Random Forest Regressor Machine Learning Model Developed for Mental Health Prediction Based on Mhi-5, Phq-9 and Bdi Scale

Mental health is the new hot area of study, research and experimentation in the current decade scenario from both medical science perspective and Technology domain. As there is observed a sudden spike and familiarity in Mental health related cases and issues, this invites a variety of research in the field. There exist multiple procedures to diagnose mental disorders but no one method has gained momentum to diagnose and treat mental health issues with remarkable accuracy. Since the symptoms and factors influencing mental health problems vary a lot, human tendency seems to be less compactible and hence there exists need to introduce Artificial intelligence-based systems to assist psychiatrists and monitor on micro level basis. The research focuses Mental Health Data collected through online forms consisting of 3 Questionnaires(MHI-5,BDI,PHQ-9) consisting of 26 questions about various factors influencing mental disorders, Each Questionnaire is used to train an individual model using random forest regressor, random forest classifiers followed by Hyper parameter Optimization techniques namely Grid Search CV, Randomized Search CV, Bayesian Optimization -Automate Hyperparameter Tuning (Hyperopt),Optuna- Automate Hyperparameter Tuning, Genetic Algorithms (TPOT) Classifier - for the purpose to optimize the model result. MHI-5 based model gives accuracy of 95.83% using Bayesian Optimization, whereas PHQ-9 with 82.61% using Optuna and BDI model with 83.33 using Bayesian Optimization, Randomize Search Cv, Grid Search Cv each.