Deep Learning for Walking Behaviour Detection in Elderly People Using Smart Footwear

BodyFlow: An Open-Source Library for Multimodal Human Activity Recognition

Human activity recognition is a critical task for various applications across healthcare, sports, security, gaming, and other fields. This paper presents BodyFlow, a comprehensive library that seamlessly integrates human pose estimation and multiple-person estimation and tracking, along with activity recognition modules. BodyFlow enables users to effortlessly identify common activities and 2D/3D body joints from input sources such as videos, image sets, or webcams. Additionally, the library can simultaneously process inertial sensor data, offering users the flexibility to choose their preferred input, thus facilitating multimodal human activity recognition. BodyFlow incorporates state-of-the-art algorithms for 2D and 3D pose estimation and three distinct models for human activity recognition.

IoT-Based Wireless System for Gait Kinetics Monitoring in Multi-Device Therapeutic Interventions

This study presents an IoT-based gait analysis system employing insole pressure sensors to assess gait kinetics. The system integrates piezoresistive sensors within a left foot insole, with data acquisition managed using an ESP32 board that communicates via Wi-Fi through an MQTT IoT framework. In this initial protocol study, we conducted a comparative analysis using the Zeno system, supported by PKMAS as the gold standard, to explore the correlation and agreement of data obtained from the insole system. Four volunteers (two males and two females, aged 24-28, without gait disorders) participated by walking along a 10 m Zeno system path, equipped with pressure sensors, while wearing the insole system. Vertical ground reaction force (vGRF) data were collected over four gait cycles. The preliminary results indicated a strong positive correlation (r = 0.87) between the insole and the reference system measurements. A Bland-Altman analysis further demonstrated a mean difference of approximately (0.011) between the two systems, suggesting a minimal yet significant bias. These findings suggest that piezoresistive sensors may offer a promising and cost-effective solution for gait disorder assessment and monitoring. However, operational factors such as high temperatures and sensor placement within the footwear can introduce noise or unwanted signal activation. The communication framework proved functional and reliable during this protocol, with plans for future expansion to multi-device applications. It is important to note that additional validation studies with larger sample sizes are required to confirm the system's reliability and robustness for clinical and research applications.

Recent Innovations in Footwear and the Role of Smart Footwear in Healthcare-A Survey

Smart shoes have ushered in a new era of personalised health monitoring and assistive technologies. Smart shoes leverage technologies such as Bluetooth for data collection and wireless transmission, and incorporate features such as GPS tracking, obstacle detection, and fitness tracking. As the 2010s unfolded, the smart shoe landscape diversified and advanced rapidly, driven by sensor technology enhancements and smartphones' ubiquity. Shoes have begun incorporating accelerometers, gyroscopes, and pressure sensors, significantly improving the accuracy of data collection and enabling functionalities such as gait analysis. The healthcare sector has recognised the potential of smart shoes, leading to innovations such as shoes designed to monitor diabetic foot ulcers, track rehabilitation progress, and detect falls among older people, thus expanding their application beyond fitness into medical monitoring. This article provides an overview of the current state of smart shoe technology, highlighting the integration of advanced sensors for health monitoring, energy harvesting, assistive features for the visually impaired, and deep learning for data analysis. This study discusses the potential of smart footwear in medical applications, particularly for patients with diabetes, and the ongoing research in this field. Current footwear challenges are also discussed, including complex construction, poor fit, comfort, and high cost.

Preparation of Elastic Macroporous Graphene Aerogel Based on Pickering Emulsion Method and Combination with ETPU for High Performance Piezoresistive Sensors

