In developing countries, the management of chronic diseases in the elderly is challenging because of limited medical resources and a large population. Hospitals cannot quickly provide treatment for chronic diseases for the elderly. This article discusses a mobile-medical-based system that uses effective data decision-making and wireless network communication to assist the elderly in implementing remote medical services and proposes a deep learning model-combined sparse autoencoder (CSAE) in the decision-making module. The model uses sparse autoencoders to process patient detection data and monitoring data collaboratively to achieve multisource data association and fusion. The prediction probability of diseases is obtained through deep neural networks and a classifier, and the best disease prediction sequence is output to realize the prediction and early warning of chronic diseases. In the experimental part, we compare CSAE with the other four machine learning classification algorithms in terms of accuracy, sensitivity, and specificity. The results prove that CSAE is superior to other algorithms in the single-disease prediction, with the accuracy and specificity exceeding 85%, and the sensitivity not lower than 75%. When predicting multiple diseases, our proposed algorithm's performance is the most stable, with an accuracy higher than other algorithms by more than 10%.
1 December 2021 in Expert Systems with Applications
By shifting the requested content to the edge in the Internet of Vehicles (IoV), edge caching is expected to be an effective solution to satisfy the low latency and high-reliability requirements of IoV users for multimedia services. However, the edge node’s coverage area and storage space are limited. Moreover, since vehicles have high mobility and in-vehicle multimedia applications require sequential delivery for contents, we need to address two main issues: 1) how to optimize the proactive content caching decision (i.e., the placement of cached content chunks) among edge nodes (ENs) to provide better Quality of Services (QoS) for IoV users and 2) how to ensure that vehicles can download the required contents sequentially to improve Quality of Experience (QoE). In this article, we propose a mobility-aware proactive edge caching scheme (MSTPS), where the spatial and temporal prediction of vehicles are taken into account for content deployment and scheduling. Specifically, we optimize the caching decision based on predicting the vehicle’s driving trajectory and travel preference. The scheme learns the vehicle’s travel preferences to cope with mobility uncertainty by combining users with similar travel patterns. Meanwhile, the proposed scheme can support the sequential downloading of content chunks. Furthermore, in order to deal with the dynamic characteristics and unpredictable challenges of the IoV, we design a system recovery strategy, which can avoid the degradation of the proposed scheme due to the failure of prediction. Finally, by using real mobility data sets and scenarios, we explore the impact of the number of ENs deployed in advance for each vehicle’s request when the cache needs to be updated on system performance. In addition, we evaluate the effectiveness of the proposed scheme. Our proposed scheme can achieve the best cache hit ratio and decrease caching costs compared to the existing mobility-aware in-order caching schemes.
18 September 2020 in IEEE Journal of Biomedical and Health Informatics
Objective: Accurate segmentation and partitioning of lesions in PET images provide computer-aided procedures and doctors with parameters for tumour diagnosis, staging and prognosis. Currently, PET segmentation and lesion partitioning are manually measured by radiologists, which is time consuming and laborious, and tedious manual procedures might lead to inaccurate measurement results. Therefore, we designed a new automatic multiprocessing scheme for PET image pre-screening, noise reduction, segmentation and lesion partitioning in this study. PET image pre-screening can reduce the time cost of noise reduction, segmentation and lesion partitioning methods, and denoising can enhance both quantitative metrics and visual quality for better segmentation accuracy. For pre-screening, we propose a new differential activation filter (DAF) to screen the lesion images from whole-body scanning. For noise reduction, neural network inverse (NN inverse) as the inverse transformation of generalized Anscombe transformation (GAT), which does not depend on the distribution of residual noise, was presented to improve the SNR of images. For segmentation and lesion partitioning, definition density peak clustering (DDPC) was proposed to realize instance segmentation of lesion and normal tissue with unsupervised images, which helped reduce the cost of density calculation and completely deleted the cluster halo. The experimental results of clinical data demonstrate that our proposed methods have good results and better performance in noise reduction, segmentation and lesion partitioning compared with state-of-the-art methods.
in social opportunistic networks, mobile devices can be regarded as socialization nodes. Furthermore, they carry and stin social opportunistic networks, mobile devices can be regarded as socialization nodes. Furthermore, they carry and store useful information. Mobile devices can select destination nodes and deliver messages through social opportunistic networks because messages can be securely and conveniently stored, carried, and transmitted with nodes. according to such characteristics, communities can be established. However, many communities may deliver messages only depending on one or two nodes. If those nodes are not enough cache and over-flooding, data transmission in communities may wait for a long time. To find many cooperation neighbor nodes can share much useful information. This work proposes a method that characteristic interests with neighbors are selected or pushed service by users. According to establish interest characteristic probability predicted method, interest neighbors can be selected and receive or send information. Compared with the social opportunistic network algorithms, the new research method achieves better results in accuracy selected and delivery ratio, transmission delay, and routing overhead. According to the simulation experiments, the average delivery ratio of new research method is 0.88, which is 40% higher than that of Epidemic algorithm; the end-to-end delay reduces 38% with Spray and wait algorithm.