Blood pressure estimation method and system based on multiple classification heads and weighted mean square error

The invention discloses a blood pressure estimation method and system based on a multi-classification head and a weighted mean square error, and the method comprises the steps: preprocessing obtained photoelectric volume pulse wave data, and inputting the data into a blood pressure estimation model, so as to output predicted systolic pressure and predicted diastolic pressure; the training process of the blood pressure estimation model is as follows: performing time sequence operation on acquired photoelectric volume pulse wave data through a sliding window to obtain photoelectric volume pulse wave data with time sequence information; a weight n1 is given to data, smaller than a lower limit threshold value, in the photoelectric volume pulse wave data with the time sequence information, a weight n2 is given to data located between the lower limit threshold value and an upper limit threshold value, and the photoelectric volume pulse wave data with the time sequence information after being given with the weight i