Optimized Predictive Models in Health Care Using Machine Learning

9Revolutionizing Pneumonia Diagnosis and Prediction Through Deep Neural Networks

Abhishek Bhola¹ and Monali Gulhane²*

¹Department of CSE, Koneru Lakshmaiah Education Foundation, A.P., India

²Symbiosis Institute of Technology (SIT) Nagpur, Symbiosis International (Deemed University) (SIU), Pune, Maharashtra, India

Abstract

Pneumonia is a common and potentially fatal respiratory illness that can be difficult to diagnose accurately. In the digital era, the deep learning-based module has shown great promise in predicting diseases from medical images. In this work, we propose using a ResNet50-based advance neural network to predict pneumonia from given pictures. The ResNet50 model was trained on a standard dataset of 5,856 images, including 2,538 typical cases and 3,318 pneumonia cases. The trained model achieved an average accuracy of 97% on a commonly available dataset. ResNet50 can be an effective tool for predicting pneumonia from chest X-ray images, with potential applications in clinical settings.

Keywords: Pneumonia diseases, ResNet50, X-ray, histogram

9.1 Introduction

Pneumonia is a respiratory illness that can significantly impact daily life. It can cause various symptoms interfering with daily activities, such as coughing, fatigue, shortness of breath, chest pain, and fever [1]. These symptoms can make working, exercising, or daily tasks difficult, leading to missed school or work days. Pneumonia can lead to hospitalization or death for older adults or those with underlying ...