Identifying and evaluating clinical subtypes of Alzheimer’s disease in care electronic health records using unsupervised machine learning

Background Alzheimer’s disease (AD) is a highly heterogeneous disease with diverse trajectories and outcomes observed in clinical populations. Understanding this heterogeneity can enable better treatment, prognosis and disease management. Studies to date have mainly used imaging or cognition data and have been limited in terms of data breadth and sample size. Here we examine the clinical heterogeneity of Alzheimer's disease patients using electronic health records (EHR) to identify and characterise disease subgroups using multiple clustering methods, identifying clusters which are clinically actionable. Methods We identified AD patients in primary care EHR from the Clinical Practice Research Datalink (CPRD) using a previously validated rule-based phenotyping algorithm. We extracted and included a range of comorbidities, symptoms and demographic features as patient features. We evaluated four different clustering methods (k-means, kernel k-means, affinity propagation and latent class analysis) to cluster Alzheimer’s disease patients. We compared clusters on clinically relevant outcomes and evaluated each method using measures of cluster structure, stability, efficiency of outcome prediction and replicability in external data sets. Results We identified 7,913 AD patients, with a mean age of 82 and 66.2% female. We included 21 features in our analysis. We observed 5, 2, 5 and 6 clusters in k-means, kernel k-means, affinity propagation and latent class analysis respectively. K-means was found to produce the most consistent results based on four evaluative measures. We discovered a consistent cluster found in three of the four methods composed of predominantly female, younger disease onset (43% between ages 42–73) diagnosed with depression and anxiety, with a quicker rate of progression compared to the average across other clusters. Conclusion Each clustering approach produced substantially different clusters and K-Means performed the best out of the four methods based on the four evaluative criteria. However, the consistent appearance of one particular cluster across three of the four methods potentially suggests the presence of a distinct disease subtype that merits further exploration. Our study underlines the variability of the results obtained from different clustering approaches and the importance of systematically evaluating different approaches for identifying disease subtypes in complex EHR.

A Full Coverage Reconstruction Algorithm for the Electromagnetic Map Based on Affinity Propagation Clustering

Considering the reconstruction of electromagnetic maps without the prior information of electromagnetic propagation environment in the target area, a new algorithm based on affinity propagation clustering is proposed to complete the electromagnetic map reconstruction of the target area from points to surfaces and then from points and surfaces to a larger surface. Firstly, according to the actual situation, the target area is reasonably divided into grids. Electromagnetic data is sampled by distributed sensing nodes, and a certain number of sample points are selected for affinity propagation clustering to determine the locations of centers of sample points. Secondly, for the incomplete sample data, the Kriging algorithm is used to reconstruct the small circular electromagnetic maps. The class center is the center of the circle and the radius is certain. After that, the obtained small area electromagnetic map data and the data obtained from the sample points are used for domain mapping processing, and the electromagnetic data of a larger area of the target area is obtained. Finally, the overall electromagnetic map is reconstructed through data fusion. The simulation results show that the proposed algorithm is better than several interpolation algorithms. When sample points account for 0.1 of total data points, the RMSE of the result is less than 1.5.

Distributed Energy Efficient Clustering Routing Protocol for Wireless Sensor Networks Using Affinity Propagation And Fuzzy Logic

Organizing nodes into clusters and forwarding data to the Base Station (BS) in clustering routing protocols have been widely utilized to improve the energy efficiency, scalability and stability of Wireless Sensor Networks (WSN). Making decisions on how many clusters are formed, which nodes are selected as Cluster Heads (CHs) and who become the relay nodes significantly impact the network performance. Therefore, a Distributed clustering routing protocol combined Affinity Propagation (AP) with Fuzzy Logic called DAPFL is proposed in this paper, which considers not only energy efficiency but also energy balance to extend the network lifetime. In DAPFL, AP is firstly used to determine the number of clusters and select the best CHs simultaneously based on residual energy, distance between nodes. Then the optimal next-hop CHs are chosen by using fuzzy logic system with residual energy, data length and distance to BS as descriptors. Simulations in different scenarios are carried out to verify the effectiveness of DAPFL, and the results show that DAPFL exhibits the promising performance in terms of network energy consumption, standard deviation of residual energy, network throughput and lifetime, compared with the up-to-date distributed clustering routing protocols EEFUC, EEFRP, LEACH-AP and APSA.

