MENTAL STRESS CLASSIFICATION USING IMPORVED RANDOM FOREST WITH THRESHOLD FEATURE EXTRACTION ON REAL TIME DATASET

Abstract

ABSTRACT: Mental stress is big issue today’s society and everyone right from kids to adults are facing this problemthat causes a variety of chronic illnesses such as depression, malignancy, and cardiovascular disease(CVD). As a result, it is critical to handle and track a person's stress on a regular basis.We present a Random Forest (RF) method that can reliably predict mental stress levels by applying an enhanced band pass filtration technique to remove noise and exact features of R-Speak from Electrocardiogram (ECG) data in this work.When a person is stressed, a shift takes place in the ECG, which is an electrocardiogram (ECG) signal. Stress signals may be classified by analyzing ECG data and extracting particular characteristics.The suggested model obtained 98.0% accuracy based on the performance assessment criteria of the stress categorization model, confusion matrices, and precision-recall, which is a 14.7% improvement over prior research results. The suggested model was additionally validated using real-time datasets, and the model's accuracy was 98.0%.We categorised the model's output into three categories: normal, stress level 1, and stress level 2. As a result, our methodology can assist persons with stress manage their mental health.