Fri, Feb 8, 2019
Experimental design and multiscale measurements. Latent model – how do you deal with certain things? Every minute you might see different types of data. Output of classifier.
You have these risk sensors, hand to mouth gestures. A little point process you see over the course of the day.
There is a chance that this is part of a smoking algorithm. An indicator where all these sequences of ones are part of a smoking episode. On top of that, a self-report throughout the day, which is three times per day.
We’re focused on how you smoke, moderating the risk. People get paid for how certain things work, a strong indicator that they smoked. Self-assessment and reflection. Map all this information onto the latent smoking process.
Question, Evidence, Conclusion.
There will be a window of smoking lengths, a probability that there will be part of a smoking gestures. Hand in mouth being part of that. One could try to use that information, very noisy, complex data. Everything is good in this respect.