Need help on ECG classification with LSTM

I have dataset from physionet for detection of myocardial infarction but the accuracy of detection isn't so good. Who can help me about that . I am using keras deep learning framework