Multiple datasets from separate trials in NARX Time-Series

I am trying to train a dynamic NARX time-series from multiple datasets gathered from separate trials measured at the same time.

Here is the scenario:

There are five people taking data at exactly the same time but in different locations around the world and this data is measured on a daily basis for 250 days: Martha, Joan, Rupak, Marlow, Vladamir are taking the data. For arguments sake they are doing the pollution example from the narxnet example set.

Each person measures the following 8 different measurements at each time but in a different location:

1. Temperature 2. Relative humidity 3. Carbon monoxide 4. Sulfer dioxide 5. Nitrogen dioxide 6. Hydrocarbons 7. Ozone 8. Particulates

The output or target is only total mortality.

1. Total mortality

Let's say that the delays are 1:5 for both input and feedback delays.

The group are going to measure on each day at the same time. Now let's say Martha is out sick on day 5 and Joan and Marlow are out on day 7. How do we set up the NARX network for this problem given that the total number of trials varies on each day. In terms of a dynamic network this won't work correctly. Also can somebody give guidance on how to setup this problem. My understanding is that the NARX needs a for loop essentially to create a network that will operate on each persons data separately and not altogether? If the trials are all combined then the network would by default not use each persons set of data and only the first persons.

There is a huge amount of confusion on how to do this properly and there is no guidance in the documentation. Any advice or help would be hugely appreciated.

Thanks.