Categorization of fistula condition in hemodialysis patients using spectral analysis of audio signals

Patients on hemodialysis need to constantly monitor the condition of their fistula, either independently or by visiting a doctor. This can be challenging, as each person may perceive the condition of their arteriovenous fistula differently. In this paper, we present a machine learning model to categorize fistula conditions. We considered various feature filtering methods to enhance the accuracy of the categorization. Additionally, we proposed a new method using spectral features, which allows for a more precise determination of the condition of patients undergoing hemodialysis.