A Two-Level Detection and Recovery Recommendation System for Depression.

Abstract

Depression is an emerging problem in public health. Various socio-demographic factors like age, sex, earning status, living spouse and family type, etc. are responsible for depression among people. Some co-morbid conditions like visual problems, hearing difficulties, and mobility problems also influence the disease. But depression can be diagnosed at the earliest using predictive modeling with various influencing input variables. The paper focuses on developing a system that can identify depression on multiple levels and provide a model for recovery. Waikato Environment for Knowledge Analysis (WEKA), a data mining tool was used for prediction based on machine learning classifiers. Also, five machine learning classifiers were compared with respect to three test options where extracted facial numerical attributes were iterated twice using a Multilayer Perceptron (MLP) classifier, with an accuracy of 82% and 89%, and precision of 87.2% and 89.3%. The system has the potential to lower depression statistics by employing a more efficient detection algorithm that takes into account both visual and text inputs including a well-designed rehabilitation model.