Comparative Analysis of Spiking Neural Networks and SNN-Diffusion Hybrid Models for Voice-Based Heart Failure Detection
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
Heart failure alters vocal characteristics due to cardiopulmonary fatigue, leading to measurable acoustic deviations.This study compares a Spiking Neural Network (SNN) and a hybrid SNN-Diffusion model for early heart failure detection from speech recordings.The SNN captures temporal-spike dynamics via leaky integrate-andfire neurons, while the diffusion model enhances latent feature denoising and generalization.Using normalized Mel-spectrograms and MFCC encodings, both architectures were evaluated on a curated heart-failure speech dataset.The hybrid SNN-Diffusion achieved higher accuracy (95.2%), precision (94.8%), recall (96.1%), and F1-score (95.4%) than the standalone SNN (89.6%).The results confirm that integrating bio-plausible spiking computation with diffusion-based generative modelling enhances robustness to noise and data scarcity, enabling more reliable lowenergy biomedical audio analytics.
Fields of Study & Subjects
Phonocardiography and Auscultation Techniques
Artificial Intelligence in Healthcare
Voice and Speech Disorders
Publication details
- Journal / source
- Advances in intelligent systems research/Advances in Intelligent Systems Research
- Pages
- 585–599
- Publication date
- 1 April 2026
- Language
- EN
- Document type
- Book chapter