The oscillations of the human heart rate are inherently complex and non-linear-they are best described by mathematical chaos, and they present a challenge when applied to the practical domain of cardiovascular health monitoring in everyday life. In this work, we study the non-linear chaotic behavior of heart rate through mutual information and introduce a novel approach for enhancing heart rate estimation in real-life conditions. Our proposed approach not only explains and handles the non-linear temporal complexity from a mathematical perspective but also improves the deep learning solutions when combined with them. We validate our proposed method on four established datasets from real-life scenarios and compare its performance with existing algorithms thoroughly with extensive ablation experiments. Our results demonstrate a substantial improvement, up to 40%, of the proposed approach in estimating heart rate compared to traditional methods and existing machine-learning techniques while reducing the reliance on multiple sensing modalities and eliminating the need for post-processing steps.

Deep learning models lack shift invariance, making them sensitive to input shifts that cause changes in output. While recent techniques seek to address this for images, our findings show that these approaches fail to provide shift-invariance in time series, where the data generation mechanism is more challenging due to the interaction of low and high frequencies. Worse, they also decrease performance across several tasks. In this paper, we propose a novel differentiable bijective function that maps samples from their high-dimensional data manifold to another manifold of the same dimension, without any dimensional reduction. Our approach guarantees that samples---when subjected to random shifts---are mapped to a unique point in the manifold while preserving all task-relevant information without loss. We theoretically and empirically demonstrate that the proposed transformation guarantees shift-invariance in deep learning models without imposing any limits to the shift. Our experiments on six time series tasks with state-of-the-art methods show that our approach consistently improves the performance while enabling models to achieve complete shift-invariance without modifying or imposing restrictions on the model's topology. The source code is available on GitHub.

Self-supervised learning (SSL) has emerged as a powerful paradigm for learning representations without labeled data. Most SSL approaches rely on strong, well-established, handcrafted data augmentations to generate diverse views for representation learning. However, designing such augmentations requires domain-specific knowledge and implicitly imposes representational invariances on the model, which can limit generalization. In this work, we propose an unsupervised representation learning method that replaces augmentations by generating views using orthonormal bases and overcomplete frames. We show that embeddings learned from orthonormal and overcomplete spaces reside on distinct manifolds, shaped by the geometric biases introduced by representing samples in different spaces. By jointly leveraging the complementary geometry of these distinct manifolds, our approach achieves superior performance without artificially increasing data diversity through strong augmentations. We demonstrate the effectiveness of our method on nine datasets across five temporal sequence tasks, where signal-specific characteristics make data augmentations particularly challenging. Without relying on augmentation-induced diversity, our method achieves performance gains of up to 15--20\% over existing self-supervised approaches. Source code: \url{https://github.com/eth-siplab/Learning-with-FrameProjections}

Today's fitness bands and smartwatches typically track heart rates (HR) using optical sensors. Large behavioral studies such as the U.K. Biobank use activity trackers without such optical sensors and thus lack HR data, which could reveal valuable health trends for the wider population. In this paper, we present the first dataset of wrist-worn accelerometer recordings and electrocardiogram references in uncontrolled at-home settings to investigate the recent promise of IMU-only HR estimation via ballistocardiograms. Our recordings are from 42 patients during the night, totaling 310 hours. We also introduce a frequency-based method to extract HR via curve tracing from IMU recordings while rejecting motion artifacts. Using our dataset, we analyze existing baselines and show that our method achieves a mean absolute error of 0.88 bpm-76% better than previous approaches and the first to surpass established medical standards for heart rate monitors. Our results validate the potential of IMU-only HR estimation as a key indicator of cardiac activity in existing longitudinal studies to discover novel health insights.

