The work described in this thesis is based on the results of a clinical trial conducted by the research team at the Medical Informatics Unit of the University of Cambridge, which show that the Ballistocardiogram (BCG) has prognostic value in detecting impaired left ventricular function before it becomes clinically overt as myocardial infarction leading to sudden death. The objective of this study is to develop and demonstrate a framework for realising an on-line BCG signal classification model in a portable device that would have the potential to find pathological signs as early as possible for home health care. Two new on-line automatic BeG classification models for time domain BeG classification are proposed. Both systems are based on a two stage process: input feature extraction followed by a neural classifier. One system uses a principal component analysis neural network, and the other a discrete wavelet transform, to reduce the input dimensionality. Results of the classification, dimensionality reduction, and comparison are presented. It is indicated that the combined wavelet transform and MLP system has a more reliable performance than the combined neural networks system, in situations where the data available to determine the network parameters is limited. Moreover, the wavelet transfonn requires no prior knowledge of the statistical distribution of data samples and the computation complexity and training time are reduced. Overall, a methodology for realising an automatic BeG classification system for a portable instrument is presented. A fully paralJel neural network design for a low cost platform using field programmable gate arrays (Xilinx's XC4000 series) is explored. This addresses the potential speed requirements in the biomedical signal processing field. It also demonstrates a flexible hardware design approach so that an instrument's parameters can be updated as data expands with time. To reduce the hardware design complexity and to increase the system performance, a hybrid learning algorithm using random optimisation and the backpropagation rule is developed to achieve an efficient weight update mechanism in low weight precision learning. The simulation results show that the hybrid learning algorithm is effective in solving the network paralysis problem and the convergence is much faster than by the standard backpropagation rule. The hidden and output layer nodes have been mapped on Xilinx FPGAs with automatic placement and routing tools. The static time analysis results suggests that the proposed network implementation could generate 2.7 billion connections per second performance.

This work demonstrates the viability of the ballistocardiogram (BCG) signal derived from a head-worn device as a biometric modality for authentication. The BCG signal is the measure of an individual's body acceleration as a result of the heart's ejection of blood. It is a characterization of an individual's cardiac cycle and can be derived non-invasively from the measurement of subtle movements of a person's extremities. Through the use of accelerometer and gyroscope sensors on a Smart Eyewear (SEW) device, derived BCG signals are used to train a convolutional neural network (CNN) as an authentication model, which is personalized for each wearer. This system is evaluated using data from 12 subjects, showing that this approach has an equal error rate of 3.5% immediately after training, and only marginally degrades to 13% after about 2 months, in the worst case. We also explore the use of our authentication approach for individuals with severe motor disabilities, and observe that the results fall only slightly short of those of the larger population, with immediate EER values at 11.2% before rising to 21.6%, again in the worst case.. Overall, we demonstrate that this model presents a longitudinally-viable authentication solution for passive biometric authentication.

No description available.

The population of the world is aging. In Sweden alone, almost 20% of the population is 65 years or older. As people get older, problems with sleep disturbances and sleep quality tends to increase, as do the risks of falling injuries. In this thesis, methods for calculating sleep quality and if a person is about to leave a bed were devised. A bed sensor, measuring ballistocardiographical signals, was used to measure activity in bed and vital signs of the occupant. The Cole-Kripke algorithm, used to calculate sleep quality based on activity from a wrist worn sensor, was adapted to the bed sensor system and compared to results from the ActiGraph wGT3X-BT activity monitor, which is frequently used in research. The bed sensor systems sleep quality estimations showed strong correlation with the ActiGraph, with a Pearson correlation coefficient of 0.946. Two approaches were made to estimate if a subject was about to leave the bed, one by training a neural network on labeled night data, and one using a linear equation with each term consisting of activity data, optimized by linear regression. The neural network approach suffered from limited data, but the linear method showed more promise, with accuracy, specificity and sensitivity all over 70%.

