Human Activity Recognition (HAR) is the automatic detection and understanding of human motion behavior based on data extracted from video camera, ambient sensors or wearable sensors which particularly has recently attracted increased attention from both researchers and industry. However, for running practical HAR systems on wearable devices, there are some requirements such as design and development of small, lightweight, powerful, and low-cost smart sensors. In this context, data must be continuously collected, and edge computing is a viable solution, which is an energy-efficient technique, offering real-time response and privacy requirements for HAR applications. Rather than sending data to the cloud, edge computing is a local process that minimizes the data transmission time and responds with low latency. Recently, HAR system designers have adopted deep learning techniques inspired by their outstanding performance in many application areas and achieved relevant gain in activity recognition performance, however, these techniques were not demonstrated suitable for running on resource constrained devices. Thus, designing energy-efficient deep learning models is critical for realizing efficient HAR for mobile applications. In this work, we present a lightweight framework for the deployment of low-power but accurate HAR systems for these devices. We also implement and run the system in a microcontroller and analyze computational cost and energy consumption, and how different system configurations and deep learning model complexity influence on that.

COVID-19 occurs due to infection through respiratory droplets containing the SARS-CoV-2 virus, which are released when someone sneezes, coughs, or talks. The gold-standard exam to detect the virus is Real-Time Polymerase Chain Reaction (RT-PCR); however, this is an expensive test and may require up to 3 days after infection for a reliable result, and if there is high demand, the labs could be overwhelmed, which can cause significant delays in providing results. Biomedical data (oxygen saturation level—SpO2, body temperature, heart rate, and cough) are acquired from individuals and are used to help infer infection by COVID-19, using machine learning algorithms. The goal of this study is to introduce the Integrated Portable Medical Assistant (IPMA), which is a multimodal piece of equipment that can collect biomedical data, such as oxygen saturation level, body temperature, heart rate, and cough sound, and helps infer the diagnosis of COVID-19 through machine learning algorithms. The IPMA has the capacity to store the biomedical data for continuous studies and can be used to infer other respiratory diseases. Quadratic kernel-free non-linear Support Vector Machine (QSVM) and Decision Tree (DT) were applied on three datasets with data of cough, speech, body temperature, heart rate, and SpO2, obtaining an Accuracy rate (ACC) and Area Under the Curve (AUC) of approximately up to 88.0% and 0.85, respectively, as well as an ACC up to 99% and AUC = 0.94, respectively, for COVID-19 infection inference. When applied to the data acquired with the IMPA, these algorithms achieved 100% accuracy. Regarding the easiness of using the equipment, 36 volunteers reported that the IPMA has a high usability, according to results from two metrics used for evaluation: System Usability Scale (SUS) and Post Study System Usability Questionnaire (PSSUQ), with scores of 85.5 and 1.41, respectively. In light of the worldwide needs for smart equipment to help fight the COVID-19 pandemic, this new equipment may help with the screening of COVID-19 through data collected from biomedical signals and cough sounds, as well as the use of machine learning algorithms.

This article describes the methodology used to train and test a Deep Neural Network (DNN) with Photoplethysmography (PPG) data performing a regression task to estimate the Respiratory Rate (RR). The DNN architecture is based on a model used to infer the heart rate (HR) from noisy PPG signals, which is optimized to the RR problem using genetic optimization. Two open-access datasets were used in the tests, the BIDMC and the CapnoBase. With the CapnoBase dataset, the DNN achieved a median error of 1.16 breaths/min, which is comparable with analytical methods in the literature, in which the best error found is 1.1 breaths/min (excluding the 8 % noisiest data). The BIDMC dataset seems to be more challenging, as the minimum median error of the literature’s methods is 2.3 breaths/min (excluding 6 % of the noisiest data), and the DNN based approach achieved a median error of 1.52 breaths/min with the whole dataset.