Online ISSN: 2515-8260

Author : Arunjeet Kaur, Vikramjit S. Dhaliwal

Identification of Cardiovascular diseases (CVDs) using machine learning and analysis on risk factors for CVD

Vikramjit S. Dhaliwal Arunjeet Kaur

European Journal of Molecular & Clinical Medicine, 2020, Volume 7, Issue 10, Pages 3466-3476

Background: Cardiovascular diseases (CVDs) comprise a range of dis-eases impacting heart and blood vessels. The early detection of CVDs in high-risk patients can certainly help them to recover from fatal diseases and to resume their routine life. The healthcare industry has repository of medical data which is sufficient enough to make predictions of cardiovascular diseases using machine learning algorithms at early stages. In this research paper, the data has been collected from Fortis healthcare centres. The data pre-processing techniques have been used to remove noisy data, to handle missing values, fill-ing default values if applicable and then machine learning based algorithms have been applied to predict whether the patient is suffering from Cardiovas-cular diseases or not. We have also presented analysis on the risk attributes with respect to the sample population in this paper.
Results: The performance of the ML based algorithms have been evaluated using confusion matrix, ROC, F1 score, precision and recall scores. The potential attributes have been highlighted and discussed that contribute in the occurrence of CVDs amongst the patients. We have analyzed the risk-factors such as consumption of alcohol and tobacco, blood pressure levels, blood sugar levels, cholesterol measures, other medical history such as diabetes, family his-tory, age and weight of the diseased, and measure of per day physical activity for analyzing the impact of each risk factor on the CVD patients.
  Conclusions: In this paper, we have collected data of 70,000 patients from North-India region for the research study. We have attempted to apply classification algorithms for the identification of the patients suffering from cardiovascular diseases. It is observed that automated machine learning techniques produce accurate results and can assist the doctors to diagnose the diseased patients in a fast manner. We have also analyzed the risk-factors
related to CVDs such as diabetes, consumption of alcohol, smoking, obesity, blood pressure, cholesterol levels, physical activity and family history with chronic CVDs. We have also observed the impact of each respective risk-attribute with respect to cardiovascular diseases. This study will assist the physicians for identifying cardiovascular diseases in patients at early stage using proposed automated system and in analyzing the risk-factors with respect to CVDs. This study also provides insights into gender-wise analysis of cardiovascular cases.