One in ten Japanese suffers from back pain.Back pain may worsen if you sit for long periods of time, even if you maintain the correct posture.If you can predict the occurrence of back pain, you can take actions such as stretching and exercising to avoid it.Until now, the posture and sitting time that are prone to chronic low back pain have been determined using a chair equipped with a sensor, but a technique for predicting subjective low back pain from changes in the pressure center associated with sitting for several minutes to several tens of minutes. There was no.

　The research group collected data for three months from a total of 22 office workers who actually worked in the office and agreed to the study.About low back pain I had a tablet terminal record the degree of subjective low back pain three times a day.From the load fluctuation data when sitting, deep learning was used to identify similar signal changes common to 3 people, and further detected a combination pattern that is strongly related to the worsening of low back pain that appears in conjunction with them.As a result, high prediction accuracy (sensitivity and specificity of about 3%) was achieved, and results at a practical level were obtained.

　This time, by applying the latest time-series signal processing and AI technology, it has become possible to predict events such as low back pain, which is a problem, from a probabilistic signal system derived from a living body.This has the potential to greatly expand the utility value of sensors in wearables and in daily life.In addition, the discovery of small movements that prevent posture fixation is expected to further elucidate the causes of symptoms called "indefinite complaints" such as stiff shoulders, headaches, and arthralgia, and to develop coping strategies.

Paper information:[Frontiers in Physiology] Low Back Pain Exacerbation Is Predictable Through Motif Identification inCenter of Pressure Time Series Recorded During Dynamic Sitting