An Adaptive Neuro-Fuzzy Inference System to predict Acute Lymphocytic Leukemia prognosis in children

Background and aims: Leukemia is the most common cancer in children, and the most commonleukemia is Acute Lymphocytic Leukemia (ALL). Prognosis can play an important role intimely treatment and reduction of subsequent complications. Therefore, the aim of this study wasto develop an Adaptive Neuro-Fuzzy Inference System to predict Acute Lymphocytic Leukemiaprognosis in children.Method: This study was conducted in ۲۰۲۲ and consisted of two main phases. In the first phase,the prognostic parameters were determined based on the literature review and by investigatingspecialists’ perspectives by using a questionnaire. The face and content validity of the questionnairewas approved by the experts in the field of medicine. The reliability of the questionnairewas calculated by using the test-retest method. In the second phase, this system was designed andimplemented by using ANFIS toolbox in MATLAB. Then, system performance was evaluated.An online dataset was used to test the sensitivity, specificity, and accuracy of the system.Results: According to the physicians, the most important parameters were age, sex, white bloodcell count at diagnosis, rate of decrease in cancer blood cell count after initial treatment, whetherthe source of cancer blood cells is B lymphocytes or T lymphocytes, specific changes in lymphocytechromosomes, Leukemia was found in the brain and spinal cord, the child had Downsyndrome, the duration of the cancer, and the type of treatment. The system was designed and theevaluation results showed that the values of sensitivity (۸۷%), specificity (۹۱%), accuracy (۸۵%)were in good condition.Conclusion: The results of the system were largely similar to the patients’ records and showedthat the designed system can be used to help physicians to predict prognosis of ALL patients andto improve the quality of care services and life for these patients.