Transfer Learning for LSTM

Question

I have trained a LSTM model for time-series binary prediction task, and now i need to re-use the pre-trained model on similar task (e.g. target task). But the problem is the target task has less features (e.g. input shape (4,13)) than the source task (e.g. input shape (4,16)). In this case when i need to train the model on the target task, the input shape is incompatible and i cannot do the training. Is there any way i can use the pre-trained model of LSTM on target task (e.g. model pruning or model configuration).

Regards,