Master of Science / Department of Electrical and Computer Engineering / Steven Warren / Polysomnography is a method to obtain physiological signals from individuals with potential sleep disorders. Such physiological data, when acquired from children with autism spectrum disorders, could allow caregivers and child psychologists to identify sleep disorders and other indicators of nighttime well-being that affect their quality of life and ability to learn. Unfortunately, traditional polysomnography is not well suited for children with autism spectrum disorder because they commonly have an aversion to unfamiliar objects – in this case, the numerous wires and electrodes required to perform a full polysomnograph. Therefore, an innovative, unobtrusive method for gathering relevant physiological data must be designed. This report discusses several methods for obtaining a ballistocardiogram (BCG), which is a representation of the ballistic forces created by the heart during the cardiac cycle. A ballistocardiograph design is implemented using an electromechanical film placed under the center of a bed sheet. While an individual sleeps on the bed, the circuitry attached to the film extract and amplify the BCG data, which are then streamed to a computer through a LabVIEW interface and stored in a text file. These data are analyzed with a MATLAB algorithm which uses autocorrelation and linear predictive coding in the time domain to sharpen the signal. Frequency-domain peaks are then extracted to determine average heart rate every ten seconds. Initial tests involved four participants (student members of the research team) who laid in four positions: on their back, stomach, right side, and left side, yielding 16 unique data sets. Each participant laid in at least one position that allowed for accurate tracking of heart rate, with seven of the 16 signals demonstrating heart rates with less than 2% error when compared to heart rates acquired with a commercial pulse oximeter. The stomach position appeared to offer the lowest total error, while lying on the right side offered the highest total error. Overall, heart rates acquired from this initial set of participants exhibited an average error of approximately 2.5% for all four positions.

A WIoT is a wireless network of low-power sensing nodes placed on the human body. While operating, these networks routinely collect physiological signals to send to offsite medical professionals for review. In this manner, these networks support a concept known as pervasive healthcare in which patients can be continuously monitored and treated remotely. Given that these networks are used to guide medical treatment and depend on transmitting sensitive data, it is important to ensure that the communication channel remains secure. Symmetric pairwise cryptography is a traditional scheme that can be used to provide such security. The scheme functions by sharing a cryptographic key between a pair of sensors. Once shared, the key can then be used by both parties to encrypt and decrypt all future messages. To configure a WIoT to support the use of symmetric pairwise cryptography a key distribution protocol is required. Schemes for pre-deployment are often used to perform this distribution. These schemes usually require inserting key information into WIoT devices before they can be used in the network. Unfortunately, this need to manually configure WIoT devices can decrease their usability. In this thesis we propose and evaluate an alternative approach to key distribution that uses physiological signals derived from accelerometer and gyroscope sensors. The evaluation of our approach indicates that more study is required to determine techniques that will enable ballistocardiography-derived physiological signals to provide secure key distribution.

