Showing papers by "Kumar Dharmarajan published in 2022"
13 Oct 2022
TL;DR: In this article , the results of the Random Forest (RF) classifier along with different wrapper methods were used to predict diabetes in an early stage and the Hyperparameter tuning method showed the more accuracy when compared to Random Forest classifier.
Abstract: Diabetics Mellitus is a common disease in the real world. It creates major health issues. Predicting diabetes in an early stage can save a human life. Health care contains huge data to access data with machine learning techniques. Diabetes occurs due to a lack of physical work and food habits. This research study predicts diabetes by using Pima Indian Dataset. This paper focuses on the results of the Random Forest (RF) classifier along with different wrapper methods. The Hyperparameter tuning method shows the more accuracy when compared to Random Forest (RF) classifier.
20 Jul 2022
TL;DR: The system's goal is to provide a comprehensive monitoring and analysis platform that provides a one-stop solution for patient monitoring systems that utilizing Gaussian Expectation Maximization (GEM) enabled Deep Transnet (GEDT) algorithm, for effective detection of patient abnormality.
Abstract: A patient monitoring system is necessary to monitor patient data remotely and precisely, at all times. It is necessary to give a high degree of care to individuals who are at high risk following post-surgical situations. Recent patient monitoring systems are being used for further diagnostic analyses and treatments that are open sourced and might benefit people all around the world. The system's goal is to provide a comprehensive monitoring and analysis platform that provides a one-stop solution for patient monitoring systems. The major goal is to develop a model utilizing Gaussian Expectation Maximization (GEM) enabled Deep Transnet (GEDT) algorithm, for effective detection of patient abnormality. Various physiological parameters are considered for analysis. The data acquired for ECG analysis is compared to the MIT BIH dataset from Physio Net. The standard values for heart rate, temperature, pressure, and oxygen saturation are used. Real-time testing is suggested, with volunteers helping to evaluate the gear. Only a subset of real-time values is used to collect training data. 70% are utilized for training, 15% for testing, and 15% for validation.