High-performance pressure sensors provide the necessary conditions for smart shoe applications. In this paper, the elastic Macroporous Graphene Aerogel (MGA) was synthesized via the modified Hummers' method, and it was further combined with Expanded-Thermoplastic polyurethane (ETPU) particles to assemble MGA-ETPU flexible sensors. The MGA-ETPU has a low apparent density (3.02 mg/cm3), high conductivity (0.024 S/cm) and fast response time (50 ms). The MGA-ETPU has a large linear sensing range (0-10 kPa) and consists of two linear regions: the low-pressure region (0 to 8 kPa) and the high-pressure region (8 to 10 kPa), with sensitivities of 0.08 kPa-1, and 0.246 kPa-1, respectively. Mechanical test results show that the MGA-ETPU sensor showed 19% reduction in maximum stress after 400 loading-unloading compression cycles at 40% strain. Electrical performance tests showed that the resistance of MGA-ETPU sensor decreased by 12.5% when subjected to sudden compression at 82% strain and returned to its original state within 0.05 s. Compared to existing flexible sensors, the MGA-ETPU sensors offer excellent performance and several distinct advantages, including ease of fabrication, high sensitivity, fast response time, and good flexibility. These remarkable features make them ideally suited as flexible pressure sensors for smart shoes.

DeepHAR: a deep feed-forward neural network algorithm for smart insole-based human activity recognition

Health monitoring, rehabilitation, and fitness are just a few domains where human activity recognition can be applied. In this study, a deep learning approach has been proposed to recognise ambulation and fitness activities from data collected by five participants using smart insoles. Smart insoles, consisting of pressure and inertial sensors, allowed for seamless data collection while minimising user discomfort, laying the baseline for the development of a monitoring and/or rehabilitation system for everyday life. The key objective has been to enhance the deep learning model performance through several techniques, including data segmentation with overlapping technique (2 s with 50% overlap), signal down-sampling by averaging contiguous samples, and a cost-sensitive re-weighting strategy for the loss function for handling the imbalanced dataset. The proposed solution achieved an Accuracy and F1-Score of 98.56% and 98.57%, respectively. The Sitting activities obtained the highest degree of recognition, closely followed by the Spinning Bike class, but fitness activities were recognised at a higher rate than ambulation activities. A comparative analysis was carried out both to determine the impact that pre-processing had on the proposed core architecture and to compare the proposed solution with existing state-of-the-art solutions. The results, in addition to demonstrating how deep learning solutions outperformed those of shallow machine learning, showed that in our solution the use of data pre-processing increased performance by about 2%, optimising the handling of the imbalanced dataset and allowing a relatively simple network to outperform more complex networks, reducing the computational impact required for such applications.

mHealth in sub-Saharan Africa and Europe: A systematic review comparing the use and availability of mHealth approaches in sub-Saharan Africa and Europe

Background

mHealth can help with healthcare service delivery for various health issues, but there's a significant gap in the availability and use of mHealth systems between sub-Saharan Africa and Europe, despite the ongoing digitalization of the global healthcare system.

Objective

This work aims to compare and investigate the use and availability of mHealth systems in sub-Saharan Africa and Europe, and identify gaps in current mHealth development and implementation in both regions.

Methods

The study adhered to the PRISMA 2020 guidelines for article search and selection to ensure an unbiased comparison between sub-Saharan Africa and Europe. Four databases (Scopus, Web of Science, IEEE Xplore, and PubMed) were used, and articles were evaluated based on predetermined criteria. Details on the mHealth system type, goal, patient type, health concern, and development stage were collected and recorded in a Microsoft Excel worksheet.

Results

The search query produced 1020 articles for sub-Saharan Africa and 2477 articles for Europe. After screening for eligibility, 86 articles for sub-Saharan Africa and 297 articles for Europe were included. To minimize bias, two reviewers conducted the article screening and data retrieval. Sub-Saharan Africa used SMS and call-based mHealth methods for consultation and diagnosis, mainly for young patients such as children and mothers, and for issues such as HIV, pregnancy, childbirth, and child care. Europe relied more on apps, sensors, and wearables for monitoring, with the elderly as the most common patient group, and the most common health issues being cardiovascular disease and heart failure.

Conclusion

Wearable technology and external sensors are heavily used in Europe, whereas they are seldom used in sub-Saharan Africa. More efforts should be made to use the mHealth system to improve health outcomes in both regions, incorporating more cutting-edge technologies like wearables internal and external sensors. Undertaking context-based studies, identifying determinants of mHealth systems use, and considering these determinants during mHealth system design could enhance mHealth availability and utilization.

