How to train an LSTM model with data that has multiple input rows per day but only one row of label/ground-truth (output) data per day

I am doing a sleep data science experiment and I need a model that outputs multiple columns sleep quality measurement values (that are decimal numbers) for each input.

For training, I collected data using a smartphone (for input data) and a smartwatch (for label or output data). The smartphone collects MULTIPLE rows of sensor data such as accelerometer and gyroscope for ONE night. The smartwatch generates a SINGLE row of sleep quality indicator values (such as TWAK, NWAK, WASO, etc…) for ONE night. The structure of data looks as follows:

The above is a representation of my data. The left side is the phone feature data (with an undefined number of records per day based on how long the person was sleeping on that day) and the right side is the sleep quality measure data which is only one value per day. I need to train a model with this data and online sources seem to suggest LSTM is the way to go. Can you please give me a suggestion on how to do this for my type of dataset structure?

Thank you in advance! I can explain it however many times you need so feel free to ask me any question.