Valutazione dell’attività di sollevamento utilizzando le caratteristiche estratte da sensori indossabili

Abstract

The manual lifting tasks, which occur in the vast majority of workplaces can cause work-related low-back disorders (WLBDs), that are the most common musculoskeletal problems. Recently, to identify the relationship between WLBDs and risk factors, wearable monitoring devices-based biomechanical risk assessments have been proposed. The purpose of this study is to characterize from a biomechanical point of view, using wearable devices other lifting conditions to define, in the future, a risk classification tool that can be applied in each lifting condition. To do this, we recorded electromyographic data of workers during lifting tasks designed to have a growing lifting index (LI=1,2 and 3) by means of revised NIOSH lifting equation. Each lifting condition (LI=1 or LI=2 or LI=3) was obtained in three different ways modifying the asymmetry angle. We acquired data by using Wi-Fi transmission surface electromyograph (sEMG). From the sEMG signals, analyses of time and frequency domains were performed within the lifitng cycle to extract maximum value, the average rectified value, the mean frequency defined as the gravity center frequency of the power spectrum of the signal. The results show that these sEMG data grew significantly with the LI and that all the lifting condition pairs are discriminated. We will test whether machine-learning techniques used for mapping features extracted from wearable sensors on LI levels can improve the biomechanical risk estimation during these tasks. These findings suggest the use of kinematic and sEMG features to assess biomechanical risk associate with work activities can be integrated with methods already used for biomechanical risk assessment.