SCG With Your Phone: Diagnosis of Rhythmic Spectrum Disorders in Field Conditions

TL;DR

An enhanced U-Net v3 architecture is developed that integrates multi-scale convolutions, residual connections, and attention gates, enabling reliable segmentation of noisy SCG signals, and a novel adaptive 3D-to-1D projection method ensures robustness to arbitrary smartphone orientation.

Abstract

Aortic valve opening (AO) events are crucial for detecting frequency and rhythm disorders, especially in real-world settings where seismocardiography (SCG) signals collected via consumer smartphones are subject to noise, motion artifacts, and variability caused by device heterogeneity. In this work, we present a robust deep-learning framework for SCG segmentation and rhythm analysis using accelerometer recordings obtained with consumer smartphones. We develop an enhanced U-Net v3 architecture that integrates multi-scale convolutions, residual connections, and attention gates, enabling reliable segmentation of noisy SCG signals. A dedicated post-processing pipeline converts probability masks into precise AO timestamps, whereas a novel adaptive 3D-to-1D projection method ensures robustness to arbitrary smartphone orientation. Experimental results demonstrate that the proposed method achieves consistently high accuracy and robustness across various device types and unsupervised data-collection conditions. Our approach enables practical, low-cost, and automated cardiac-rhythm monitoring using everyday mobile devices, paving the way for scalable, field-deployable cardiovascular assessment and future multimodal diagnostic systems.

Figures (4)

• Figure 1: Overview of the dataset statistics: (a) distribution of mobile device vendors and (b) distribution of accelerometer sampling frequencies.
• Figure 2: Labeled SCG.
• Figure 3: Modified U-Net v3 Architecture.
• Figure 4: Interpretability analysis of the best-performing model (U-Net v3 + Dice+BCE) on smartphone SCG recordings.