Authors: Yi Li*, Rebecca A Frederick*, Daniel George, Stuart F Cogan, Joseph J Pancrazio, Leonidas Bleris☨, and Ana G Hernandez-Reynoso☨
(*Co-First Authors, ☨Corresponding Authors) | Published 27 June 2024 | J. Neural Eng. 21 036054
https://doi.org/10.1088/1741-2552/ad593e
Supplementary_Scripts/MLP-Partial-10-16.pkl: Candidate predictive model using the partial feature set and Multilayer Perceptron.
Supplementary_Scripts/MLP-Total-5-26.pkl: Candidate predictive model using the semi-complete feature set and Multilayer Perceptron.
Supplementary_Scripts/RF-Partial-19.pkl: Candidate predictive model using the partial feature set and Random Forest.
Supplementary_Scripts/RF-Total-89.pkl: Candidate predictive model using the semi-complete feature set and Random Forest.
Supplementary_Scripts/RF-Partial-Level-8.pkl: Candidate predictive model using the partial feature set, Random Forest, and scaled outputs.
Supplementary_Scripts/data_boxplot.py: Python script for boxplot graphing for numerical features.
Supplementary_Scripts/feature_importance.py: Python script for calculating importance scores of features.
Supplementary_Scripts/pca_training.py: Python script for data visualization using PCA.
Supplementary_Scripts/tsne_training.py: Python script for data visualization using t-sne.
Supplementary_Scripts/umap_training.py: Python script for data visualization using UMAP.
Supplementary_Scripts/shannon_equation.py: Python script for modeling using original Shannon equation.
Supplementary_Scripts/knn_trianing.py: Python script for modeling using KNN algorithm.
Supplementary_Scripts/randomForest_trianing.py: Python script for modeling using Random Forest algorithm.
Supplementary_Scripts/logisticRegression.py: Python script for modeling using logistic regression algorithm.
Supplementary_Scripts/MLP_trianing.py: Python script for modeling using MLP algorithm.
Supplementary Data 1. Neuromodulation dataset for meta-analyses. 385 neuromodulation experimental results were compiled from 58 publications.
Supplementary Data 2. The training dataset. The training dataset contained 308 entries, of which 151 were with label 0, and 157 were with label 1.
Supplementary Data 3. The testing dataset. The testing dataset contained 77 entries, of which 44 were with label 0, and 33 were with label 1.
Supplementary Data 4. The predictive performances of the 14 candidate models which passed both filtering conditions for random forest and semi-complete feature set.
Supplementary Data 5. The predictive performances of the 46 candidate models which passed both filtering conditions for MLP and semi-complete feature set.
Supplementary Data 6. The predictive performances of the 46 candidate models which passed both filtering conditions for random forest and Partial feature set.
Supplementary Data 7. The predictive performances of the 9 candidate models which passed both filtering conditions for MLP and Partial feature set.
Supplementary Data 8. The predictive performances of candidate models using One-Hot encoding, random forest, Partial feature set, and non-binary labels (levels 0-4).
Supplementary Data 9. Database of new parameters for validation with binary labels and predictions.
Supplementary Scripts. The RF-Partial-19 model, and other main python scripts used in building and evaluating the predictive models in this study.