A Multimodal Dataset for Automatic Edge-AI Cough Detection.

Counting the number of times a patient coughs per day is an essential biomarker in determining treatment efficacy for novel antitussive therapies and personalizing patient care. Automatic cough counting tools must provide accurate information, while running on a lightweight, portable device that protects the patient's privacy. Several devices and algorithms have been developed for cough counting, but many use only error-prone audio signals, rely on offline processing that compromises data privacy, or utilize processing and memory-intensive neural networks that require more hardware resources than can fit on a wearable device. Therefore, there is a need for wearable devices that employ multimodal sensors to perform accurate, privacy-preserving, automatic cough counting algorithms directly on the device in an edge Artificial Intelligence (edge-AI) fashion. To advance this research field, we contribute the first publicly accessible cough counting dataset of multimodal biosignals. The database contains nearly 4 hours of biosignal data, with both acoustic and kinematic modalities, covering 4,300 annotated cough events from 15 subjects. Furthermore, a variety of non-cough sounds and motion scenarios mimicking daily life activities are also present, which the research community can use to accelerate machine learning (ML) algorithm development. A technical validation of the dataset reveals that it represents a wide variety of signal-to-noise ratios, which can be expected in a real-life use case, as well as consistency across experimental trials. Finally, to demonstrate the usability of the dataset, we train a simple cough vs non-cough signal classifier that obtains a 91% sensitivity, 92% specificity, and 80% precision on unseen test subject data. Such edge-friendly AI algorithms have the potential to provide continuous ambulatory monitoring of the numerous chronic cough patients.

A semi-supervised algorithm for improving the consistency of crowdsourced datasets: The COVID-19 case study on respiratory disorder classification.

BACKGROUND AND OBJECTIVE: Cough audio signal classification is a potentially useful tool in screening for respiratory disorders, such as COVID-19. Since it is dangerous to collect data from patients with contagious diseases, many research teams have turned to crowdsourcing to quickly gather cough sound data. The COUGHVID dataset enlisted expert physicians to annotate and diagnose the underlying diseases present in a limited number of recordings. However, this approach suffers from potential cough mislabeling, as well as disagreement between experts. METHODS: In this work, we use a semi-supervised learning (SSL) approach - based on audio signal processing tools and interpretable machine learning models - to improve the labeling consistency of the COUGHVID dataset for 1) COVID-19 versus healthy cough sound classification 2) distinguishing wet from dry coughs, and 3) assessing cough severity. First, we leverage SSL expert knowledge aggregation techniques to overcome the labeling inconsistencies and label sparsity in the dataset. Next, our SSL approach is used to identify a subsample of re-labeled COUGHVID audio samples that can be used to train or augment future cough classifiers. RESULTS: The consistency of the re-labeled COVID-19 and healthy data is demonstrated in that it exhibits a high degree of inter-class feature separability: 3x higher than that of the user-labeled data. Similarly, the SSL method increases this separability by 11.3x for cough type and 5.1x for severity classifications. Furthermore, the spectral differences in the user-labeled audio segments are amplified in the re-labeled data, resulting in significantly different power spectral densities between healthy and COVID-19 coughs in the 1-1.5 kHz range (p=1.2×10-64), which demonstrates both the increased consistency of the new dataset and its explainability from an acoustic perspective. Finally, we demonstrate how the re-labeled dataset can be used to train a COVID-19 classifier, achieving an AUC of 0.797. CONCLUSIONS: We propose a SSL expert knowledge aggregation technique for the field of cough sound classification for the first time, and demonstrate how it can be used to combine the medical knowledge of multiple experts in an explainable fashion, thus providing abundant, consistent data for cough classification tasks.

The COUGHVID crowdsourcing dataset, a corpus for the study of large-scale cough analysis algorithms.

Cough audio signal classification has been successfully used to diagnose a variety of respiratory conditions, and there has been significant interest in leveraging Machine Learning (ML) to provide widespread COVID-19 screening. The COUGHVID dataset provides over 25,000 crowdsourced cough recordings representing a wide range of participant ages, genders, geographic locations, and COVID-19 statuses. First, we contribute our open-sourced cough detection algorithm to the research community to assist in data robustness assessment. Second, four experienced physicians labeled more than 2,800 recordings to diagnose medical abnormalities present in the coughs, thereby contributing one of the largest expert-labeled cough datasets in existence that can be used for a plethora of cough audio classification tasks. Finally, we ensured that coughs labeled as symptomatic and COVID-19 originate from countries with high infection rates. As a result, the COUGHVID dataset contributes a wealth of cough recordings for training ML models to address the world's most urgent health crises.

Towards Continuous and Ambulatory Blood Pressure Monitoring: Methods for Efficient Data Acquisition for Pulse Transit Time Estimation.

