What question did this study set out to answer?

This application aims to early detect mental health disorders using various data inputs for better self-monitoring.

May 12, 2026Open Access

AI Health application

Key Points

This application aims to early detect mental health disorders using various data inputs for better self-monitoring.
Developed a multi-modal system using voice recordings, text analysis, and behavioral data.
Employed CNN for audio analysis, LSTM for behavioral patterns, and Transformers for text sentiment.
Utilized a cloud-native architecture for scalability and a secure handling of user data.
The application categorizes risk into four bands: Low, Moderate, High, and Critical.
Provides personalized recommendations based on the user's risk profile.
Automatically directs critical cases to mental health resources.

Abstract

AI-Powered Mental Health Diagnosis Application** is a multi-modal intelligent system designed to detect early signs of mental health disorders, including depression, anxiety, and stress, through the automated analysis of voice recordings, textual input, and behavioral data. The system leverages a combination of Convolutional Neural Networks (CNN) for acoustic feature extraction from speech, Long Short-Term Memory (LSTM) networks for longitudinal behavioral pattern recognition, and Transformer-based models for natural language processing and sentiment analysis of user-written text. The application targets adult users seeking an accessible, stigma-free tool for continuous mental health self-monitoring. Upon providing voice samples, journal entries, and passive behavioral data, the system processes each modality through a dedicated AI pipeline and synthesizes the results using a weighted fusion engine that generates a composite Risk Index, categorized into four bands: Low, Moderate, High, and Critical. Based on this risk profile, the system delivers personalized, evidence-based recommendations and, in critical cases, automatically signposts users to professional mental health resources. The system is built on a layered cloud-native architecture comprising a cross-platform mobile and web frontend, a RESTful backend API, a containerized AI inference layer, and a secure health data storage component. End-to-end encryption, GDPR-compliant data handling, and a privacy-first design ensure that sensitive user health data is protected at every stage. The application is designed for scalability, supporting large concurrent user bases through auto-scaling microservices, and for usability, adhering to Material Design guidelines to minimize interaction barriers for non-specialist users This work was conducted at Arab International University (AIU), Syria. The official website of the university is: https://www.aiu.edu.sy

Read Full Paperexternally

Mark Helpful

Bookmark

Relay

View Full Paper