On Design and Deployment of ML Model for Cardiac Disease Prediction

Abstract

Cardiovascular disease is among the disorders that pose the greatest hazard to life. Its high death rate is responsible for around 17 million fatalities globally. Early diagnosis helps to treat the illness at the appropriate time to prevent death. Numerous machine learning and deep learning methods can be used to analyze the presence or absence of the disease. This review paper discusses the entire process of creating an intelligent model. In this study, logistic regression techniques are utilized to classify heart illness using the UCI dataset. Pre-processing the data by decluttering the dataset, finding missing values, and selecting features by correlating each feature with the target value were done to improve the model's performance. The features that showed a strong positive correlation were selected. Next, classification is carried out by dividing the dataset in an 80:20 split-up ratio between training and testing data. By utilizing the Sklearn framework and Python scripting, the proposed model achieved an accuracy of 83.6%. Later the model is prepared with no code AI platform (Akkio) with production quality mode and obtain an efficiency of 86.9%.  The effectiveness of the proposed model was thoroughly gratifying and was competent to predict evidence of having heart disease in a certain individual which exhibited quality precision 77%(87% in Akkio) and recall 89%(91% in Akkio). This prediction system for cardiac disease increases access to care while lowering costs. This research report provides us with important information that can assist us in determining whether or not the patient has a cardiac problem. The assessment of the suggested work was completed in Google Colab and afterwards in Akkio. The joblib library in Colab and an Akkio Google sheet were used for deployment.