The invention provides a training method, a recognition method, a
system and equipment for a
fistula anomaly recognition model. The training method comprises the steps that a training
data set is acquired, the training
data set comprises a plurality of pieces of sample data, and each piece of sample data comprises
fistula anomaly recognition influence parameters and corresponding sample anomaly recognition results; wherein the
fistula abnormity identification influence parameters comprise fistula tremor signals and
blood flow and / or
blood vessel diameter; and training a preset model based on the plurality of sample data to obtain a fistula anomaly recognition model. According to the method, the fistula tremor
signal, the
blood flow and / or the
blood vessel diameter and the corresponding sample anomaly recognition result are jointly used as the sample data to
train the model, the accuracy of the fistula anomaly recognition model is improved, anomaly recognition can be performed on the fistula in real time, accurate evaluation about the internal
arteriovenous fistula blood vessel function is provided for doctors, and the accuracy of the fistula anomaly recognition model is improved. Therefore, various complications of internal
arteriovenous fistula can be timely and accurately treated, and the safety of a patient is guaranteed.