Ralf Seepold

Automatic sleep stage classification is essential for enabling non‐invasive, at‐home monitoring. However, current methods often rely on electroencephalogram (EEG) signals and ad‐hoc development approaches that limit reproducibility. We present a reproducible engineering framework for a deep learning model based on the U‐Net architecture that classifies sleep into five stages (Wake, N1, N2, N3 and REM) or four (Wake, Light Sleep, Deep Sleep and REM) using only three easily acquired physiological signals: oxygen saturation (SpO), heart rate (HR) and abdominal respiratory effort (AbdRes). In contrast to most previous studies, our model provides sleep stage predictions on a per‐second basis, thus overcoming the limitations associated with fixed 30‐s epochs. The model was trained on the Sleep Heart Health Study—Visit 2 (SHHS2) dataset and externally validated on the Multi‐Ethnic Study of Atherosclerosis (MESA). Optimisation of the model was achieved via Keras Tuner with the Hyperband algorithm. The study achieved weighted F1‐scores of 68% (five‐stage) and 71% (four‐stage) with Cohen's Kappa of 0.61 and 0.67 on SHHS2, with consistent performance on MESA. These results demonstrate strong generalisation and suggest that this lightweight, EEG‐free approach offers a practical path towards scalable, clinically relevant sleep monitoring.

Sleep deprivation is a significant contributor to various diseases, leading to poor cognitive function, decreased performance, and heart disorders. Insomnia, the most prevalent sleep disorder, requires more effective diagnosis and screening for proper treatment. Actigraphic data and its combination with physiological sensors like electroencephalogram (EEG), electrocardiogram (ECG), and body temperature have proven significant in predicting insomnia using machine learning methods. Studies focusing solely on actigraphic data achieved an accuracy of 84%, combining it with other wearable devices increased accuracy to 88%, and 2-channel EEG alone yielded an accuracy of 92%, but limits scalability and practicality in real-world settings. Here we show that using the hybrid approach of incorporating both recursive feature elimination (RFE) and principal component analysis (PCA) on sleep and heart data features yields outstanding results, with the multi-layer perception (MLP) achieving an accuracy of 95.83% and an F1 score of 0.93. The top-ranked features are predominantly sleep-related and time-domain RR interval. Our findings emphasize the importance of tailoring feature sets and employing appropriate reduction techniques for optimal predictive modeling in sleep-related studies. Our results demonstrate that the ensemble classifiers generalize well on the dataset regardless of the feature count, while other algorithms are hindered by the curse of dimensionality.

With the emergence of new sensor technologies, such as fiber optic sensors (FOSs), compared to traditional mechanical sensors, unobtrusive sleep monitoring has been a research focus for decades. This work aims to provide a guide to current bed-based sensor technologies with diverse applications in various settings. We conducted a retrospective literature review, summarizing the state-of-the-art research over the past decade on non-contact bed-based sensor technology in sleep monitoring. We developed a three-category terminology: unobtrusive sensor technology, application, and subject. A total of 263 unique articles were acquired from three databases and screened for relevance, resulting in 21 papers selected for in-depth analysis. The findings revealed eight types of sensors: six mechanical sensors (pressure, accelerometer, piezoelectric, load cell, electromechanical film (EMFI), and hydraulic) and two FOSs (fiber Bragg grating and microbend FOS) that are integrated with or positioned under the bed at three levels of unobtrusiveness. We identified 15 parameters, with heart rate (14) and respiratory rate (13) being the most frequently measured. These parameters are generally categorized into three applications: disease-related diagnosis (18), general sleep analysis (9), and general well-being (11). The results indicated that sleep apnea (5) and insomnia (2) were the most frequently detected sleep disorders. Additionally, 59.1% (13) of the systems were tested in a lab environment, with only one undergoing clinical trials. In summary, there is a clear lack of convincing proof of the systems’ effectiveness in continuous in-home sleep monitoring.

