Sleep monitoring method and system

Abstract

The invention discloses a sleep monitoring method. The method comprises the following steps: in accordance with acquired sleep characteristic data, acceleration data and heart rate data, running a statistical empirical model, and acquiring a first predicating result corresponding to the statistical empirical model; in accordance with the heart rate data and the acceleration data, running a logistic regression model, and acquiring a second predicting result corresponding to the logistic regression model; and implementing weighted calculation on the first predicting result and the second predicting result, and obtaining a sleep state in accordance with a calculation result. With the application of the sleep monitoring method, user's sleep state can be monitored more accurately.

Description

technical field [0001] The invention belongs to the technical field of wearable devices, in particular to a sleep monitoring method and system. Background technique [0002] Wearable smart devices have become a trend of the current era. Due to different functions, various smart wearable devices have different focuses. Some focus on sports step counting, some focus on heart rate monitoring, and some focus on sleep monitoring. [0003] In the prior art, the general flow of the sleep monitoring algorithm is as follows: the characteristic data for detecting sleep are collected in sections, and the collected characteristic data include the number of times the arm swings from rest to the swing, the number of times the arm is in a stationary state, and the time of continuous arm movement , the total time of arm activities in each stage; by reading the three-axis data of the register in the acceleration sensor, the criterion for identifying the arm posture is obtained; by obtaining...

AI Technical Summary

Problems solved by technology

[0004] The sleep monitoring method in the prior art obtains the parameters of the algorithm model in most scenarios through offline testing, and then monitors all situations with fixed parameters. There may be abnormal situations such as sleep analysis data after the wristband is left standing for one night. Data analysis is not accurate enough

Since the characteristics of the model only include four characteristic parameters of three accelerations and heart rate, this algorithm does not carry out sufficient feature engineering, and the analysis of the sleep state is not accurate, which is the shortcoming of the current sleep monitoring algorithm.

Images