Virtual reality (VR) presents immersive opportunities across many applications, yet the inherent risk of developing cybersickness during interaction can severely reduce enjoyment and platform adoption. Cybersickness is marked by symptoms such as dizziness and nausea, which previous work primarily assessed via subjective post-immersion questionnaires and motion-restricted controlled setups. In this paper, we investigate the dynamic nature of cybersickness while users experience and freely interact in VR. We propose a novel method to continuously identify and quantitatively gauge cybersickness levels from users' passively monitored electroencephalography (EEG) and head motion signals. Our method estimates multitaper spectrums from EEG, integrating specialized EEG processing techniques to counter motion artifacts, and, thus, tracks cybersickness levels in real-time. Unlike previous approaches, our method requires no user-specific calibration or personalization for detecting cybersickness. Our work addresses the considerable challenge of reproducibility and subjectivity in cybersickness research. In addition to our method's implementation, we release our dataset of 16 participants and approximately 2 hours of total recordings to spur future work in this domain. Source code: https://github.com/eth-siplab/EEG_Cybersickness_Estimation_VR-Beyond_Subjectivity.

Multi-channel photoplethysmography (PPG) sensors have found widespread adoption in wearable devices for monitoring cardiac health. Channels thereby serve different functions-whereas green is commonly used for metrics such as heart rate and heart rate variability, red and infrared are commonly used for pulse oximetry. In this paper, we introduce a novel method that simultaneously fuses multi-channel PPG signals into a single recovered PPG signal that can be input to further processing. Via signal fusion, our learning-based method compensates for the artifacts that affect wavelengths to different extents, such as motion and ambient light changes. We evaluate our method on a novel dataset of multi-channel PPG recordings and electrocardiogram recordings for reference from 10 participants over the course of 13 hours during real-world activities outside the laboratory. Using the fusion PPG signal our method recovered, participants' heart rates can be calculated with a mean error of 4.5 bpm (23% lower than from green PPG signals at 5.9 bpm).

Detection of periodic patterns of interest within noisy time series data plays a critical role in various tasks, spanning from health monitoring to behavior analysis. Existing learning techniques often rely on labels or clean versions of signals for detecting the periodicity, and those employing self-supervised methods are required to apply proper augmentations, which is already challenging for time series and can result in collapse—all representations collapse to a single point due to strong augmentation. In this work, we propose a novel method to detect the periodicity in time series without the need for any labels or requiring tailored positive or negative data generation mechanisms. We mitigate the collapse issue by ensuring the learned representations retain information from the original samples without imposing any variance constraints on the batch. Our experiments in three time-series tasks against state-of-the-art learning methods show that the proposed approach consistently outperforms prior works, achieving performance improvements of more than 45--50%, showing its effectiveness.

Massage therapy is helpful for the rehabilitation of various diseases, such as headaches caused by migraines and stress. Existing robotic systems have focused on massage therapy on the torso and limbs, but performing massage motions through suitable actuation on a person’s head has been a challenge. In this paper, we present MiBOT, a head-worn massage robot that actuates two soft tactors to produce touch motions mimicking human massage. A key design principle behind MiBOT is its silent actuation, which we achieve through pneumatic artificial muscles in conjunction with a controller loop to respond to contact pressure. We evaluated the effectiveness of MiBOT in a controlled study and assessed subjects’ blood pressure and heart rate levels while applying MiBOT. We found that our mechanical system generated positive and conclusive quantitative outcomes that are similar to the human-administered massage, decreasing participants’ mean systolic and diastolic blood pressure by 2.8 mmHg and 1.7 mmHg, respectively, as well as calming their heart rate by 8–10% on average.

Reflective photoplethysmography (PPG) has become the default sensing technique in wearable devices to monitor cardiac activity via a person’s heart rate (HR). However, PPG-based HR estimates can be substantially impacted by factors such as the wearer’s activities, sensor placement and resulting motion artifacts, as well as environmental characteristics such as temperature and ambient light. These and other factors can significantly impact and decrease HR prediction reliability. In this paper, we show that state-of-the-art HR estimation methods struggle when processing representative data from everyday activities in outdoor environments, likely because they rely on existing datasets that captured controlled conditions. We introduce a novel multimodal dataset and benchmark results for continuous PPG recordings during outdoor activities from 16 participants over 13.5 hours, captured from four wearable sensors, each worn at a different location on the body, totaling 216 hours. Our recordings include accelerometer, temperature, and altitude data, as well as a synchronized Lead I-based electrocardiogram for ground-truth HR references. Participants completed a round trip from Zurich to Jungfraujoch, a tall mountain in Switzerland over the course of one day. The trip included outdoor and indoor activities such as walking, hiking, stair climbing, eating, drinking, and resting at various temperatures and altitudes (up to 3,571 m above sea level) as well as using cars, trains, cable cars, and lifts for transport—all of which impacted participants’ physiological dynamics. We also present a novel method that estimates HR values more robustly in such real-world scenarios than existing baselines.Dataset & code for HR estimation: https://siplab.org/projects/WildPPG

