Heart Failure Prediction System Using Machine Learning (HFPS using ML)

Abstract

There are many death cases related to heart and their counting is increasing day by day. Heart Failure, alternatively known as cardiovascular disease, encases various conditions that impact the heart and is the primary basis of death worldwide over the span of the past few decades. It associates many risk factors in heart Failure and a need of the time to get accurate, reliable, and sensible approaches to make an early diagnosis. To deal with the problem there is essential need of prediction system for awareness about Heart Failure. Diagnosis and prediction of heart related diseases requires more perfection and correctness because a little mistake can cause death of the person. Machine learning is subset of Artificial intelligence. It provides large support in any kind of event. I am using machine learning, Data Analysis with Python, Data Science, Python, Flask Framework, Flask my SQL bd and JavaScript. First, I picked data set from Kaggle website. This data set contains 13 features that can be used to predict mortality by heart failure. Data Set Name: Heart Failure prediction. Data Set Link: https://www.kaggle.com/andrewmvd/heart-failure-clinical-data Data Set Features: (age, anemia, creatinine phosphokinase, diabetes, ejection fraction, high blood pressure, platelets, serum creatinine, serum sodium, sex, smoking, time, Death Event). Jupyter notebook (Anaconda Environment) is used for data analysis, visualization, and model building. First, I import data set in my jupyter notebook. Then I perform data cleaning operations on my data set after that I perform exploratory data analysis with visualization on my data set and bring insights out of it. For Visualization, I used different libraries named as (Matplotlib, seaborn, Plotly express). I used some Panda SQL queries to fetch some data from data frame to perform analysis. Then I find correlation between each feature of my data set and build a heatmap of correlation. Dividing Data Set: Divide my data set into training and testing sections. 70% for training the model and 30% for testing the model. Model Building: According to my data set Heart Failure Prediction. It is a classification problem So; I have applied all types of machine learning classification Models. Random Forest Classifier got the best accuracy. Deployment: First, I have created a separate environment in Anaconda. Then I have installed all libraries which I am going to import in jupyter notebook. I even install the Jupyter Notebook Kernel in my environment. Then I build a ML model in jupyter notebook. Then I dumped my ML model into PKL file. Created a requirement text file. I have deployed my Random Forest Classifier Model using Flask Framework. I have connected my back-end model with front-end using Python and Flask Framework. I have stored my data into Database using XAMPP server (PHP my Admin) and Flask my SQL DB. I used Spyder editor for python file which connect back-end with my front-end. I used Anaconda Prompt to execute the whole system by using command (python app.py). I used HTML, CSS, and JavaScript as a front-end. I used Visual Studio Code editor for HTML, CSS, and JavaScript. Heart Failure Prediction System: It is a web-based system. In this system user must first register to the system and then login to the system. In the system user can enter its patient history to predict the heart failure. Patient History must contain features (age, anemia, creatinine phosphokinase, diabetes, ejection fraction, high blood pressure, platelets, serum creatinine, serum sodium, sex, smoking, time). After that user can contact doctor by sending him a message through website and visualize Heart Analysis and Model Building.