Battery power is crucial for wearable devices as it ensures continuous operation, which is critical for real-time health monitoring and emergency alerts. One solution for long-lasting monitoring is energy harvesting systems. Ensuring a consistent energy supply from variable sources for reliable device performance is a major challenge. Additionally, integrating energy harvesting components without compromising the wearability, comfort, and esthetic design of healthcare devices presents a significant bottleneck. Here, we show that with a meticulous design using small and highly efficient photovoltaic (PV) panels, compact thermoelectric (TEG) modules, and two ultra-low-power BQ25504 DC-DC boost converters, the battery life can increase from 9.31 h to over 18 h. The parallel connection of boost converters at two points of the output allows both energy sources to individually achieve maximum power point tracking (MPPT) during battery charging. We found that under specific conditions such as facing the sun for more than two hours, the device became self-powered. Our results demonstrate the long-term and stable performance of the sensor node with an efficiency of 96%. Given the high-power density of solar cells outdoors, a combination of PV and TEG energy can harvest energy quickly and sufficiently from sunlight and body heat. The small form factor of the harvesting system and the environmental conditions of particular occupations such as the oil and gas industry make it suitable for health monitoring wearables worn on the head, face, or wrist region, targeting outdoor workers.

Keywords: Energy harvesting, Hybrid, Medical Wearable Sensor Nodes, Health Monitoring, Power Management

Apnea is a sleep disorder characterized by breathing interruptions during sleep, impacting cardiorespiratory function and overall health. Traditional diagnostic methods, like polysomnography, are unobtrusive, leading to noninvasive monitoring. This study aims to develop and validate a novel sleep monitoring system using noninvasive sensor technology to estimate cardiorespiratory parameters and detect sleep apnea. We designed a seamless monitoring system integrating non-contact force-sensitive resistor sensors to collect ballistocardiogram signals associated with cardiorespiratory activity. We enhanced the sensor’s sensitivity and reduced the noise by designing a new concept of edge-measuring sensor using a hemisphere dome and mechanical hanger to distribute the force and mechanically amplify the micromovement caused by cardiac and respiration activities. In total, we deployed three edge-measuring sensors, two deployed under the thoracic and one under the abdominal regions. The system is supported with onboard signal preprocessing in multiple physical layers deployed under the mattress. We collected the data in four sleeping positions from 16 subjects and analyzed them using ensemble empirical mode decomposition to avoid frequency mixing. We also developed an adaptive thresholding method to identify sleep apnea. The error was reduced to 3.98 and 1.43 beats per minute in heart rate and respiration estimation, respectively. The apnea was detected with an accuracy of 87%. We optimized the system such that only one edge-measuring sensor can measure the cardiorespiratory parameters. Such a reduction in the complexity and simplification of the instruction of use shows excellent potential for in-home and continuous monitoring.

While driving, stress is caused by situations in which the driver estimates their ability to manage the driving demands as insufficient or loses the capability to handle the situation. This leads to increased numbers of driver mistakes and traffic violations. Additional stressing factors are time pressure, road conditions, or dislike for driving. Therefore, stress affects driver and road safety. Stress is classified into two categories depending on its duration and the effects on the body and psyche: short-term eustress and constantly present distress, which causes degenerative effects. In this work, we focus on distress. Wearable sensors are handy tools for collecting biosignals like heart rate, activity, etc. Easy installation and non-intrusive nature make them convenient for calculating stress. This study focuses on the investigation of stress and its implications. Specifically, the research conducts an analysis of stress within a select group of individuals from both Spain and Germany. The primary objective is to examine the influence of recognized psychological factors, including personality traits such as neuroticism, extroversion, psychoticism, stress and road safety. The estimation of stress levels was accomplished through the collection of physiological parameters (R-R intervals) using a Polar H10 chest strap. We observed that personality traits, such as extroversion, exhibited similar trends during relaxation, with an average heart rate 6% higher in Spain and 3% higher in Germany. However, while driving, introverts, on average, experienced more stress, with rates 4% and 1% lower than extroverts in Spain and Germany, respectively.