Article: Detection from audio and text data features using modified score level fusion and improved deep convolutional neural network Journal: International Journal of Advanced Mechatronic Systems (IJAMECHS) 2026 Vol.13 No.2 pp.104 - 121 Abstract: Depression is a serious illness that requires prompt treatment to avoid negative impacts on a person's quality of life and general health. It is characterised by enduring feelings of melancholy and pessimism. Treatment options typically encompass psychotherapy, medication, or a combination designed for individual needs. Recognising the importance of early detection, this research introduces a depression detection model based on modified score level fusion-improved deep convolutional neural network, that utilises audio and text data features. The research methodology follows a systematic approach involving pre-processing, feature extraction, and depression detection process. Audio and text inputs undergo independent pre-processing and feature extraction using specialised techniques. The resulting features are then fed into a hybrid detection model, employing two IDCNN classifiers. The outcomes of IDCNN are obtained using an MSLF procedure which enhances the precision of depression detection. To validate the proposed MSLF-IDCNN model, comprehensive analyses, including simulation and experimental assessments are conducted. Inderscience Publishers - linking academia, business and industry through research

Title: Detection from audio and text data features using modified score level fusion and improved deep convolutional neural network

Authors: Janaswami Hymavathi; Chokka Anuradha

Addresses: Department of Computer science and Engineering, Koneru Lakshmaiah Education Foundation, Vaddeswaram, Guntur, Andhra Pradesh – 522302, India ' Department of Computer science and Engineering, Koneru Lakshmaiah Education Foundation, Vaddeswaram, Guntur, Andhra Pradesh – 522302, India

Abstract: Depression is a serious illness that requires prompt treatment to avoid negative impacts on a person's quality of life and general health. It is characterised by enduring feelings of melancholy and pessimism. Treatment options typically encompass psychotherapy, medication, or a combination designed for individual needs. Recognising the importance of early detection, this research introduces a depression detection model based on modified score level fusion-improved deep convolutional neural network, that utilises audio and text data features. The research methodology follows a systematic approach involving pre-processing, feature extraction, and depression detection process. Audio and text inputs undergo independent pre-processing and feature extraction using specialised techniques. The resulting features are then fed into a hybrid detection model, employing two IDCNN classifiers. The outcomes of IDCNN are obtained using an MSLF procedure which enhances the precision of depression detection. To validate the proposed MSLF-IDCNN model, comprehensive analyses, including simulation and experimental assessments are conducted.

Keywords: depression detection; bidirectional encoder representations from transformers; improved aspect term extraction; feature extraction; modified score level fusion.

DOI: 10.1504/IJAMECHS.2026.153247

International Journal of Advanced Mechatronic Systems, 2026 Vol.13 No.2, pp.104 - 121

Received: 07 Dec 2024
Accepted: 17 Mar 2025
Published online: 29 Apr 2026 *

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

Title: Detection from audio and text data features using modified score level fusion and improved deep convolutional neural network

Keep up-to-date