Feedback control of human metabolic work rate during robotics-assisted treadmill exercise

Abstract

A novel approach is presented which suggests the use of human metabolic work rate to define and regulate exercise intensity during robotics-assisted treadmill training. The work describes the design and technical validation of the new method.

A feedback structure is proposed which provides automatic regulation of metabolic work rate, in conjunction with an embedded feedback loop for volitional control of mechanical work rate. Human metabolic work rate was derived in real time from breath-by-breath measurements of oxygen uptake and carbon dioxide output.

The results show that the feedback method provides close to nominal performance for square-wave and ramp reference tracking tasks and that good disturbance rejection properties are obtained. A collateral finding of this work is an estimate of 14.5% of the metabolic efficiency of robotics-assisted treadmill exercise.

The use of feedback control of human metabolic work rate provides a direct measure of exercise intensity as perceived by the exercising human as it directly reflects the energy requirements of the working muscles. This complements previous approaches to guiding robotics-assisted treadmill training based on mechanical work rate, heart rate or oxygen uptake. The new approach based on metabolic work rate may have advantages in populations with compromised and widely varying exercise responses. This provides a new approach for driving and controlling active patient participation during robotics-assisted treadmill exercise.

Introduction

There is growing interest in the potential of rehabilitation-robotics devices for cardiopulmonary rehabilitation of patients with a range of neurological deficits [1], [2], [3]. This would provide a complement to the application of these devices for neurological rehabilitation. Hitherto, the clinical application and research focus of these devices has been the desire to promote positive neurological adaptations, plasticity and the recovery of walking function [4], [5]. The complementary application for cardiopulmonary rehabilitation raises the question of how to define and regulate exercise intensity while patients train with such devices.

Within the field of exercise science and physiology, exercise intensity can in general be characterised in a number of ways. Measures include external mechanical work rate, heart rate, oxygen uptake rate, or ratings of perceived exertion. These approaches have in part been translated to the context of robotics-assisted treadmill exercise using the Lokomat driven-gait orthosis. External mechanical work rate has been estimated using real time measurements of human–machine interaction forces and orthosis position, and this was in turn used to drive the exercise intensity [6], [7], [8]. In this setup the patient was required to maintain a target work rate using a visual feedback of measured work rate. Direct control of mechanical work rate was used during a clinical investigation of the potential of work-rate-guided Lokomat training for cardiopulmonary rehabilitation [3], [9]. In closely related work, a weighted sum of joint torques was chosen as the controlled variable and as the measure of intensity while patient interaction occurred using a visual task [10]. That work also explored feedback control of heart rate on the Lokomat using either a visual task for the patient or automated changes in treadmill/Lokomat speed. Direct feedback control of oxygen uptake rate was implemented using breath-by-breath monitoring together with a human-in-the-loop control of mechanical work rate [11], [12].

In the present work, a novel approach is presented which suggests the use of human metabolic work rate to define and regulate exercise intensity during robotics-assisted treadmill training. A feedback structure is proposed which provides automatic regulation of this variable, in conjunction with an embedded feedback loop for volitional control of mechanical work rate. Human metabolic work rate is derived in real time from breath-by-breath measurements of oxygen uptake and carbon dioxide output. This new variable provides a direct measure of exercise intensity as perceived by the exercising human as it directly reflects the energy requirements of the working muscles. This measure may have clear advantages in populations displaying low physical fitness, low metabolic efficiency or neurological deficits: in such populations a given metabolic work rate will be associated with a wide range of external work rate or heart rate values. These values may therefore poorly reflect exercise intensity.

This new method will be targeted primarily at people who are indicated for Lokomat therapy as a consequence of stroke or spinal cord injury. This new approach specifically targets cardiopulmonary rehabilitation and would complement conventional therapy or alternative exercise modalities, with impact on important markers of cardiopulmonary status.

The present work aims to design and technically validate a new method for feedback control of human metabolic work rate during robotics-assisted treadmill exercise. Availability of metabolic work rate as an indicator of exercise intensity would provide a new approach for driving and controlling active patient participation during this form of exercise.