Key result
An ensemble machine learning model combining 10 classification algorithms predicted heart failure with a test accuracy of 85.2% and a test recall of 87.50%.
Why the study?
Cardiovascular diseases are the leading cause of global mortality, requiring early warning systems so precautionary measures can be taken well before time.
Does an ensemble machine learning model accurately predict heart failure using Framingham Heart study data?
Does an ensemble machine learning model accurately predict heart failure using Framingham Heart study data?
An ensemble machine learning model combining 10 algorithms can predict heart failure with 85.2% accuracy using Framingham Heart study data.
Should not yet alter clinical practice; leaves open ensemble ML utility for heart failure prediction in population datasets.
According to WHO, cardiovascular diseases are the number 1 cause of death globally. It causes the death of more than 12 million people every year worldwide. The main issue that needs to be resolved is that one should be warned well before time to take precautionary measures. Thus, in this paper, we propose a radical solution based on ensemble learning combining 10 different classification algorithms namely AdaBoost, CatBoost, Decision Trees, KNN, Logistic regression, Light GBM, Gaussian Naïve Bayes, Random Forest, SVM and XGBoost. This ensemble model was able to achieve a test accuracy of 85.2% and test recall of 87.50%. We used the data collected from the Framingham Heart study which includes 15 attributes and 4200+ records. Moreover, we performed extensive Exploratory Data Analysis to understand the importance of each attribute in causing heart failure.
No takes yet. Share an insight, caveat, or question.
Agrawal et al. (2021) studied Heart failure (n=4,200). Ensemble machine learning model was evaluated on Heart failure prediction (Accuracy 85.2%, Recall 87.50%). An ensemble machine learning model combining 10 classification algorithms predicted heart failure with a test accuracy of 85.2% and a test recall of 87.50%.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: