Novel metabolomics approach combined with machine learning for the diagnosis of influenza from nasopharyngeal specimens
To train models with cross validation and test on a held-out set of samples, run train_and_test.py. Usage: ./train_and_test.py -o <output_dir> -d <data_csv> -l <labels_csv>.
To validate models trained with retrospective data on an unseen prospective dataset, run prospective_train.py to produce the necessary model checkpoints, followed by prospective_evaluate.py to evaluate the models on the prospective set. The scripts automatically use the datasets in the data/ folder. Usage:
python prospective_train.py
python prospective_evaluate.py