Scheme 1.

Whole analysis pipeline of the machine learning model of the MALDI-TOF spectrum. After collection of MALDI-TOF raw spectrum data, data preprocessing (quality control, smoothing, baseline correction, intensity calibration, and peak detection) was performed. Peak adjustment and merging process (spectra alignment, spectra/peak binning) were performed, and the feature matrix was constructed. The machine learning algorithm was trained using the training data set; tuning and evaluation with the validation data was conducted to complete the model. The final performance of the machine learning model was evaluated using the test dataset.