THALASSEMIA PREDICTION USING HYBRID MACHINE LEARNING APPROACHES

ARTIFICIAL INTELLEGENCE AND MACHINE LEARNING

This study proposes a workflow for preparing and training a neural network model to classify medical data related to thalassemia diagnoses. Thalassemia is a genetic blood disorder that affects the production of hemoglobin, leading to anemia and other complications. Early diagnosis and treatment are essential for improving the quality of life and survival of patients. However, conventional methods of diagnosis are often invasive, expensive, and timeconsuming. In this paper, we propose a novel workflow for preparing and training a neural network model to classify medical data related to thalassemia diagnoses. Our workflow consists of four main steps: data loading, cleaning, and processing; data splitting and oversampling; feature scaling; and model definition, compilation, and training. We use TensorFlow to implement our neural network model, which has several dense layers with ReLU activation functions, dropout regularization, and a SoftMax output layer. We evaluate our model on a publicly available dataset of 420 patients with different types of thalassemia. Our results show that our model achieves an accuracy of 96.67% on the test set, outperforming previous methods based on logistic regression, decision tree, and support vector machine. We conclude that our workflow provides a comprehensive pipeline for data preprocessing and training a neural network model for thalassemia diagnosis classification, addressing class imbalance and ensuring proper data preparation.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.