Non-contact blood pressure detection model training method and blood pressure detection method and device

Abstract

The invention provides a non-contact blood pressure detection model training method and a blood pressure detection method and device. Periodic pulses are sensed in a non-contact mode based on wireless signals of a millimeter wave radar, a two-stage blood pressure detection model is adopted, and a pre-training process based on long sequence supervision information is introduced; parameters obtained after training of the first encoder are directly applied to the second encoder in a transfer learning mode, and the perception ability of the model to blood pressure related features is enhanced. A multi-level attention mechanism of a convolutional self-attention model is introduced to obtain global features and local context related features, and the blood pressure feature extraction capability can be remarkably improved. The method comprises the following steps: performing short-time Fourier transform on a reflected frequency-modulated continuous wave to perform segmentation processing on a reflected wave of each sub-region in a three-dimensional space, and sensing an artery pulse signal based on a maximum beam forming processing algorithm of an autocorrelation coefficient, so as to accurately sense the position of a wrist artery; and the pulse wave signal of the wrist artery can be sensed and recovered stably and accurately.

Description

technical field [0001] The invention relates to the technical field of vital sign detection, in particular to a non-contact blood pressure detection model training method, a blood pressure detection method and a device. Background technique [0002] Hypertension, a new type of disease, has become more and more common in people's lives with the improvement of living standards, and it is seriously harmful to health. Accurate blood pressure measurement is a prerequisite for prescribing the right medicine. For the elderly, the measurement method needs to be more reliable and convenient, and in order to realize the monitoring of vital signs, blood pressure must be continuously measured without interruption. [0003] The most accurate method of measuring blood pressure today is invasive arterial pressure measurement through a vascular cannula, in which a trocar is placed in an artery (usually radial, femoral, dorsal foot, or brachial) to measure directly Arterial pressure, but t...

AI Technical Summary

Problems solved by technology

[0007] In view of this, the embodiment of the present invention provides a non-contact blood pressure detection model training method, blood pressure detection method and device, to eliminate or improve one or more defects in the prior art, and solve the problem of wearable blood pressure in the prior art. The equipment is inconvenient to detect blood pressure, the accuracy is poor, and it is easy to be interfered by environmental factors

Images