An Intelligent chatbot implementation for early detection and intervention for anorexia nervosa

Abstract

Anorexia nervosa (AN), an exhausting and potentially fatal eating disorder, has long been a significant public health concern. Characterized by extreme eating habits and an intense fear of gaining weight, this disorder may lead to other mental illnesses such as depression, obsessive compulsive disorder (OCD), borderline personality disorder (BPD) and sometimes self-harm. Notwithstanding its devastating consequences and prevalence amongst adolescence and youths especially women, its early detection and intervention remains challenging. This research presents a novel approach in addressing anorexia through utilization of random forest, a machine learning algorithm and natural language processing to create an intelligent chatbot for detection and provision of personalized intervention for anorexia patients. The chatbot is built based on RASA framework and it is deployed on Telegram, a social media platform where it can engage users in supportive dialogues to detect potential risk factors and deliver timely intervention for anorexia nervosa. The implications of this research underscore the value of machine learning in mental health detection and treatment. Besides provision of a toolkit that could be used by medical practitioners, it introduces an accessible means of reaching individuals who may not seek help through the conventional means. Additionally, it connects individuals to health care professionals and support networks enhancing early detection and reducing further complications. To assess the feasibility of the proposed concept, a functional chatbot prototype was developed using a Rapid Application Development (RAD) approach. The training and testing data were split n an 80/20 ratio, and the Telegram messaging platform was utilized for user interaction testing. While the study presents promising results, some of the limitations included constraints of limited data set and the need for ongoing refinement of the chatbot’s algorithm. There is also limited research regarding anorexia. Future studies could investigate refining the technology, expanding the dataset, and addressing ethical concerns around mental health with an aim to contribute to more effective and accessible mental health support.

Keywords: Anorexia, Chatbot, Machine Learning, Natural Language Processing, Eating Disorder, Sentimental Analysis