Machine learning analysis of colchicine treatment in Familial Mediterranean Fever

In this study, the detection of Familial Mediterranean Fever (FMF) disease is challenging and can be achieved after complex processes and procedures. Many patients, despite exhibiting symptoms of the disease for an extended period, cannot receive a diagnosis. The main reason for this is that many internal medicine doctors harbor suspicions of different diseases when presented with patients showing symptoms of the disease. This is because Familial Mediterranean Fever has not been commonly encountered among diseases until today. This study encompasses research on the methods and application techniques that can be employed for the detection of Familial Mediterranean Fever disease. The learning method is utilized, and machine learning methods are employed initially to identify the crucial criteria present in the disease. Subsequently, it is envisioned that the patient will provide information to the doctor about the likelihood of having Familial Mediterranean Fever. However, due to the inadequacy of the dataset, the study has been revised to investigate the response or lack of response of patients to colchicine, one of the most important drugs used in the treatment of the disease. In this context, a customized dataset containing demographic information, clinical symptoms, and genetic variants related to Familial Mediterranean Fever was analyzed. In the feature selection process, 8 critical features were identified using the SelectKBest algorithm based on ANOVA F-value. In the analysis conducted on these features, the Logistic Regression model yielded noteworthy results. This model determined an average accuracy rate of 85.42%, average precision of 85.81%, average recall of 99.17%, and an average F1 score of 92.00. These results provide the opportunity to develop a model that operates with high accuracy and precision in the diagnosis of Familial Mediterranean Fever. These findings can contribute significantly to the early diagnosis of FMF and the development of personalized treatment strategies. The study establishes a robust foundation for broader research in this field and can be considered a significant step in the field.