Explore chapters and articles related to this topic
IoT in the Healthcare Sector
Published in Govind Singh Patel, Seema Nayak, Sunil Kumar Chaudhary, Machine Learning, Deep Learning, Big Data, and Internet of Things for Healthcare, 2023
A.K. Awasthi, Sanjeev Kumar, Arun Kumar Garov
The basic topology of IoT in healthcare consists of various components that are coherently interconnected in the healthcare environment, commonly known as HIoT. Processing and dissemination of data with the help of publisher, broker, and subscribers in different stages are shown in Figure 7.1. The publisher corresponds to a network of connected sensors which record the patient’s important information. The broker receives this information from the publisher continuously and processes the stored data in the cloud. In the final stage, the subscriber continuously monitors the patient’s information, which can be accessed with the help of smartphones or laptops. The feedback about the patient’s health condition is given by the publisher. This structure is not only the topology that is used everywhere but in the past decade, numerous other IoT architectures have also been proposed for the healthcare environment. The foremost requirement is to follow the medical rules and steps in the diagnosis procedure.
Solutions Using Machine Learning for Diabetes
Published in Punit Gupta, Dinesh Kumar Saini, Rohit Verma, Healthcare Solutions Using Machine Learning and Informatics, 2023
Jabar H. Yousif, Kashif Zia, Durgesh Srivastava
The topology can be classified mainly as single layer, multilayer, and self-organizing. A single-layer network is a simple topology of a neural network that has no in-between (hidden) layers. One layer of input nodes directly connects the input layer to the next layer (output layer). A multilayer network is a complex topology of a neural network that has one or more hidden layers. The input layers connect to the output layers not directly but through intermediate layers called hidden layers. The self-organizing topology is a competitive unsupervised machine learning that aims to reduce the higher-dimensional data to a low-dimensional (two-dimensional) structure.
Tracking, Modelling, and Understanding of Pandemic Outbreak with Artificial Intelligence and IoT
Published in Ram Shringar Raw, Vishal Jain, Sanjoy Das, Meenakshi Sharma, Pandemic Detection and Analysis Through Smart Computing Technologies, 2022
Sapna Kataria, Anjali Chaudhary, Neeta Sharma
Except ZigBee, the second most used technology in times of pandemic is Wi-Fi based on the features, strengths, and weaknesses. Wi-Fi also provides flexibility of choosing any topology for developing wireless networks; in the current situation, mesh topology could be proved a better option. Employing wireless communication techniques and sensing devices in smart cities for surveillance and predicting future events by evaluating the collected data will make us capable of detecting the next pandemic as well as providing enough information for handling the current pandemic situation. Researchers and scientists are gradually increasing the use of big data, ML, AI, and NLP to track coronavirus (CoVs), and to gain a further inclusive understanding of this virus. From starting months when COVID-19 start to spread out of China, researchers are working very hard trying to discover the physiology of coronavirus. Some questions are really need to be answered like why this virus has different effects on some people than others, what measures should be followed to decrease the spread, and what level this disease is likely to go next, what vaccine can cure or control the effects of coronavirus. All this research is based on the big data collected directly from corona patients.
Network pharmacology approach to investigate the multitarget mechanisms of Zhishi Rhubarb Soup on acute cerebral infarction
Published in Pharmaceutical Biology, 2022
Yuejia Shao, Yue Zhang, Rongrong Wu, Lurui Dou, Fengjiao Cao, Yuqing Yan, Yuming Tang, Chi Huang, Yang Zhao, Jinghua Zhang
Protein–protein interactions (PPIs) for each target were generated from the STRING database to help understand cell functions (Szklarczyk et al. 2017). The common targets of the drug and disease were input into the String database (https://string-db.org/cgi/input.pl) to construct the PPI network, and the species was set as ‘Homo sapiens’ to obtain the PPI network. The PPI network was imported into Cytoscape 3.6.0 software for a topology analysis and clustering analysis. Degree, betweenness centrality, average shortest path length and closeness centrality were used as the reference standards and sorted by degree. Genes with scores higher than the average were selected as key targets. The MCODE module was used to analyse gene clusters and screen core targets. Based on this information, the component-disease-target network was constructed using Cytoscape 3.8.0 software. The topology analysis of the network was performed, and the components were sorted by degree.
In-silico analysis of outflow graft implantation orientation and cerebral thromboembolism incidence for full LVAD support
Published in Computer Methods in Biomechanics and Biomedical Engineering, 2022
Ray Prather, Eduardo Divo, Alain Kassab, William DeCampli
Once each geometry is imported in the destination analysis software the mesh is generated. The meshing scheme provides for a polyhedral mesh with 3 prism layers at the wall boundary, with curvature and surface refinements to better capture the topology, having a maximum size of 1.5 mm and a minimum of 0.375 mm. Given a patient-specific geometry, the shape of a polyhedral cell is more suitable to retain high mesh quality and improving numerical stability without compromise in computational accuracy (Sosnowski et al. 2017), offering (1) improved convergence properties (2) resulting in approximately a 77% reduction in cell count and a 60% decrease in iterations as compared to tetrahedral meshes (Spiegel et al. 2011; Garimella et al. 2014). In our study, the average converged mesh accounts for nearly 0.6 M polyhedral cells, following a grid convergence study in which velocity and pressure are targeted as convergence markers with convergence established based on the Roache’s grid convergence index. The corresponding tetrahedral cell count from which this polyhedral mesh is constructed would be well over 1.6 M tetrahedral cells (Garimella et al. 2014; Prather et al. 2021).
Customized designs of short thumb orthoses using 3D hand parametric models
Published in Assistive Technology, 2022
Chih-Hsing Chu, I-Jan Wang, Jing-Ru Sun, Chien-Hsiou Liu
The cross-parameterization technique was then used to achieve the same mesh topology, under the condition that the numbers of meshes and mesh connectivity were the same (Kraevoy & Sheffer, 2004). First, one model was selected as the example mesh (source) and its mesh connectivity served as a template for adjusting the other models (target). A precondition was that the source and target models must have similarly shaped-features to preserve the structure of the two models by establishing correspondence between their features. The mesh model was divided into a number of patches by connecting those features along the shortest paths. The patches usually exhibited a smooth surface variation; thus, they could easily be locally parameterized using mean value parameterization (Kraevoy & Sheffer, 2004). The mesh topology of the source model was made identical to that of the target by mapping the two corresponding patches parametrically. As shown in Table 1, 19 feature points were selected to divide a mesh model into patches. The training data from an additional 119 hand data sets, except the target mesh, served as the source mesh for the sequence process, and one data set was selected for each process.