Health telemonitoring systems are constrained by the computational and data transmission load resulting from the large volumes of various measured signals, e.g. in the fall detection application. Nevertheless, the trend of movement and the implementation of computer intelligence in intelligent devices ensure an intelligent and convenient method for continuous real‐time telemonitoring of health conditions. In this paper, fall detection is presented while leveraging edge computing integrated on a multi‐level architecture combines the Wireless Sensors Network and the Internet of Things. Particularly, we present a complete study and implementation scenarios while investigating the performances of machine learning algorithms to distinguish between different fall patterns and activities of daily living using a set of significant extracted features from measured acceleration and angular velocity signals. For low computational requirements and to improve the classification performances, the Linear Discriminant Analysis is used to reduce the dimensionality of extracted features. The experimental results assess the performances of the proposed approach in fall detection that show the highest accuracy of 99.92% provided using the KNN classifier and accuracy of 97.5% for fall pattern recognition using the SVM classifier. Also, the online classification on the Fog device reached an accuracy of 94.42% using the SVM classifier.

Plant diseases are a major threat to food security. Thus, the identification of these diseases is crucial to alleviate the problem. Deep learning combined with image processing has proven to be efficient in order to identify such diseases accurately, automatically and quickly. This article presents a general architecture for plant disease prediction based on the Hybrid Wireless Sensor Network, the Internet of Things, and Deep Learning algorithms. In this work, the focus is on learning models for plant disease detection to be implemented afterward on the structure for actual trials. Thus, several deep learning architectures have been tested, i.e. ResNet50, 101, and 152, AlexNet, VGG, SqueezeNet, DenseNet, and Inception. In addition to that, a new parallel architecture has been developed and evaluated by combining two deep learning architectures. Several optimizers such as SGD and ADAM have been leveraged in order to improve the performances. The public published database PlantVillage is used for model learning, which contains 54.306 images of agricultural plant leaves. As a result, the best validation accuracy score of 98,09% is achieved using the presented para model with ADAM optimizer running with the 80/20 splitting data on the 14th epoch. Copyright © 2021 Praise Worthy Prize - All rights reserved.

The continuous monitoring of oxygen saturation (SpO₂) and respiratory rates (RRs) are major clinical issues in many cardio-respiratory diseases and have been of tremendous importance during the COVID-19 pandemic. The early detection of hypoxemia was crucial since it precedes significant complications, and SpO₂ follow-up allowed early hospital discharge in patients needing oxygen therapy. Nevertheless, fingertip devices showed some practical limitations. In this study, we investigated the reliability of the new Multisense^® pulse oximetry system compared to a reference pulse oximeter (Vyntus CPX Pulse Oximeter) during hypoxia. In a population of sixteen healthy male subjects (mean age: 31.5 ± 7.0 years, BMI: 24.9 ± 3.6 kg/m², and 35% with darker skin tones), simultaneous SpO₂ and RR measurements were collected over 12.4 h, during which FiO₂ was progressively reduced from 21% to 10.5%. The average root mean square error (ARMS) of SpO₂ for Multisense^® placed on the back and chest was 2.94% and 2.98%, respectively, with permutation testing confirming a significant ARMS below 3.5% for both positions and no statistically significant difference in the ARMS between patch placements. Positive correlations and acceptable accuracy between devices were observed at both locations (r = 0.92, p < 0.001 and r = 0.90, p < 0.001 for back and chest placements, respectively). Bland–Altman analysis further indicated limits of agreement that support consistency across placements, with similar agreement levels noted across skin tones. Similar findings were obtained with the RR measurements. In conclusion, Multisense^® demonstrated robust accuracy in measuring SpO₂ and RRs during hypoxia in humans comparable to standard hospital-grade equipment. The effectiveness of the findings suggests that this wearable device is a valuable tool for the continuous monitoring of SpO₂ and RRs, potentially enhancing patient safety and optimizing hospital resource allocation. Nevertheless, to overcome study limitations and allow generalized use, further work on a larger population sample, including more subjects with a high phototype and desaturation below 80%, would be useful.