Evidence shows that that nearly 75% of people over age 70 have some type of balance disorder. Given the ongoing growth of the aging population, this represents a public health burden and threat to the healthcare system. Balance assessments play an important role in accurately diagnosing potential disorders, identifying fall risks, and developing treatment plans. Traditional balance assessments rely on clinician’s judgments, which can be subjective. These authors looked at using equipment-based and digitalized assessment methods to eliminate subjective opinions from balance assessments. They combined 3D skeleton data from the Kinect system in 10 healthy recruits with deep convolutional neural networks (DCNN), a popular machine learning technology, and compared their findings with those of traditional assessments during walking from 300 sample cases. Their results suggest that 3D skeleton data and DCNN can be used for balance assessment with decent accuracy. The proposed method should be useful in early screening balance impaired people. It can partially replace commonly used balance measures and reduce the influence of subjective factors.
Source: Ma X, Zeng B and Xing Y. Combining 3D skeleton data and deep convolutional neural network for balance assessment during walking. Front. Bioeng. Biotechnol. 11:1191868. doi: 10.3389/fbioe.2023.1191868
