A comprehensive machine learning approach to prognose pulmonary disease from home

Abstract

This paper proposes a machine learning based prognosis for rehabilitating the COPD patients to be monitored from home in real time. Wearable sensor Technology (WST) is utilized to collect the physiological status of the pulmonary patient from home dynamically and communicated to the healthcare centre. The proposed approach applies a comprehensive predictive model employing a time series forecasting using condensed polynomial neural network with swarm intelligence. Discrete particle swarm optimization (DPSO) filters out the relevant neurons and continuous particle swarm optimization (CPSO) reduces the computational overheads. The time series prediction is further strengthened by using multimodal genetic algorithm. Control measures such as sensitivity, specificity and reliability are applied meticulously to validate the predicted state of the patient.