A Comprehensive Review on Smart Health Care: Applications, Paradigms, and Challenges with Case Studies

Growth and advancement of the Deep Learning (DL) and the Internet of Things (IoT) are figuring out their way over the modern contemporary world through integrating various technologies in distinct fields viz, agriculture, manufacturing, energy, transportation, supply chains, cities, healthcare, and so on. Researchers had identified the feasibility of integrating deep learning, cloud, and IoT to enhance the overall automation, where IoT may prolong its application area through utilizing cloud services and the cloud can even prolong its applications through data acquired by IoT devices like sensors and deep learning for disease detection and diagnosis. This study explains a summary of various techniques utilized in smart healthcare, i.e., deep learning, cloud-based-IoT applications in smart healthcare, fog computing in smart healthcare, and challenges and issues faced by smart healthcare and it presents a wider scope as it is not intended for a particular application such aspatient monitoring, disease detection, and diagnosing and the technologies used for developing this smart systems are outlined. Smart health bestows the quality of life. Convenient and comfortable living is made possible by the services provided by smart healthcare systems (SHSs). Since healthcare is a massive area with enormous data and a broad spectrum of diseases associated with different organs, immense research can be done to overcome the drawbacks of traditional healthcare methods. Deep learning with IoT can effectively be applied in the healthcare sector to automate the diagnosing and treatment process even in rural areas remotely. Applications may include disease prevention and diagnosis, fitness and patient monitoring, food monitoring, mobile health, telemedicine, emergency systems, assisted living, self-management of chronic diseases, and so on.

Machine Learning Algorithm to Predict Acidemia Using Electronic Fetal Monitoring Recording Parameters

Background: Electronic fetal monitoring (EFM) is the universal method for the surveillance of fetal well-being in intrapartum. Our objective was to predict acidemia from fetal heart signal features using machine learning algorithms. Methods: A case–control 1:2 study was carried out compromising 378 infants, born in the Miguel Servet University Hospital, Spain. Neonatal acidemia was defined as pH < 7.10. Using EFM recording logistic regression, random forest and neural networks models were built to predict acidemia. Validation of models was performed by means of discrimination, calibration, and clinical utility. Results: Best performance was attained using a random forest model built with 100 trees. The discrimination ability was good, with an area under the Receiver Operating Characteristic curve (AUC) of 0.865. The calibration showed a slight overestimation of acidemia occurrence for probabilities above 0.4. The clinical utility showed that for 33% cutoff point, missing 5% of acidotic cases, 46% of unnecessary cesarean sections could be prevented. Logistic regression and neural networks showed similar discrimination ability but with worse calibration and clinical utility. Conclusions: The combination of the variables extracted from EFM recording provided a predictive model of acidemia that showed good accuracy and provides a practical tool to prevent unnecessary cesarean sections.

A Clinical Decision Web to Predict ICU Admission or Death for Patients Hospitalised with COVID-19 Using Machine Learning Algorithms

The purpose of the study was to build a predictive model for estimating the risk of ICU admission or mortality among patients hospitalized with COVID-19 and provide a user-friendly tool to assist clinicians in the decision-making process. The study cohort comprised 3623 patients with confirmed COVID-19 who were hospitalized in the SALUD hospital network of Aragon (Spain), which includes 23 hospitals, between February 2020 and January 2021, a period that includes several pandemic waves. Up to 165 variables were analysed, including demographics, comorbidity, chronic drugs, vital signs, and laboratory data. To build the predictive models, different techniques and machine learning (ML) algorithms were explored: multilayer perceptron, random forest, and extreme gradient boosting (XGBoost). A reduction dimensionality procedure was used to minimize the features to 20, ensuring feasible use of the tool in practice. Our model was validated both internally and externally. We also assessed its calibration and provide an analysis of the optimal cut-off points depending on the metric to be optimized. The best performing algorithm was XGBoost. The final model achieved good discrimination for the external validation set (AUC = 0.821, 95% CI 0.787-0.854) and accurate calibration (slope = 1, intercept = -0.12). A cut-off of 0.4 provides a sensitivity and specificity of 0.71 and 0.78, respectively. In conclusion, we built a risk prediction model from a large amount of data from several pandemic waves, which had good calibration and discrimination ability. We also created a user-friendly web application that can aid rapid decision-making in clinical practice.