Non-invasive remote detection of cardiac and blood displacements is an important topic in cardiac telemedicine. Here we propose kinocardiography (KCG), a non-invasive technique based onmeasurement of body vibrations produced by myocardial contraction and blood flow through thecardiac chambers and major vessels. KCG is based on ballistocardiography and seismocardiographyand measures 12 degrees-of-freedom (DOF) of body motion. The integral of kinetic energy (iK)and maximum Power (Pmax) obtained from the linear and rotational SCG/BCG signals, was computedover the cardiac cycle, and used as a marker of cardiac mechanical function. We showedthat KCG metrics show high repeatability, can be computed on 50 Hz and 1 kHz SCG/BCG signalsindifferently, that most of the metrics were highly similar when computed on different sensors,and with less than 5% of error when computed on record length longer than 60 s. Finally, weshow that KCG metrics allow detecting dobutamine-induced haemodynamic changes with a highaccuracy and present a major improvement over single axis ballistocardiography or seismocardiography.These results suggest that KCG may be a robust and non-invasive method to monitorcardiac inotropic activity. / La détection à distance et non invasive des déplacements cardiaques et sanguins est un sujet important en télémédecine. Nous proposons ici la kinocardiographie (KCG), une technique non invasive basée sur mesure des vibrations corporelles produites par la contraction du myocarde et par le flux sanguin au travers des cavités cardiaques et des principaux vaisseaux sanguins. La KCG est basée sur la balistocardiographie et la seismocardiographie et mesure 12 degrés de liberté (DOF) de mouvement corporel. L'intégrale de l'énergie cinétique (iK) et la puissance maximale (Pmax) obtenue à partir des signaux SCG / BCG linéaire et rotationnel, a été calculée au cours du cycle cardiaque, et sont utilisées comme marqueur de la fonction mécanique cardiaque. Ce travail montre que les métriques KCG sont caractérisées par une répétabilité élevée, peuvent être calculées sur des signaux SCG / BCG à 50 Hz et à 1 kHz indifféremment, que la plupart des métriques étaient très similaires lorsqu'elles étaient calculées sur différents capteurs, et avec moins de 5% d'erreur lors du calcul sur une longueur d'enregistrement supérieure à 60 s. Enfin ce travail montre que les métriques KCG permettent de détecter les changements hémodynamiques induits par la dobutamine avec précision et présentent une amélioration majeure par rapport à la balistocardiographie à un seul axe ou à la seismocardiographie. Ces résultats suggèrent que la KCG peut être une méthode robuste et non invasive pour surveiller l'activité inotrope du coeur. / Doctorat en Sciences de l'ingénieur et technologie / La défense publique a eu lieu le 05/05/2021. Cet upload remplace l'upload pécédent et contient les derniers commentaires du jury après la défense publique. / info:eu-repo/semantics/nonPublished

The 21st century brought tremendous changes in technologies, sweeping away every aspect of daily life, from social to private life, from education to professional world, at a breakneck pace. Medical science has not been spared by this colossal wave of changes: e-health, e-patients, e-physicians and e-medical students have already made their first appearance in the routine medical practice. Surgical robots, 3D printing, Artificial-Intelligence based imaging, smartwatches, wearable sensors are already used to improve diagnosis, personalize treatments and monitor patients’ health. Ballistocardiography (BCG) and seismocardiography (SCG) are ancient techniques which estimate the mechanical performance of the heart through measuring myocardial contraction-induced vibrations transmitted to the skin surface. They made their first appearance into clinic at the beginning of the 20th century to help in the diagnosis of cardiovascular diseases, but their popularity drastically declined in the middle of the 70s and they never had their place in clinical practice: cumbersome and complex equipment, ambiguity in the signal interpretation, unclear understanding of the physiological genesis of the signal, the advent of high-performing technologies (cardiac MRI, echocardiography) are a few of the reasons for their clinical failure. Thanks to the tremendous improvements in technologies, these techniques of the past came back to the medical world as wearable biosensors, first holding the promise of a remote and continuous monitoring of the cardiovascular status. Displacement, velocity and acceleration of blood mass flowing into cardiac chambers and main extracardiac vessels are recorded for each heartbeat along three cardinal axes and in two dimensions, a linear and a rotational one, by the renewed BCG and SCG, with 6 degrees-of-freedom (6D-BCG and 6D-SCG). By