Explainable AI-based Alzheimer's prediction and management using multimodal data.

BACKGROUND: According to the World Health Organization (WHO), dementia is the seventh leading reason of death among all illnesses and one of the leading causes of disability among the world's elderly people. Day by day the number of Alzheimer's patients is rising. Considering the increasing rate and the dangers, Alzheimer's disease should be diagnosed carefully. Machine learning is a potential technique for Alzheimer's diagnosis but general users do not trust machine learning models due to the black-box nature. Even, some of those models do not provide the best performance because of using only neuroimaging data. OBJECTIVE: To solve these issues, this paper proposes a novel explainable Alzheimer's disease prediction model using a multimodal dataset. This approach performs a data-level fusion using clinical data, MRI segmentation data, and psychological data. However, currently, there is very little understanding of multimodal five-class classification of Alzheimer's disease. METHOD: For predicting five class classifications, 9 most popular Machine Learning models are used. These models are Random Forest (RF), Logistic Regression (LR), Decision Tree (DT), Multi-Layer Perceptron (MLP), K-Nearest Neighbor (KNN), Gradient Boosting (GB), Adaptive Boosting (AdaB), Support Vector Machine (SVM), and Naive Bayes (NB). Among these models RF has scored the highest value. Besides for explainability, SHapley Additive exPlanation (SHAP) is used in this research work. RESULTS AND CONCLUSIONS: The performance evaluation demonstrates that the RF classifier has a 10-fold cross-validation accuracy of 98.81% for predicting Alzheimer's disease, cognitively normal, non-Alzheimer's dementia, uncertain dementia, and others. In addition, the study utilized Explainable Artificial Intelligence based on the SHAP model and analyzed the causes of prediction. To the best of our knowledge, we are the first to present this multimodal (Clinical, Psychological, and MRI segmentation data) five-class classification of Alzheimer's disease using Open Access Series of Imaging Studies (OASIS-3) dataset. Besides, a novel Alzheimer's patient management architecture is also proposed in this work.

Silicene Quantum Capacitance Dependent Frequency Readout to a Label-Free Detection of DNA Hybridization- A Simulation Analysis.

The use of deoxyribonucleic acid (DNA) hybridization to detect disease-related gene expression is a valuable diagnostic tool. An ion-sensitive field-effect transistor (ISFET) with a graphene layer has been utilized for detecting DNA hybridization. Silicene is a two-dimensional silicon allotrope with structural properties similar to graphene. Thus, it has recently experienced intensive scientific research interest due to its unique electrical, mechanical, and sensing characteristics. In this paper, we proposed an ISFET structure with silicene and electrolyte layers for the label-free detection of DNA hybridization. When DNA hybridization occurs, it changes the ion concentration in the surface layer of the silicene and the pH level of the electrolyte solution. The process also changes the quantum capacitance of the silicene layer and the electrical properties of the ISFET device. The quantum capacitance and the corresponding resonant frequency readout of the silicene and graphene are compared. The performance evaluation found that the changes in quantum capacitance, resonant frequency, and tuning ratio indicate that the sensitivity of silicene is much more effective than graphene.

CoAR: Congestion-Aware Routing Protocol for Low Power and Lossy Networks for IoT Applications.

The IPv6 routing protocol for low power and lossy networks (RPL) was designed to satisfy the requirements of a wide range of Internet of Things (IoT) applications, including industrial and environmental monitoring. In most scenarios, different from an ordinary environment, the industrial monitoring system under emergency scenarios needs to not only periodically collect the information from the sensing region, but also respond rapidly to some unusual situations. In the monitoring system, particularly when an event occurs in the sensing region, a surge of data generated by the sensors may lead to congestion at parent node as data packets converge towards the root. Congestion problem degrades the network performance that has an impact on quality of service. To resolve this problem, we propose a congestion-aware routing protocol (CoAR) which utilizes the selection of an alternative parent to alleviate the congestion in the network. The proposed mechanism uses a multi-criteria decision-making approach to select the best alternative parent node within the congestion by combining the multiple routing metrics. Moreover, the neighborhood index is used as the tie-breaking metric during the parent selection process when the routing score is equal. In order to determine the congestion, CoAR adopts the adaptive congestion detection mechanism based on the current queue occupancy and observation of present and past traffic trends. The proposed protocol has been tested and evaluated in different scenarios in comparison with ECRM and RPL. The simulation results show that CoAR is capable of dealing successfully with congestion in LLNs while preserving the required characteristics of the IoT applications.

An Energy Centric Cluster-Based Routing Protocol for Wireless Sensor Networks.

Clustering is an effective way to prolong the lifetime of a wireless sensor network (WSN). The common approach is to elect cluster heads to take routing and controlling duty, and to periodically rotate each cluster head's role to distribute energy consumption among nodes. However, a significant amount of energy dissipates due to control messages overhead, which results in a shorter network lifetime. This paper proposes an energy-centric cluster-based routing mechanism in WSNs. To begin with, cluster heads are elected based on the higher ranks of the nodes. The rank is defined by residual energy and average distance from the member nodes. With the role of data aggregation and data forwarding, a cluster head acts as a caretaker for cluster-head election in the next round, where the ranks' information are piggybacked along with the local data sending during intra-cluster communication. This reduces the number of control messages for the cluster-head election as well as the cluster formation in detail. Simulation results show that our proposed protocol saves the energy consumption among nodes and achieves a significant improvement in the network lifetime.

An Eccentricity Based Data Routing Protocol with Uniform Node Distribution in 3D WSN.

Due to nonuniform node distribution, the energy consumption of nodes are imbalanced in clustering-based wireless sensor networks (WSNs). It might have more impact when nodes are deployed in a three-dimensional (3D) environment. In this regard, we propose the eccentricity based data routing (EDR) protocol in a 3D WSN with uniform node distribution. It includes network partitions called 3D subspaces/clusters of equal member nodes, an energy-efficient routing centroid (RC) nodes election and data routing algorithm. The RC nodes election conducts in a quasi-static nature until a certain period unlike the periodic cluster heads election of typical clustering-based routing. It not only reduces the energy consumption of nodes during the election phase, but also in intra-communication. At the same time, the routing algorithm selects a forwarding node in such a way that balances the energy consumption among RC nodes and reduces the number of hops towards the sink. The simulation results validate and ensure the performance supremacy of the EDR protocol compared to existing protocols in terms of various metrics such as steady state and network lifetime in particular. Meanwhile, the results show the EDR is more robust in uniform node distribution compared to nonuniform.