This file contains a brief description of the source codes and data files that appear in this folder and were employed to obtain the results reported in the article "Assessing the risk of falling in community-dwelling older adults through cognitive domains and Machine Learning techniques".

------------------data files-----------------------
1)The file Planilla_pruebas contains the database used in this study, which is composed of six variables and 50 samples.
2) The files models0_v3, models2_v3, and models02_v3 each contain 100 LR model instances obtained through bootstrapping sampling. models0_v3, models2_v3, and models02_v3 correspond to LR with Edu input, LR with TMT input, and LR with Edu and TMT input, respectively.
3) The files matrices_conf_test0_v3, matrices_conf_test2_v3, and matrices_conf_test02_v3 contain 100 confusion matrices measuring the performance of the previous model instances.   

------------------Python code files----------------------- 

0 MBT_histograms: Obtain histograms to describe the variables involved in the study (Section 4.1.2).
1 FDR: Obtain FDR values to describe the individual discriminative power of the input variables (Section 4.1.2). 
2 MBT_correlations: Obtain the correlation matrix between the input variables to describe the variables involved in the study (Section 4.1.2).
3 MBT_clasif_superv: Train and assess three ML models. For each model, we assess different sets of input variables and parameter values. Model assessment is carried out through an aggregated confusion matrix and typical metrics. The models that perform well are saved using the pickle library (Sections 4.2 and 4.3).
4 Wald_test: Perform the Wald test to determine if some input variables can be dropped from the models. The models were obtained during the training/assessing stage (Section 4.3).


