Artificial Intelligence-Based Early Detection of Dengue Using CBC Data

Abstract

A method for early dengue identification utilizing complete blood count (CBC) data that is based on artificial intelligence is provided. We use a number of feature selection methods to find the most important features, such as Pearson Correlation, ExtraTree, Chi-Square (Chi2), Recursive Feature Elimination (RFE) using Random Forest, SelectKBest, and more. Logistic Regression, Support Vector Machine (SVM), Naive Bayes, Random Forest, AdaBoost, XGBoost, Multi-Layer Perceptron (MLP), LightGBM, and ensemble methods such as a Stacking Classifier (XGB + LR + MLP with LightGBM) and a Voting Classifier (Boosted Decision Tree + ExtraTree) are utilized. Also included are various deep learning and machine learning algorithms. Some of the deep learning architectures used include ANNs, CNNs, GRUs, Bi-LSTMs, FNNs, Transformers, and hybrid models like CNN + LSTM. The Voting Classifier achieved an F1 Score and accuracy of 98% by combining predictions from individual models using ensemble approaches, which increases resilience and accuracy. Another factor that enhances the system's effectiveness is the utilization of hybrid models, specifically CNN + LSTM. The method is built to be user-friendly and secure, with a Flask-based interface that allows authentication. It has a high predictive accuracy and is easy to use.