Disease Prediction using Chest X-ray Images in Serverless Data pipeline Framework

Abstract

Serverless architecture is a rapidly emerging trend in the field of cloud computing that promises increased flexibility, scalability, and cost-effectiveness compared to traditional server-based approaches. Leveraging machines to automatically analyze and predict the disease using image data such as chest X-ray images is becoming a challenging task for various contemporary applications. Serverless computing is a cloud computing execution model that provides and manages resources based on the requirements of the users/applications. Besides that, modern data-intensive applications require the power to manage the flow of data between different components in a serverless platform. Motivated by that, in this paper, we develop a new serverless data pipeline framework for predicting disease using chest Xray images. The system utilizes Deep Learning (DL)-based image classification models hosted on Google serverless platform for COVID-19 diagnosis. For disease prediction, we incorporate a transfer learning technique over three popular DL models, namely VGG-16, DenseNet121, and ResNet50. The experimental analysis demonstrates that the proposed serverless data pipeline framework achieves high accuracy, reliability, and speed during COVID-19 disease diagnosis. As per the simulation results, the VGG-16 model outperforms the existing DL models and achieves 97.66% accuracy.