This paper presents a novel cancellation method of Intermodulation Distortions (IMDs) for earbud speakers used to measure Distortion Product Otoacoustic Emissions (DPOAE). Speakers’ non-linear behaviour is a significant problem for earbuds with small loudspeakers due to limitations in cone movement. Linear and non-linear speaker modelling enables us to inject exact distortion inverse to cancel what is introduced by speakers’ non-linearities. Our proposed method is compared against state-of-the-art related works in terms of harmonic reduction ratio. Simulation results and evaluation on real hardware show a 77% to 95% reduction in the harmonic distortions of a focused frequency region at the output of the loudspeaker, outperforming existing works by 6% to 14%.

The success of contrastive learning is well known to be dependent on data augmentation. Although the degree of data augmentations has been well controlled by utilizing pre-defined techniques in some domains like vision, time-series data augmentation is less explored and remains a challenging problem due to the complexity of the data generation mechanism, such as the intricate mechanism involved in the cardiovascular system. Moreover, there is no widely recognized and general time-series augmentation method that can be applied across different tasks. In this paper, we propose a novel data augmentation method for time-series tasks that aims to connect intra-class samples together, and thereby find order in the latent space. Our method builds upon the well-known data augmentation technique of mixup by incorporating a novel approach that accounts for the non-stationary nature of time-series data. Also, by controlling the degree of chaos created by data augmentation, our method leads to improved feature representations and performance on downstream tasks. We evaluate our proposed method on three time-series tasks, including heart rate estimation, human activity recognition, and cardiovascular disease detection. Extensive experiments against the state-of-the-art methods show that the proposed method outperforms prior works on optimal data generation and known data augmentation techniques in three tasks, reflecting the effectiveness of the presented method. The source code is available at double-blind policy.

We present a novel learning-based method that achieves state-of-the-art performance on several heart rate estimation benchmarks extracted from photoplethysmography signals (PPG). We consider the evolution of the heart rate in the context of a discrete-time stochastic process that we represent as a hidden Markov model. We derive a distribution over possible heart rate values for a given PPG signal window through a trained neural network. Using belief propagation, we incorporate the statistical distribution of heart rate changes to refine these estimates in a temporal context. From this, we obtain a quantized probability distribution over the range of possible heart rate values that captures a meaningful and well-calibrated estimate of the inherent predictive uncertainty. We show the robustness of our method on eight public datasets with three different cross-validation experiments.

This article presents a resource-efficient adaptive sampling methodology for classifying electrocardiogram (ECG) signals into different heart rhythms. We present our methodology in two folds: ( i ) the design of a novel real-time adaptive neural network architecture capable of classifying ECG signals with different sampling rates and ( ii ) a runtime implementation of sampling rate control using deep reinforcement learning (DRL). By using essential morphological details contained in the heartbeat waveform, the DRL agent can control the sampling rate and effectively reduce energy consumption at runtime. To evaluate our adaptive classifier, we use the MIT-BIH database and the recommendation of the AAMI to train the classifiers. The classifier is designed to recognize three major types of arrhythmias, which are supraventricular ectopic beats (SVEB), ventricular ectopic beats (VEB), and normal beats (N). The performance of the arrhythmia classification reaches an accuracy of 97.2% for SVEB and 97.6% for VEB beats. Moreover, the designed system is 7.3× more energy-efficient compared to the baseline architecture, where the adaptive sampling rate is not utilized. The proposed methodology can provide reliable and accurate real-time ECG signal analysis with performances comparable to state-of-the-art methods. Given its time-efficient, low-complexity, and low-memory-usage characteristics, the proposed methodology is also suitable for practical ECG applications, in our case for arrhythmia classification, using resource-constrained devices, especially wearable healthcare devices and implanted medical devices.

