An adaptive ml model for covid-19 diagnosis in a smart hospital environment

Abstract

Abstract The Covid-19 pandemic has since 2019 caused worldwide socio-economic unrest, fear, and panic among all individuals, nations, races, and continents thereby forcing governments to introduce This necessitated the integration of predictive models into the healthcare support system for effective diagnostic and prediction of Covid-19. The need for modeling existing models to provide satisfactory models, give a clear understanding of the existing model contribution and further improve these models has become significantly necessary since the lack of confidence in predictive health systems would slow the early diagnostics and detection of Covid-19 in the smart health environment and in the world at large. This study is an adaptive study to experiment with existing models to ascertain and confirm the effectiveness of the model and further attempt to improve the performance of existing models to give healthcare system designers the edge to build and increase the effectiveness of Covid-19 predictive systems in a smart hospital environment. The study model 3 separate Arthurs conducted to produce a real-time intelligent Covid-19 predictive model using dataset from the Kaggle dataset repository, which can be implemented in smart hospitals to help eliminate physical contact treatment by healthcare professionals, prevent long queues which lead to long waiting at the healthcare facility. The experimental result confirms the efficacy of the models proposed by the authors and a further moderation to implement the stacking ensemble classifier techniques outperformed the modeling studies by producing an accuracy result of 96.00% and scoring an error rate of 0.040 representing 4%, having 1% higher than previous studies which used random forest with an accuracy of 95%. The study, therefore, confirms and recommends the models by the previous Arthurs as effective predictive models for diagnosing and predicting COVID-19 in a smart hospital environment.