Predict-Mental-Health-Crisis-using-Surveillance-Data-from-BRFSS-2021

This is a report which relates to a simple classification model of machine learning technique used in BRFSS 2021 data from CDC.

Methods

Dataset

The analysis is based on a public dataset of CDC, collected bya random annual phone-based survey which tracks health risk behaviors, chronic diseases, access to healthcare, and the use of preventive health services in the United States.

Mental Health Crisis is characterized by individuals who had current depression/anxiety, a lifetime diagnosis of depression, and/or a lifetime diagnosis of anxiety and the class atribute (’Mental crisis”) was compiled based on the answeres from the questionnaires.

all these analysis have been done usng R programming language.

Visualizations

Below is the comparison of the number of classes before and after applying SMOTE into the dataset

Below is the ROC curve, and it is reported that the AUROC and performance measures show 0.87 AUC from the decision tree (C50) model.