How would you approach a heat map that would show different level of activity?

I am interested in making an AI-inspired after the alien isolation AI. The AI there has 2 parts. A pointer and a searcher. The pointer only gives the AI a rough location of where you are. The difference in my pointer and alien isolation's pointer is that mine will be more of a "heat map" showing which areas have more activity by either having too many people walk, or sprinting or causing any sort of noise will cause the alien to see that your area is lighting up in the heat map. For example, sprinting would cause your area to light up but will also cause other areas to do also giving the AI more uncertainty. So there are multiple ways you can approach this. My idea was 10x10x10 chunks that would be assigned with "point" values. The higher they're, the more likely there is a human there. Sprinting in a hallway would cause the entire hallway in the AI's heat map view to turn bright then quickly turn unbright. From that information, the AI would be alarmed and switch into a searching mode attempting to search for the running player. Since the player is running and causing a trail behind him, the AI could easily follow it. Another thing is that the AI could re-evaluate the target value. A single person running away from it is not worth as much as an entire group of people moving which would cause a large effect on the AI's heat map. The problem is that this is not really "dynamic" and is stuck to 10x10x10 cubes. On top of that, having to constantly create arrays over and over again can become a lag issue. My question is are there better ways of approaching this issue or is my current plan better?