applying the Newtonian principles to the recorded signals, signal processing algorithms automatically compute the kinetic energy (KE) and its temporal integral (iK) for each cardiac cycle. This work first analyzed the influence of normal and pathological respiration as well as the effects of sympathetic overactivity on the genesis of the 6D-BCG and 6D-SCG signals and the iK parameters; secondary, it tested the usefulness of 6D-BCG and 6D-SCG techniques in the detection of cardiac dysfunction and hemodynamic impairment during acute myocardial infarction and reperfusion in an animal model for acute coronary syndrome. While breathing normally mildly affects cardiac iK parameters, pathological respiration profoundly alters them. During a sustained end-expiratory apnea, as it happens in patients suffering from central sleep apnea, the iK generated within a contractile cycle acutely increases at the end of the apnea, strictly depending on the magnitude of sympathetic activity; inspiring against a resistance, as it happens in patients suffering from obstructive sleep apnea, acutely increases the cardiac iK and this surge is related to the acute external force afterloading the left ventricle. So, whether it is a central apnea or an obstructive one, myocardial mechanical function as expressed in terms of iK is profoundly impaired, suggesting the myocardium is enduring a sustain endeavor during these pathological respirations. During an experimental acute myocardial infarction, in a context of mechanical ventilation without major respiratory events, iK parameters drastically drop during coronary occlusion and does not improve during reperfusion, along with systemic blood pressures and cardiac output, thus holding the promise to continuously monitor the cardiac contractile function and the hemodynamic profile both during acute coronary occlusion and reperfusion. Renewed and wearable 6D-BCG and 6D-SCG may prove useful in the detection and continuous monitoring of cardiac dysfunction and hemodynamic impairment in patients suffering from sleep disordered breathing and may be used in the mid-long-term follow-up of patients with myocardial dysfunction of ischemic origin. There is still a lot of work to do before validating these renewed technologies in the practice of cardiovascular medicine, but evidences are there to consider them as next generation medical devices. / Doctorat en Sciences médicales (Médecine) / info:eu-repo/semantics/nonPublished

Les approches actuelles pour diagnostiquer les troubles du sommeil sont lourdes, intrusives et peuvent influer sur la qualité du sommeil du patient. Il y a donc un besoin crucial de systèmes moins encombrants pour diagnostiquer les problèmes liés au sommeil. Nous proposons d'utiliser un nouveau système de suivi du sommeil non intrusif basé sur un tapis à fibre optique à microflexion placée sous le matelas de lit. La qualité du sommeil est évaluée en fonction de différents paramètres, y compris la fréquence cardiaque, le rythme respiratoire, les mouvements du corps, l’heure du réveil, la durée du sommeil, le mouvement nocturne et l’heure du coucher. Le système proposé a été validé dans un environnement de santé et de bien-être, en plus d'un environnement clinique comme suit. Dans le premier cas, la fréquence cardiaque est mesurée à partir de signaux ballistocardiogramme bruités acquis de 50 volontaires en position assise à l'aide d'une chaise de massage. Les signaux sont recueillis discrètement à partir d'un capteur de fibre optique microflexible intégrée dans l'appui-tête de la chaise, puis transmis à un ordinateur par une connexion Bluetooth. La fréquence cardiaque est calculée à l'aide de l'analyse multi-résolution de la transformée discrète en ondelettes à chevauchement maximal. L'erreur entre la méthode proposée et électrocardiogramme de référence est estimée en battements par minute en utilisant l'erreur absolue moyenne où le système a obtenu des résultats relativement bons (10.12±4.69) malgré la quantité remarquable d'artefact de mouvement produit en raison des fréquents mouvements corporels et/ou vibrations de la chaise de massage pendant le massage de soulagement du stress. Contrairement à l'algorithme complet de décomposition du mode empirique de l'ensemble, précédemment utilisé pour l'estimation de la fréquence cardiaque, le système proposé est beaucoup plus rapide. Par conséquent, il peut être utilisé dans les applications temps réel. Dans ce dernier cas, nous avons évalué la capacité du capteur de fibre optique microflexible pour suivre la fréquence cardiaque et la respiration d’une manière discrète. En outre, nous avons testé la capacité du capteur dans la discrimination entre la respiration superficielle et pas de respiration. Le capteur proposé a été comparé à un dispositif de surveillance portatif à trois canaux (ApneaLink) dans