Cardiac disease prediction using AI algorithms with SelectKBest.

Area of Science:

• Cardiovascular disease research
• Artificial intelligence in healthcare
• Machine learning for medical diagnosis

Background:

• Atherosclerotic cardiovascular disease (ASCVD) is a leading global cause of mortality.
• Current AI research for heart disease prediction faces challenges with data diversity and model interpretability.
• Early detection of heart disease is crucial for reducing mortality rates.

Purpose of the Study:

• To propose an AI-driven cardiac disease prediction model using SelectKBest.
• To address limitations in current AI heart disease prediction methods.
• To improve the accuracy and interpretability of heart disease diagnosis.

Main Methods:

• Feature standardization, balancing, and selection using StandardScaler, SMOTE, and SelectKBest.
• Evaluation of various machine learning (SVM, KNN, DT, LR, AB, NB, RF, ET) and deep learning (LSTM variants, DNN) models.
• Utilized Alizadeh Sani, combined (Cleveland, Hungarian, Switzerland, Long Beach VA, Stalog), and Pakistan heart failure datasets.

Main Results:

• The proposed deep neural network (DNN) model with SelectKBest demonstrated promising heart disease prediction.
• Achieved unweighted accuracy rates of 99% on Alizadeh Sani, 98% on combined, and 97% on Pakistan datasets.
• Results were validated through tenfold cross-validation experiments.

Conclusions:

• The developed AI approach, particularly the DNN with SelectKBest, is effective for early heart disease diagnosis.
• The method offers a potential solution to current challenges in AI-based cardiovascular disease prediction.
• This approach can aid in the timely diagnosis and management of heart conditions.