Abstract:
Dengue fever, a common mosquito-borne viral infection, often goes undiagnosed until
severe symptoms appear. Early detection is important for effective management and
prevention of complications. This study explores the potential of machine learning (ML)
and artificial intelligence (AI) to develop predictive models for dengue fever and severity
assessment. This study used secondary datasets including patient data including medical
history, environmental factors, symptoms and possible genetic predisposition to train
various machine learning algorithms such as CNN, Linear Regression, Random Forest and
K-Nearest Neighbors (KNN). This method involves data collection, preprocessing, feature
selection, model training, cross-validation, and hyperparameter tuning to optimize
performance and avoid overfitting. Models are evaluated based on metrics such as accuracy
and AUC-ROC, including attribute importance scores and shape quality analysis. The
dataset consisted of 531 patient records with 20 attributes each, which ensured data quality
through preprocessing steps such as handling missing values and encoding categorical
variables. A high-performance computer, cloud resources, powerful hardware and software
including Python, TensorFlow, PyTorch and Skit-Learn are essential for implementation.
Effective project management and financial analysis ensure research success and
sustainability. Studies have shown that linear regression and random forest models achieve
the highest accuracy (99%), followed by CNN (97%) and KNN (54%). These findings
underscore the potential of machine learning to enhance dengue prediction, improve patient
outcomes, and inform public health strategies.
