Vitamin deficiency predicted by machine-learning

This repository contains codes used to produce most of the machine-learning results of the following paper.

• Efficient Prediction of Vitamin B Deficiencies via Machine-learning Using Routine Blood Test Results in Patients With Intense Psychiatric Episode. Hidetaka Tamune*, Jumpei Ukita*, Yu Hamamoto, Hiroko Tanaka, Kenji Narushima and Naoki Yamamoto. Frontiers in Psychiatry, 2020. Paper Link. (*: co-first author)

Usage

1. Modify load_dataset in data_utils.py to read the dataset.
2. Optimize hyperparameters by running grid_search.py.
3. Calculate the prediction accuracy: e.g. python test.py --target folate --modelType rf
4. To estimate the sample size necessary for a high prediction accuracy, run sample_size.py