<![CDATA[Anaemia is a widespread blood disorder characterized by a deficiency of red blood cells or hemoglobin, which can lead to severe health complications if not diagnosed and treated promptly. This research aims to develop a machine learning model to predict anaemia based on hemoglobin levels and image pixel distributions, leveraging a dataset from Kaggle. The dataset includes features such as percentages of red, green, and blue pixels in images and hemoglobin levels. We applied a Random Forest Classifier, a robust machine learning algorithm, and evaluated its performance using 5-fold cross-validation. The data pre-processing involved removing irrelevant columns, encoding categorical variables, and scaling numerical features. The model achieved a mean accuracy of 97.05%, precision of 97.02%, recall of 97.05%, and F1-score of 96.88%, indicating its high reliability in predicting anaemia. Visualizations such as Correlation Heatmaps, 3D PCA, Parallel Coordinates Plots, 3D t-SNE, and Violin Plots were used to understand feature relationships and distributions. These results underscore the potential of machine learning in providing a non-invasive, cost-effective diagnostic tool for anaemia, especially in resource-limited settings. Future research should address dataset imbalance and potential biases, explore additional features, and test other machine learning models to further enhance the predictive accuracy. This study contributes to the field of medical diagnostics by demonstrating the efficacy of integrating hemoglobin levels and image data for anaemia prediction, paving the way for improved early detection and treatment strategies.Keywords: Anaemia, Hemoglobin, Machine Learning, Random Forest.]]>
Copyrights © 2024
