Applying the Binary Logistic Regression Model for Prediction of the Ischemic Heart Disease

Chaithra N

Ms. Chaithra N, Assistant Professor, Division of Medical Statistics, Faculty of Life Sciences, JSS Academy of Higher Education & Research, Mysuru-15, Karnataka, India.

Manuscript received on 29 October 2020 | Revised Manuscript received on 07 November 2021 | Manuscript Accepted on 15 December 2020 | Manuscript published on 30 December 2020 | PP: 28-32 | Volume-1 Issue-1, December 2020 | Retrieval Number: A1002061120/2020©LSP

Open Access | Ethics and Policies | Indexing and Abstracting
© The Authors. Published by Lattice Science Publication (LSP). This is an open-access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)

Abstract: Ischemic Heart Disease (IHD) has been taxing the healthcare systems with a huge economic burden by being the major cause of deaths globally. In that, India contributes to about one-fifth of such deaths. Statistics suggest that for every 1000 people in a population of at least 7.5 people suffer from this condition with an average age ranging between 30 and 69 years. In the present study, we evaluated individual contributions of M-Mode two dimensional echocardiographic parameters to determine the presence of IHD in a large segment of the Indian population using a logistic model. This model can be a predictive step in assisting the junior cardiologists/echo technicians to diagnose IHD patients well in advance and model could be used in software applications for the medical field and estimating the impact of health interventions in developing countries. Methodology: A total of 7304 echo records were selected for performing the logistic regression from Electronic Health Records (EHRs) at the Department of Cardiology, JSS Hospital. The data set included 6191 patients without IHD and 1113 patients with IHD. The study included one dichotomous variable and fifteen explanatory variables that were taken during the transthoracic echocardiography examination.log-likelihood Statistic, Cox and Snell R2 , Nagelkerke R2 , Akaike Information Criterion are the tests to find the Goodness of fit for testing the fitness of the model. We used Likelihood ratio and Wald tests for testing the statistical significance of regression co-efficient. The classification table and Receiver Operating Characteristic (ROC) curve are the method to evaluate the predictive accuracy of the logistic regression. Results: This study is the first to apply a large sample from echo data, to determine how well a predictive model would perform based only upon patients M-mode echocardiography measurements without clinical risk factors or physical exam findings. All the variables exhibited statistically significant variation between IHD patients and non-IHD patients. The prevalence of IHD was significantly higher in men than in women. Our model was constructed by a Likelihood ratio forward method and Iteration History shows that estimation was terminated at iteration number 8 with 9 Steps (Model) because the parameter estimates did not change by more than 0.001. Conclusion: The present study estimates the efficiency of the logistic model to investigate the factors contributing significantly to enhancing the risk of IHD and the resulting model has a higher accuracy rate (96.7%), which makes it a handy tool for junior cardiologists and echo technicians to screen the patients who have a high probability of having the disease and transfer those patients to senior cardiologists for further clinical evaluation.

Keywords: Ischemic Heart Disease, Echocardiographic Data, Goodness of Fit, Prediction Model Indexes, Logistic Regression.
Scope of the Article: Ischemic Heart Disease