Predicting cervical spine disease with explainable artificial intelligence

Cervical spine straightening disease is a condition that occurs due to the loss of natural curvatures in the cervical spine, leading to restricted movement and pain. It negatively affects the course of life with its characteristics. In this thesis, grading of neck straightness is done with the help of explainable artificial intelligence (XAI) methods. The aim of the research is to present the disease diagnosis and grading process in a transparent and comprehensible manner. In the study, machine learning algorithms such as Random Forest and Decision Tree were used and hyperparameter optimization was used and improved with these models. The attributes in the data set have been carefully selected to enhance the model's performance in determining the severity of the disease. Metrics such as accuracy, precision, sensitivity and F1 score were used and compared in the performance of the models. When we look at the accuracy rate, the Random Forest model has proven the highest performance with 88% and balanced classification ability. It was chosen in clinical decision processes due to its simplicity and ease of explanation in the Decision Tree algorithm. As a result, it has been observed that XAI methods have a significant effect on the detection and grading of cervical straightening disease. While these investigations strengthen medical decision times, experts bring reliable and understandable information, and introduce new solutions in the field of healthcare.