AUTHOR=Mishra Sushruta , Tripathy Hrudaya Kumar , Kumar Thakkar Hiren , Garg Deepak , Kotecha Ketan , Pandya Sharnil 

TITLE=An Explainable Intelligence Driven Query Prioritization Using Balanced Decision Tree Approach for Multi-Level Psychological Disorders Assessment

JOURNAL=Frontiers in Public Health

VOLUME=Volume 9 - 2021

YEAR=2021

URL=https://www.frontiersin.org/journals/public-health/articles/10.3389/fpubh.2021.795007

DOI=10.3389/fpubh.2021.795007

ISSN=2296-2565

ABSTRACT=Emotions have a critical impact on human psychology as it affects their psychological health to a great level. Positive emotions relates to health improvement while negative emotions may aggravate various psychological disorders like anxiety, stress and depression. In recent times, these mental health concerns have become predominant throughout society. Diagnosing these mental health issues at an early stage is crucial. Though there exists different computational methods that successfully predicted psychological disorders but most of them lacks a concrete explanation and reasoning behind such predictions. It leads to a black-box view of uncertainty. This research involves an in-depth assessment of such mental risks which further leads to development of a novel intelligent predictive model for multi class psychological risk recognition well equipped with an accurate explainable interface. Online standard questionnaires are utilized as dataset. Q-Prioritization is a new approach used to drop insignificant questions from dataset. Repetitive oversampling enabled balanced decision tree is a novel technique applied for training and testing of the model. Predictive nature along with its contributing factors are explained and interpreted with three techniques which include Permuted Feature Importance method, Contrastive Explanation method and Counterfactual method which together forms a reasoning engine. Prediction outcome generated an impressive performance with an aggregated accuracy of 98.25\%. The mean precision, recall and f-score metric recorded were 0.98, 0.977 and 0.979 respectively. Also it was noted that without applying Q-Prioritization, the accuracy significantly drops to 90.25\%. The error rate observed with our model was only 0.026. Most importantly, the reasoning engine helped in more precise and transparent interpretation of predicted results by not only generating priority ranked symptoms but also highlighting critical contributing factors for a certain psychological risk prediction. Also it recommends possible risk regulatory solutions to patients. Hence the proposed multi-level psychological disorder predictive model can successfully serve as an assistive deployment for medical experts in effective treatment of mental health concerns supported by an in-built decision explanation functionality.