Abstract

A particular kind of spinal condition called kyphosis is defined by an aberrant curvature of the top vertebrae, which makes the back rounded or stooped. For timely intervention and treatment planning, early identification and case classification are essential in cases of kyphosis. This study proposes a hybrid machine-learning approach for kyphosis sickness classification and prediction that combines Random Forest (RF) and Gradient Boosting (GB). Clinical features like age, the number of afflicted vertebrae, and the curvature angle are used in this work. Preprocessing involves splitting the dataset into two training and testing sets and fixing any missing values. Different RF and GB classifiers were trained independently on the training dataset. The hybrid model is created by the suggested way by combining the predictions from both classifiers using a majority vote strategy. The final prediction for each occurrence is determined by the class that is predicted the most frequently. Analyze the hybrid model's effectiveness using the F1-score, accuracy, precision, and recall metrics. The results demonstrate that the hybrid model outperforms individual classifiers in classification accuracy, attaining 86%. Using the advantages of both algorithms, the hybrid technology provides a more robust and accurate classification model for kyphosis illness by combining their predictive capabilities. This study advances the field of medical diagnostics by demonstrating the effectiveness of combining machine literacy algorithms to improve disease prognosis. This study employs a hybrid mode that can potentially improve treatment outcomes for patients with kyphosis by aiding physicians in making accurate and prompt decisions.