A Cluster based Intelligent Method to Manage Load of Controllers in SDN-IoT Networks for Smart Cities

Software Defined Network (SDN) is a programmable network which separates the control logic-plane and hardware data-plane. The SDN centrally manages different Internet of Things (IoT) enabled smart devices like, actuators and sensors connected in the networks. Smart city infrastructure is an application of IoT network which purpose is to manage the city network without human interventions. To collect the real time data, such smart devices generate large amount of data and increasing the traffic in network. To maintain the quality of services (QoS) of smart city IoT networks, the SDN needs to deploy the multi-controllers. But the communication performance reduces due to unbalance load distribution on controllers. To balance the traffic load of controller an intelligent cluster based Grey Wolf Optimization Affinity Propagation (GWOAP) Algorithm is proposed when deploying the multiple controllers in SDN-IoT enabled smart city networks. The proposed algorithm is simulated and the experimental results able to calculates the minimum overall communication cost in comparison with Genetic Algorithm (GA), Particle Swarm Optimization (PSO) and Affinity Propagation (AP). The proposed GWOAP better balance the IoT enabled smart switches among clusters and node equalization is balanced for each controller in deployed topology. By using the proposed methodology, the traffic load of IoT enabled devices in smart city networks intelligently better balance among controllers.

The Effects of Munsell Neutral Grey Backgrounds on the Colour of Chrysoprase and the Application of AP Clustering to Chrysoprase Colour Grading

Chrysoprase is a popular gemstone with consumers because of its charming apple green colour but a scientific classification of its colour has not yet been achieved. In this research, we determined the most effective background of the Munsell Chart for chrysoprase colour grading under a 6504 K fluorescent lamp and applied an affinity propagation (AP) clustering algorithm to the colour grading of coloured gems for the first time. Forty gem-quality chrysoprase samples from Australia were studied using a UV-VIS spectrophotometer and Munsell neutral grey backgrounds. The results determined the effects of a Munsell neutral grey background on the observed colour. It was found that the Munsell N9.5 background was the most effective for colour grading in this case. The observed chrysoprase colours were classified into five groups: Fancy Light, Fancy, Fancy Intense, Fancy Deep and Fancy Dark. The feasibility of the colour grading scheme was verified using the colour difference formula DE2000.

K-affinity propagation clustering algorithm for the classification of part-time workers using the internet

There has been a significant increase in the number of part-time workers in the last 3 years. Data collected from sakernas BPS showed that the number of part-time workers was 125,443,748 in the second period of 2016. This number rapidly increased in 2017, 2018 and 2019 in the same period, by 128,062,746, 131,005,641, and 133,560,880 workers. Based on the increase in the last 3 years, East Java province has the highest number of part-time workers that use the internet. This research aims to determine the number of part-time workers that use the internet by using the k-affinity propagation (K-AP) clustering. This method is used to produce the optimal number of cluster points (exemplar) is the affinity propagation (AP). Three clusters were used to determine the sum of the smallest value ratio. The result showed that clusters 1, 2, and 3 have 3, 23, and 5 members in Bondowoso, Jombang, and Surabaya districts.

Improved Indoor Fingerprinting Localization Method Using Clustering Algorithm and Dynamic Compensation

Many indoor fingerprinting localization methods are based on signal-domain distances with large localization error and low stability. An improved fingerprinting localization method using a clustering algorithm and dynamic compensation was proposed. In the offline stage, the fingerprint database was built and clustered based on offline hybrid distance and an affinity propagation clustering algorithm. Furthermore, clusters were adjusted using transition regions and a given radius, as well as updating the corresponding position and fingerprint of the cluster centroid. In the online stage, the lost received signal strength (RSS) in the reference fingerprint would be dynamically compensated by using a minimum RSS value, rather than a fixed one. Online signal-domain distance was calculated for cluster identification based on RSS readings and compensated reference fingerprint. Then, K reference points with minimum online signal-domain distances were selected, and affinity propagation clustering was reused by position-domain distances to choose the position-concentrated sub-cluster for location estimation. Experimental results show that the proposed method outperforms state-of-the-art fingerprinting methods, with the mean error of 2.328 m, the root mean square error of 1.865 m and the maximum error of 10.722 m in a testbed of 3200 square meters. The improvement rates, in terms of accuracy and stability, are more than 21% and 13%, respectively.