Clustering based Contact Tracing Analysis and Prediction of SARS-CoV-2 Infections

Abstract

INTRODUCTION: Contact tracing is a method to track the victims, which have been infected from the host with any particular disease. Therefore, clustering based machine learning techniques can be employed for contact tracing. Contact tracing can be automated by using technology and thus helps us in producing much more accurate and efficient results.

OBJECTIVES: This work aims at finding usefulness of clustering techniques for contact tracing. Two different clustering techniques namely density-based clustering and partitioning-based clustering have been used to analyse corresponding results for COVID-19 infected cases. The dataset is generated from a mock data generator with certain assumptions.

RESULTS: The paper compares DBSCAN and K-means for contact tracing for COVID-19 Pandemic. The comparative analysis of two algorithms is presented.

CONCLUSION: The effectiveness of certain clustering algorithms in COVID-19 contact tracing is analysed. DBSCAN performs well for clustering tasks. This work only focuses on possible techniques useful for contact tracing and does not claim any medical accuracy.