un milieu clinique au cours d'une endoscopie sous anesthésie. Parmi les dix patients recrutés pour notre étude, le système a obtenu des résultats satisfaisants quant à la fréquence cardiaque moyenne et quant à la fréquence respiratoire moyenne avec une erreur de 0.55 ± 0.59 battements/minute et de 0.38 ± 0.32 respirations/minute, respectivement. De plus, le coefficient de corrélation Pearson entre le capteur proposé et le dispositif de référence était de 0.96 et 0.78 pour la fréquence cardiaque et la respiration, respectivement. Au contraire, le capteur proposé a fourni une très faible sensibilité (24.24 ± 12.81%) et une spécificité relativement élevée (85.88 ± 6.01%) pour la détection de l'apnée du sommeil. On s'attend à ce que cette recherche préliminaire ouvre la voie vers la détection discrète de l'apnée obstructive du sommeil en temps réel. Suite à la validation réussie du système proposé, nous avons déployé avec succès notre système de surveillance du sommeil pendant plus de 6 mois dans treize appartements habités principalement par les personnes âgées. Néanmoins, dans cette recherche, nous nous concentrons sur un déploiement d'un mois avec trois résidents seniors de sexe féminin. Le système proposé montre l’accord avec l’enquête utilisateur recueillie avant l'étude. En outre, le système est intégré dans une plate-forme d’autonomie assistée existante avec une interface conviviale pour rendre plus commode pour les aidants le suivi des paramètres de sommeil des résidents. / The current approaches for diagnosing sleep disorders are burdensome, intrusive, and can affect the patient’s sleep quality. As a result, there is a crucial need for less cumbersome systems to diagnose sleep-related problems. We propose to use a novel nonintrusive sleep monitoring system based on a microbend fiber-optic mat placed under the bed mattress. The sleep quality is assessed based on different parameters, including heart rate, breathing rate, body movements, wake up time, sleep time, night movement, and bedtime. The proposed system has been validated in a health and wellness environment in addition to a clinical environment as follows. In the former case, the heart rate is measured from noisy ballistocardiogram signals acquired from 50 human volunteers in a sitting position using a massage chair. The signals are unobtrusively collected from a microbend fiber optic sensor embedded within the headrest of the chair and then transmitted to a computer through a Bluetooth connection. The heart rate is computed using the multiresolution analysis of the maximal overlap discrete wavelet transform. The error between the proposed method and the reference ECG is estimated in beats per minute using the mean absolute error where the system achieved relatively good results (10.12 ± 4.69) despite the remarkable amount of motion artifact produced owing to the frequent body movements and/or vibrations of the massage chair during stress relief massage. Unlike the complete ensemble empirical mode decomposition algorithm, previously employed for heart rate estimation, the suggested system is much faster. Hence, it can be used in real-time applications. In the latter case, we evaluated the capacity of the microbend fiber optic sensor to monitor heart rate and respiration unobtrusively. In addition, we tested the capacity of the sensor in discriminating between shallow breathing and no breathing. The proposed sensor was compared to a three-channel portable monitoring device (ApneaLink) in a clinical setting during a drug-induced sleep endoscopy. Across all ten patients recruited for our study, the system achieved satisfactory results in the mean heart rate and the mean respiratory rate with an error of 0.55±0.59 beats/minute and 0.38 ± 0.32 breaths/minute, respectively. Besides, the Pearson correlation coefficient between the proposed sensor and the reference device was 0.96 and 0.78 for heart rate and respiration, respectively. On the contrary, the proposed sensor provided a very low sensitivity (24.24 ± 12.81%) and a relatively high specificity (85.88 ± 6.01%) for sleep apnea detection. It is expected that this preliminary research will pave the way toward unobtrusive detection of obstructive sleep apnea in real-time. Following successful validation of the proposed system, we have successfully deployed our sleep monitoring system in thirteen apartments with mainly senior residents over six months. Nevertheless, in this research, we concentrate on a one-month deployment with three senior female residents. The proposed system shows an agreement with a user’s survey collected before the study. Furthermore, the system is integrated within an existing ambient assisted living platform with a user-friendly interface to make it more convenient for the caregivers to follow-up the sleep parameters of the residents.