Marco Dozza - PhD in Bioengineering, 2003-2007

Biofeedback Systems for Postural Control

III Posture Symposium, 6·9 September 2003 - Smolenice, Slovak Republic

M. Dozza, L. Chiari, R.J. Peterka, F.B. Horak, and F. Hlavacka, “Controlling posture using an audio biofeedback system”, Proc. of the Third International Posture Symposium, “Human Posture Control: Physiology, Disorders, Modeling and Balance Rehabilitation”, p.39, Smolenice, Slovak Republic, 6-9 September 2003.

[Slide Presentation]




In normal conditions, postural stability in stance is accomplished through fusion and integration of three sensory channels - visual, vestibular and somatosensory. We investigated whether hearing was also a valid channel to provide additional postural information. Recently, we demonstrated that both healthy subjects and subjects with profound bilateral vestibular loss could decrease sway area in stance using auditory biofeedback (ABF) that provided information about trunk position and motion (Dozza, et al ,2003; Horak et al, 2003). In that study, ABF provided static information and the goal was to minimize postural sway. In this new study, we present the first step toward understanding a dynamic application of ABF such that ABF guides subjects' movements.


A portable accelerometer and gyro system, mounted on the subjects' back at L5, provided trunk kinematic information that a Matlab code processed to create stereo sound representing 2-D trunk position and velocity information. For both the AP and ML directions, auditory volume indicated how far the trunk had swayed from its initial position. AP sway modulated auditory frequency and ML sway modulated auditory balance. The sound dynamics were processed as a function of anthropometric parameters. A 1 degree threshold region was defined around the natural posture of each subject. When the subjects were outside this threshold region, the sound dynamics informed them about their movements.


The subjects performed the experiment standing on a force plate with eyes closed. They were instructed to keep their sway inside the threshold region. A sine wave at different frequencies (0.05, 0.1, 0.2, 0.4, 0.6, 0.8 and 1.2 Hz) represented the path used to move the threshold region. The experiments induced either AP or ML sway. Displacement of center of pressure (CoP) and the dynamic relation of CoP with respect to trunk displacement were used to determine postural response gain (peak CoP/peak audio threshold displacement) and kinematic postural strategies.


Subjects' postural sway followed the auditory sine waves generated in the specified direction, without change in the orthogonal direction of postural sway. The gain was largest at the lowest frequencies and decreased with increasing frequency, until it became almost zero at the highest frequency tested. At the lowest frequencies (0.05Hz and 1Hz), subjects were unaware that the sound induced them to sway but subjects became increasingly aware of their self-motion as frequency increased. AP and ML sway induced different movement strategies. In the AP direction, subjects used an ankle strategy at the lowest frequencies and gradually added a more and more hip strategy as frequency increased, starting at 0.6Hz. In the ML direction, the postural sway strategy appeared more effective than in the AP direction, because in the ML direction, gain of the postural response was larger than gain in the AP direction at every frequency tested. FFT of the CoP data showed that, at the lowest frequencies, the sway-induced response was specific to the driving frequency, whereas at the highest frequencies, the power was spread broadly around the driving frequency. The power at the driving frequency gradually increased during each trial as the subjects adapted to the sway frequency. Even after the auditory signal was stopped, a peak in the CoP power spectrum at the driving frequency continued for 5-10 sec. Also, for the 30 sec following the experiment with ABF, the power at the driving frequency was higher than its power during the 30 sec preceding ABF.


These experiments demonstrate how easy ABF information is for subjects to understand and to follow, even when it involves dynamic postural motion rather than maintaining initial stance posture. Without practice, naïve subjects had no difficulty using auditory information to modify postural sway, even when they were unaware that the ABF was inducing low frequency, sinusoidal postural sway. The fact that AP feedback does not interfere with ML direction and vice versa is evidence of the specificity of ABF. The postural strategies used to follow the ABF signals were automatically limited by biomechanical constraints on postural sway frequency with ML sway capable of following much faster frequencies than AP sway.

Supported by NIH DC01849 and DC06201.

©2007 DEIS Università di Bologna, ITALY- NSI Oregon Health and Science University, Portland (OR), USA - Webmaster MD