right moment, the two can combine to either boost or dampen the muscle's response. This interaction is the foundation of a technique called paired stimulation, where researchers try to understand how the nervous system learns and adapts. For years, scientists have used this method to study the arms and hands, but the legs remain a mystery. The nerves in the legs are longer, the pathways are different, and the timing required to make them work together is likely unique to each person. If doctors want to use these techniques to help people recover from injury or manage chronic pain, they first need to know exactly when to deliver the second pulse. Too early, and the signal gets lost; too late, and the window of opportunity closes. The question is whether there is a single perfect moment that works for everyone, or if the brain's clock is set differently for every individual.

A team of researchers at the Université de Montréal set out to solve this puzzle by testing healthy adults and children. They wanted to find the precise moment when stimulating a nerve in the leg would most effectively boost the brain's command to the foot muscles. To do this, they used a device to send a mild electrical pulse to the common peroneal nerve, which runs down the side of the leg and controls the muscles that lift the foot. A few milliseconds later, they used a magnetic coil to tap the part of the brain that controls those same muscles. By changing the time gap between the electrical pulse and the magnetic tap, they could map out how the leg muscles responded. They tested twenty-six different time gaps for each person, ranging from very short delays to longer ones, to see which timing produced the strongest reaction. They did this with ten adults and eight children, carefully measuring the electrical activity in the muscles to see if the combination of stimuli made the muscles more active or less active than usual.

The results showed that the timing was everything. In the adults, the response changed dramatically depending on the delay. When the electrical pulse came just before the magnetic tap, the muscle response often dropped, as if the brain was briefly suppressing the signal. But when the delay was longer, the response grew stronger, suggesting the brain was now amplifying the command. This pattern was not the same for everyone. While the group as a whole showed a general trend, the specific moment that worked best varied from person to person. Some adults needed a shorter gap to get the best boost, while others needed a longer one. The researchers also found that the intensity of the electrical pulse mattered. When they used a stronger pulse that actually made the foot move, the timing effects were clear and strong. When they used a weaker pulse that only felt like a tingle, the brain's response was much quieter and harder to predict.

The study took a different approach with the children. Because the protocol was long and demanding, the children only underwent the stronger stimulation condition and performed fewer repetitions of each timing test. The results here were more complex. Like the adults, the children's muscles responded differently depending on the timing, but their patterns were much more scattered and varied. One child might show a strong boost at a specific delay, while another showed almost no change at the same moment. This suggests that the developing nervous system in children is not just a smaller version of the adult system; it operates with its own unique and shifting rules. The researchers found it much harder to predict the best timing for a child based on a few tests, whereas the adult patterns were smoother and easier to model.

To make sense of all this data, the researchers used a sophisticated computer method called Gaussian process regression. Think of this as a smart way to draw a smooth line through a messy set of dots. Instead of just looking at the average result for each time gap, the computer learned the shape of the curve for each individual person, filling in the gaps between the tested times to guess what would happen at any moment. This allowed them to see the full picture of how each person's brain and leg interacted, even with limited data. They then used this model to run a virtual experiment called Bayesian optimization. Imagine trying to find the highest peak in a foggy mountain range. You could walk randomly, checking every spot, which would take a long time. Or, you could use a map that gets smarter with every step you take, guiding you toward the peak more efficiently. The researchers used this "smart map" approach to see if they could find the best timing for each person with far fewer tests than the full twenty-six they originally performed.

The findings were encouraging for the adults. The smart map method successfully identified the best timing for nine out of ten adults, and it did so in about half the number of steps that a random search would have taken. On average, the smart method needed only ten tests to find the peak, while random guessing often needed twenty or more. This suggests that for adults, it is possible to personalize the treatment quickly without exhausting the patient with endless trials. However, the same method was less effective for the children. While it still found the best timing for some, it struggled more with the others, often requiring many more tests to reach the same level of confidence. The children's responses were simply too variable for the computer to predict with high certainty based on a small number of tests.

The study also tested whether knowing the average results of the whole group could help speed up the process for a new individual. They tried starting the search with a "population-informed" guess, assuming that a new person would behave like the others. This did not provide a consistent advantage. Sometimes it helped, but often it did not, and in some cases, it made no difference at all. This reinforces the idea that every person's nervous system is unique, and a one-size-fits-all starting point is not reliable. The researchers also checked that the procedure was safe and comfortable. The electrical pulses were mild, and the magnetic taps were painless. Most participants reported very little pain or fatigue, and their ability to move their legs did not weaken after the session. One child did stop the session early, but only because they lost patience, not because of any physical discomfort.

Ultimately, this work proves that the timing of nerve stimulation is a highly personal variable. There is no single "magic moment" that works for everyone. For adults, the patterns are stable enough that a smart, data-driven approach can quickly find the right timing for an individual. For children, the picture is still too blurry, and more research is needed to understand how their developing brains respond. The study does not claim to have solved the problem of how to treat neurological conditions, but it provides a crucial proof of concept. It shows that by using advanced modeling to personalize the timing of stimulation, we can move away from guesswork and toward a more efficient, tailored approach. This could one day help doctors design therapies that are not only more effective but also shorter and less burdensome for patients, whether they are adults or children. The path forward lies in recognizing that the brain's clock is set individually, and the tools to read it are finally becoming available.