Title: A smart intelligent Internet of Things framework for predicting mental health

Authors: G. Sherlin Shobitha; V. Sudarshani Kataksham; T. Nagalaxmi; V. Spandana; G. Sreelatha; V. Radha

Addresses: Department of ECE, Stanley College of Engineering & Technology for Women, Hyderabad, Telangana 500001, India ' Department of ECE, Stanley College of Engineering & Technology for Women, Hyderabad, Telangana 500001, India ' Department of ECE, Stanley College of Engineering & Technology for Women, Hyderabad, Telangana 500001, India ' Department of ECE, Osmania University, Hyderabad, Telangana, India ' Department of IT, Stanley College of Engineering & Technology for Women, Hyderabad, Telangana 500001, India ' Department of ECE, Stanley College of Engineering & Technology for Women, Hyderabad, Telangana 500001, India

Abstract: A psychiatric disorder is a global concern affecting millions and burdening the healthcare system. Current diagnosis relies on subjective symptoms and isolated clinical examinations, leading to premature diagnoses and treatments that affect millions of lives. The paper introduces a new Fossa-based graph neural network (FbGNN) to enhance predictive accuracy for mental illness. The study collected mental health data from various sources, including Reddit, Twitter, and discussion forums, and processed it using the Fossa optimisation technique. The selected features were then used in a graph neural network (GNN) model to classify various mental health diseases The data was then preprocessed, noise removed, and the model was applied to further refine the classification process. The FbGNN model outperformed traditional machine learning models across key performance metrics, achieving 98.87% accuracy, 97.85% Precision, 98.60% recall, 98.22% F1 score, and a minimal error rate of 1.13%.

Keywords: mental health data; fossa optimisation; GNN; graph neural network; preprocessing; feature analysis; classification accuracy.

DOI: 10.1504/IJNVO.2025.151510

International Journal of Networking and Virtual Organisations, 2025 Vol.33 No.3, pp.251 - 278

Received: 04 Jul 2025
Accepted: 26 Nov 2025
Published online: 03 Feb 2026 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article