The effect of training data size on real-time respiration prediction using long short-term memory model.

Aim: To investigate the optimal training dataset size (TDS) for respiration prediction accuracy using a long short-term memory (LSTM) model.

Methods: The respiratory signals of 151 patients acquired with the real-time position management system were retrospectively included in this study. Among the dataset, 101 respiratory signals were utilized to evaluate the impact of the TDS on prediction accuracy, while the remaining 50 signals were employed for setting the default hyperparameters. The prediction accuracy of the LSTM model using eight different TDSs (10 s, 20 s, 30 s, 60 s, 90 s, 110 s, 130 s, and 150 s) was examined and evaluated by the root mean square error (RMSE) between the real and predicted respiratory signals. The interplay effects of the main hyperparameters, the ahead time and the different testing data lengths using different TDSs were also measured.

Results: For the 520 ms ahead time, the root mean square error values of the LSTM model using the eight different training data sizes listed above were 0.146 cm, 0.137 cm, 0.134 cm, 0.125 cm, 0.120 cm, 0.121 cm, 0.121 cm, and 0.119 cm, respectively. The LSTM model achieved the highest prediction accuracy when the TDS was 150 s. The prediction accuracy was stable when the TDS exceeded 90 s.

Conclusions: TDS selection could influence the respiration signal prediction accuracy of the LSTM model. The relationship between TDS and the prediction accuracy of the LSTM model was not linear. The 90 s seemed to be an optimal TDS for the respiration signal prediction tasks using the LSTM model, as it was the shortest time at which a favorable prediction accuracy was maintained in this study.