What's the right model for predicting a time-series value?


Problem statement: Predict the value of a person’s A1C (the average sugar level for 3 months) given a set of conditions.

Let’s say the A1C is affected by the carb intake (grams), exercise time (minutes), current weight, and water intake. This is a made up example to illustrate the problem.

In a typical time-series prediction (stock price), we have values for several years and then build a model to learn from that data and help predict the next value. However in our case say we are performing a 9 month study where we follow 100 people and gather values for above variables.

What kind of machine learning model can be built with a data set 100 daily values for 9 months that will predict the A1C value for another person given values for the variables?

Essentially we have 270 days of data for 100 patients. I dont think we can simply “concatenate” the values to arrive at 27000 data points for training. Is my assumption correct? What’s the right model for this kind of data?