Providing health monitoring devices with machine intelligence is important for enabling automatic mobile healthcare applications. However, this brings additional challenges due to the resource scarcity of these devices. This work introduces a neural contextual bandits based dynamic sensor selection methodology for high-performance and resource-efficient body-area networks to realize next generation mobile health monitoring devices. The methodology utilizes contextual bandits to select the most informative sensor combinations during runtime and ignore redundant data for decreasing transmission and computing power in a body area network (BAN). The proposed method has been validated using one of the most common health monitoring applications: cardiac activity monitoring. Solutions from our proposed method are compared against those from related works in terms of classification performance and energy while considering the communication energy consumption. Our final solutions could reach 78.8% AU-PRC on the PTB-XL ECG dataset for cardiac abnormality detection while decreasing the overall energy consumption and computational energy by 3.7 × and 4.3 ×, respectively.

The recent developments in wearable devices and the Internet of Medical Things (IoMT) allow real-time monitoring and recording of electrocardiogram (ECG) signals. However, continuous monitoring of ECG signals is challenging in low-power wearable devices due to energy and memory constraints. Therefore, in this article, we present a novel and energy-efficient methodology for continuously monitoring the heart for low-power wearable devices. The proposed methodology is composed of three different layers: 1) a noise/artifact detection layer to grade the quality of the ECG signals; 2) a normal/abnormal beat classification layer to detect the anomalies in the ECG signals; and 3) an abnormal beat classification layer to detect diseases from ECG signals. Moreover, a distributed multioutput convolutional neural network (CNN) architecture is used to decrease the energy consumption and latency between the edge–fog/cloud. Our methodology reaches an accuracy of 99.2% on the well-known MIT-BIH Arrhythmia Data Set. Evaluation on real hardware shows that our methodology is suitable for devices having a minimum RAM of 32 kb. Moreover, the proposed methodology achieves$7\times $more energy efficiency compared to state-of-the-art works.

Human activity recognition (HAR) is one of the key applications of health monitoring that requires continuous use of wearable devices to track daily activities. This article proposes an adaptive convolutional neural network for energy-efficient HAR (AHAR) suitable for low-power edge devices. Unlike traditional adaptive (early-exit) architecture that makes the early-exit decision based on classification confidence, AHAR proposes a novel adaptive architecture that uses an output block predictor to select a portion of the baseline architecture to use during the inference phase. The experimental results show that traditional adaptive architecture suffer from performance loss whereas our adaptive architecture provides similar or better performance as the baseline one while being energy efficient. We validate our methodology in classifying locomotion activities from two data sets—1) Opportunity and 2) w-HAR. Compared to the fog/cloud computing approaches for the Opportunity data set, our baseline and adaptive architectures show a comparable weighted F1 score of 91.79%, and 91.57%, respectively. For the w-HAR data set, our baseline and adaptive architectures outperform the state-of-the-art works with a weighted F1 score of 97.55%, and 97.64%, respectively. Evaluation on real hardware shows that our baseline architecture is significantly energy efficient ($422.38\times $less) and memory-efficient ($14.29\times $less) compared to the works on the Opportunity data set. For the w-HAR data set, our baseline architecture requires$2.04\times $less energy and$2.18\times $less memory compared to the state-of-the-art work. Moreover, experimental results show that our adaptive architecture is 12.32% (Opportunity) and 11.14% (w-HAR) energy efficient than our baseline while providing similar (Opportunity) or better (w-HAR) performance with no significant memory overhead.

Edge-Cloud hierarchical systems employing intelligence through Deep Neural Networks (DNNs) endure the dilemma of workload distribution within them. Previous solutions proposed to distribute workloads at runtime according to the state of the surroundings, like the wireless conditions. However, such conditions are usually overlooked at design time. This paper addresses this issue for DNN architectural design by presenting a novel methodology, LENS, which administers multi-objective Neural Architecture Search (NAS) for two-tiered systems, where the performance objectives are refashioned to consider the wireless communication parameters. From our experimental search space, we demonstrate that LENS improves upon the traditional solution’s Pareto set by 76.47% and 75% with respect to the energy and latency metrics, respectively.