Search arXiv⌕ Search
Home/Archive/arXiv

arXiv · 2610.37132

Robust PPG-Exclusive Heart Rate Estimation During Maximal Exercise Using Population-Based Priors

Elisabetta De Giovanni·Iñigo Urteaga·Matteo Croci·Tomas Teijeiro
View original source ↗Read source document ↗Export metadata ↓

Abstract

PPG is the main technology for wearable heart rate (HR) monitoring. However, motion artifacts during intense physical activity reduce its reliability due to spectral overlap with the cardiac signal. This study investigates whether severely contaminated PPG signals can yield reliable HR estimates without auxiliary sensors by leveraging prior knowledge of HR dynamics. We acquired simultaneous PPG and ECG signals from 21 subjects during a maximal cycloergometer test. Using Leave-One-Subject-Out cross-validation, we built HR evolution priors from the ground-truth ECG and applied Dynamic Time Warping (DTW) to align normalized HR curves and estimate a barycenter average. DTW aligns normalized HR curves with minimal error across all subjects, validating robust population-based priors. For each test subject, we combined the PPG Power Spectral Density (PSD) with the prior distribution to generate a "posterior PSD". We identified and ranked candidate HR trajectories using heuristic search within the "posterior PSD" and selected the most likely path. Our approach reduced mean absolute error (MAE) in BPM relative to the ground-truth ECG by a median of 63.3% (IQR: 47.3%-74.6%) compared to a PPG+accelerometer benchmark. These results show that physiological priors and DTW-based alignment enable robust, PPG-exclusive HR monitoring during high-intensity exercise.

Explore related subjects

eess.SP

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Elisabetta De Giovanni, Iñigo Urteaga, Matteo Croci, Tomas Teijeiro. 2026-10-27. Robust PPG-Exclusive Heart Rate Estimation During Maximal Exercise Using Population-Based Priors. https://arxiv.org/abs/2610.37132

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Efficient Localization of Directional Emitters via Joint Beampattern Estimation

Localization methods designed for omnidirectional emitters degrade on directional sources, where directive beampatterns induce pronounced received signal strength (RSS) modulations. We present a direct position determination (DPD) approach that augments standard angle of arrival (AOA) and time difference of arrival (TDOA) measurements with RSS modulation to jointly estimate emitter position and beampattern parameters. Two algorithms reduce the four-dimensional search to two dimensions: a block coordinate ascent (BCA) method that alternates between position and beampattern subproblems, and a profile likelihood method that concentrates the beampattern parameters out via a generalized eigenvalue solve, a weighted least squares fit, and Newton refinement on the constrained Rayleigh quotient. We derive the Cramér--Rao lower bound (CRLB) and benchmark against an AOA-TDOA baseline and a prior-art approach that assumes known beamwidth and estimates only the orientation. The profile likelihood method approaches the CRLB, substantially outperforms all baselines at low signal-to-noise ratio (SNR), and remains robust under model mismatch with the assumed von Mises template. A sensitivity analysis shows performance peaks when receivers lie on the main-lobe edges where RSS gradients are maximized.

eess.SP↗
arXiv2026

In-Context Learning for MIMO Turbo Receivers

Multiple-input multiple-output (MIMO) turbo receivers iteratively exchange information between a soft-input soft-output detector and a decoder. Traditional implementations rely on pilot-based channel estimates, which can be poor in the case of short pilot sequences or mismatched channel models, resulting in significant performance degradation. To address this limitation, we introduce the EXTrinsic In-Context Learning Network (EXT-ICLNet), a Transformer-based detector that takes the pilots as in-context examples and is trained to approximate the symbol extrinsic posterior distribution given the pilots, without explicit channel estimation. EXT-ICLNet is thus compatible with existing turbo receivers, as it implements a log-domain prior skip connection, ensuring that the network learns only the extrinsic information. Numerical results for a $4\times4$ MIMO system with 16-QAM and a quantized front end show that EXT-ICLNet outperforms baselines based on explicit channel state information (CSI) estimation at low-to-medium signal-to-noise ratios (SNRs), as well as expectation-propagation (EP) baselines with perfect CSI at high SNRs. The results are further validated by comparing the extrinsic information transfer (EXIT) charts of EXT-ICLNet, a neural EP baseline, and EXT-ICLNet without skip connection, demonstrating how EXT-ICLNet provides a more favorable transfer of a priori and extrinsic information.

eess.SP↗
arXiv2026

FRAME-C: A knowledge-augmented deep learning pipeline for classifying multi-electrode array electrophysiological signals

Amyotrophic lateral sclerosis (ALS) is a fatal neurodegenerative disorder characterized by motor neuron degeneration, with changes to neuronal excitability being an early phenotype for ALS identified in patients and animal models. Recent advancements in induced pluripotent stem cell (iPSC) technology have enabled the generation of human iPSC-derived neuronal cultures, which, when combined with multi-electrode array (MEA) electrophysiology, provide rich spatial and temporal electrophysiological data. Traditionally, MEA data is analyzed using handcrafted features based on potentially imperfect domain knowledge, which while useful may not fully capture all the relevant characteristics inherent in the MEA data. Machine learning, in particular deep learning has the potential to automatically learn relevant characteristics (features) from raw data, without solely relying on handcrafted feature extraction. However, handcrafted features remain critical for encoding domain knowledge and improving model interpretability, especially in scenarios with limited or noisy data, as is often the case in most experimental studies. This study introduces FRAME-C, a knowledge-augmented machine learning pipeline that combines domain knowledge, raw spike waveform data, and deep learning techniques to classify MEA signals and identify ALS-specific phenotypes. FRAME-C leverages deep learning to extract important features from spike waveforms, while also incorporating handcrafted features such as spike amplitude, inter-spike interval, and spike duration, thus preserving key spatial and temporal information. We validated FRAME-C on both simulated and real-world MEA data from human iPSC-derived neuronal cultures, demonstrating its superior performance compared to existing methods for MEA classification.

eess.SP↗