Impact forces in backward falls: Subject-specific video-based rigid body simulation of backward falls

Abstract

A critical missing component in the study of real-world falls is the ability to accurately determine impact forces resulting from the fall. Subject-specific rigid body dynamic (RBD) models calibrated to video captured falls can quantify impact forces and provide additional insights into injury risk factors. RBD models were developed based on five backward falls captured on surveillance video in long-term care facilities in British Columbia, Canada. Model joint stiffness and initial velocities were calibrated to match the kinematics of the fall and contact forces were calculated. The effect of joint stiffnesses (neck, lumbar spine, hip, and knee joint) on head contact forces were determined by modifying the calibrated stiffness values ±25%. Fall duration, fall trajectories, and maximum velocities showed a close match between fall events and simulations. The maximum value of pelvic velocity difference between Kinovea (an open-source software 2D digitization software) and Madymo multibody modeling was found to be 6% ± 21.58%. Our results demonstrate that neck and hip stiffness values have a non-significant yet large effect on head contact force (t(3) = 1, p = 0.387 and t(3) = 2, p = 0.138), while lower effects were observed for knee stiffness, and the effect of lumbar spine stiffness was negligible. The subject-specific fall simulations constructed from real world video captured falls allow for direct quantification of force outcomes of falls and may have applications in improving the assessment of fall-induced injury risks and injury prevention methods.

Keywords: Fall; Madymo; joint stiffness; older adults’ fall; rigid body dynamics.

Graphical abstract

Figure 1.

Diagram showcasing the process of the comparative study. This involved generating Rigid Body Dynamics (RBD) models via visual examination of videos, followed by a validation process where these models were cross-referenced against fall kinematics derived from a standard video kinematic analysis.

Figure 2.

Comparing the simulation with final conditions against video in all five falls.

Figure 3.

Digitization procedure: (a) calibration using the faller’s height. Examples of calibrated (b) horizontal and (c) vertical position versus time data in Kinovea. (d) Vertical velocity for head and pelvis in Kinovea.

Figure 4.

Force time history of: (a) lower neck, (b) upper neck, (c) lower lumbar spine, (d) upper lumbar spine, (e) left femur, and (f) right femur.

Figure 5.

(a) Pelvic force in five falls in simulations. Hip impact force during a backward fall with different floor types using a (b) mathematical model, and (c) experimental setup on a force plate.^,

Figure 6.

The effect of the neck, lumbar spine, hip, and knee stiffnesses on the head contact force in a sample fall. The stiffness effects were evaluated by a factorial design with values of −25%, 0%, and +25% of calibrated values in Madymo (total of 27 points) for each stiffness. *Blue, red and green colors indicate − 25%, 0%, and +25% stiffness scale respectively compared to the calibrated value.