I'm pretty sure you have enough data for the training and testing of the neural network - whether from the datasheet, or the data set that someone here gave you in another comment. Another option is to conduct some battery testing to add to your data (discharge in various currents and temperatures). But you'll still need some sensors in your system- temprature and the current drawn- those are the main paramters in estimating battery performance. Then you can train and test the model.

Hi, I actually worked in battery testing about 2 years ago, and a colleague of mine was assigned to predict exactly that, only with simple regression function not full ML system. So first, good luck and PLEASE make an update post if you do find something intersting... 🙏🏾 Second, I don't know your background in electronics so I'll say that make sure you have all the data you need, mainly room temperature, battery temp., life cycles the battery has been through, estimated energy drawn from the battery at the moment etc.

Good luck!