Dynamic Distributed Clustering Approach Directed to Patient-Centric Healthcare System

Healthcare data are often obtained from different healthcare providers for improved healthcare services. Healthcare data collection and maintenance became a concern owing to data issues including high-dimensionality irregularity, and sparsity. Several scholars focused on these topics and offered several effective and flexible healthcare solutions. A few other machine learning techniques, such as clustering, are used to evaluate healthcare data. Clustering is one of the well-known and employed data mining methods in health data analysis. However, a major challenge remains on how to scale up and pace up clustering algorithms with a limited loss to clustering consistency. In this paper, a framework is proposed that utilizing an updated hierarchical distributed clustering in biomedical engineering for a large data environment. The complex essence of the method arises from not providing the number of right clusters in advance. The suggested methodology operates through two key phases: the first step constructs local clusters dependent on their portion of the entire dataset; this phase takes maximum advantage of the Spark Framework-based task parallelism paradigm; whereas the second phase aggregates local clusters to render the final clusters lightweight and reliable. The framework can help both doctors and patients. Experimental findings demonstrate that the method is flexible and provides high-quality for Healthcare Cost and Utilization Project (HCUP) dataset.