Triage determination with gradient-based machine learning methods in health field

In this thesis, the integration of machine learning algorithms into decision support systems in the process of directing patients to the relevant branch physician according to their complaints was examined and a model was developed that will enable patients to be directed to the correct health department at the hospital entrance using patient data. This thesis examines the integration of machine learning algorithms into decision support systems in the process of directing patients to the relevant branch physician according to their complaints was examined and a model was developed that will enable patients to be directed to the correct health department at the hospital entrance using patient data. The study analyzed approximately one million anonymized patient records obtained from Özel Türkiye Hospital. Various machine learning algorithms, including Naive Bayes, Support Vector Machines, Artificial Neural Networks, Random Forest, AdaBoost, Gradient Boosting, CatBoost, LightGBM and XGBoost, were utilized and their performance were compared in this research. With the developed model, among all of the examined algorithms, XGBoost algorithm have demonstrated the highest performance 97% accuracy This finding shows that the model obtained with the XGBoost algorithm can be a very effective decision support model in supporting the patient's decision to go to the right department in healthcare services. During the data processing phase, patient complaints and other text data were transformed into numerical formats using TF-IDF methods. The model's success was evaluated using accuracy metrics, and its class-level performance was analyzed with a IX confusion matrix. The findings suggest significant potential for automating triage processes and providing support to healthcare professionals. This study aims to contribute to the literature by demonstrating how artificial intelligence and machine learning technologies can be employed as decision support systems in healthcare services.