We developed a prototype for measuring physiological data for pulse transit time (PTT) estimation that will be used for ambulatory blood pressure (BP) monitoring. The device is comprised of an embedded system with multimodal sensors that streams high-throughput data to a custom Android application. The primary focus of this paper is on the hardware-software codesign that we developed to address the challenges associated with reliably recording data over Bluetooth on a resource-constrained platform. In particular, we developed a lossless compression algorithm that is based on optimally selective Huffman coding and Huffman prefixed coding, which yields virtually identical compression ratios to the standard algorithm, but with a 67-99% reduction in the size of the compression tables. In addition, we developed a hybrid software-hardware flow control method to eliminate microcontroller (MCU) interrupt-latency related data loss when multi-byte packets are sent from the phone to the embedded system via a Bluetooth module at baud rates exceeding 115,200 bit/s. The empirical error rate obtained with the proposed method with the baud rate set to 460,800 bit/s was identically equal to 0%. Our robust and computationally efficient physiological data acquisition system will enable field experiments that will drive the development of novel algorithms for PTT-based continuous BP monitoring.

A Globalized Model for Mapping Wearable Seismocardiogram Signals to Whole-Body Ballistocardiogram Signals Based on Deep Learning.

The ballistocardiography (BCG) signal is a measurement of the vibrations of the center of mass of the body due to the cardiac cycle and can be used for noninvasive hemodynamic monitoring. The seismocardiography (SCG) signals measure the local vibrations of the chest wall due to the cardiac cycle. While BCG is a more well-known modality, it requires the use of a modified bathroom scale or a force plate and cannot be measured in a wearable setting, whereas SCG signals can be measured using wearable accelerometers placed on the sternum. In this paper, we explore the idea of finding a mapping between zero mean and unit l2-norm SCG and BCG signal segments such that, the BCG signal can be acquired using wearable accelerometers (without retaining amplitude information). We use neural networks to find such a mapping and make use of the recently introduced UNet architecture. We trained our models on 26 healthy subjects and tested them on ten subjects. Our results show that we can estimate the aforementioned segments of the BCG signal with a median Pearson correlation coefficient of 0.71 and a median absolute deviation (MAD) of 0.17. Furthermore, our model can estimate the R-I, R-J and R-K timing intervals with median absolute errors (and MAD) of 10.00 (8.90), 6.00 (5.93), and 8.00 (5.93), respectively. We show that using all three axis of the SCG accelerometer produces the best results, whereas the head-to-foot SCG signal produces the best results when a single SCG axis is used.

REWARD: Design, Optimization, and Evaluation of a Real-Time Relative-Energy Wearable R-Peak Detection Algorithm.

Wearable devices are an unobtrusive, cost-effective means of continuous ambulatory monitoring of chronic cardiovascular diseases. However, on these resource-constrained systems, electrocardiogram (ECG) processing algorithms must consume minimal power and memory, yet robustly provide accurate physiological information. This work presents REWARD, the Relative-Energy-based WeArable R-Peak Detection algorithm, which is a novel ECG R-peak detection mechanism based on a nonlinear filtering method called Relative-Energy (Rel-En). REWARD is designed and optimized for real-time execution on wearable systems. Then, this novel algorithm is compared against three state-of-the-art real-time R-peak detection algorithms in terms of accuracy, memory footprint, and energy consumption. The Physionet QT and NST Databases were employed to evaluate the algorithms' accuracy and robustness to noise, respectively. Then, a 32-bit ARM Cortex-M3-based microcontroller was used to measure the energy usage, computational burden, and memory footprint of the four algorithms. REWARD consumed at least 63% less energy and 32% less RAM than the other algorithms while obtaining comparable accuracy results. Therefore, REWARD would be a suitable choice of R-peak detection mechanism for wearable devices that perform more complex ECG analysis, whose algorithms require additional energy and memory resources.

Unobtrusive Estimation of Cardiac Contractility and Stroke Volume Changes Using Ballistocardiogram Measurements on a High Bandwidth Force Plate.

Unobtrusive and inexpensive technologies for monitoring the cardiovascular health of heart failure (HF) patients outside the clinic can potentially improve their continuity of care by enabling therapies to be adjusted dynamically based on the changing needs of the patients. Specifically, cardiac contractility and stroke volume (SV) are two key aspects of cardiovascular health that change significantly for HF patients as their condition worsens, yet these parameters are typically measured only in hospital/clinical settings, or with implantable sensors. In this work, we demonstrate accurate measurement of cardiac contractility (based on pre-ejection period, PEP, timings) and SV changes in subjects using ballistocardiogram (BCG) signals detected via a high bandwidth force plate. The measurement is unobtrusive, as it simply requires the subject to stand still on the force plate while holding electrodes in the hands for simultaneous electrocardiogram (ECG) detection. Specifically, we aimed to assess whether the high bandwidth force plate can provide accuracy beyond what is achieved using modified weighing scales we have developed in prior studies, based on timing intervals, as well as signal-to-noise ratio (SNR) estimates. Our results indicate that the force plate BCG measurement provides more accurate timing information and allows for better estimation of PEP than the scale BCG (r² = 0.85 vs. r² = 0.81) during resting conditions. This correlation is stronger during recovery after exercise due to more significant changes in PEP (r² = 0.92). The improvement in accuracy can be attributed to the wider bandwidth of the force plate. ∆SV (i.e., changes in stroke volume) estimations from the force plate BCG resulted in an average error percentage of 5.3% with a standard deviation of ±4.2% across all subjects. Finally, SNR calculations showed slightly better SNR in the force plate measurements among all subjects but the small difference confirmed that SNR is limited by motion artifacts rather than instrumentation.