Share this article on:

Functional Output Improvement in FES Cycling by Means of Forced Smooth Pedaling


Medicine & Science in Sports & Exercise: May 2007 - Volume 39 - Issue 5 - pp 764-780
doi: 10.1249/mss.0b013e3180334966
CLINICAL SCIENCES: Clinically Relevant

Purpose: Investigation of the influence of forced smooth and normal (nonsmooth) pedaling on the functional output of outdoor functional neuromuscular electrical stimulation (FES)-propelled cycling of spinal cord-injured subjects.

Subjects: Twelve subjects with complete spinal cord injury (T4-T12) and limited previous FES training.

Method: Each subject participated in two separate outdoor sessions: once while pedaling a tricycle in a fixed gear, and a second time while free pedaling the same tricycle; both times with FES. Data on distance covered until exhaustion, cadence, and pedal forces were collected. Energy balance calculations led to evaluations of jerk loss and joint-related concentric/eccentric work.

Results: First-trial and total session distances were 68 and 103% longer, respectively, in the forced smooth cycling session than in the free cycling session (P < 0.001). Significantly more additional crank work (accompanied by increased concentric work production) was generated in nonsteady cycling phases to overcome increased jerk losses during free than during fixed-gear pedaling. During fixed-gear pedaling, timing and joint location of muscle work generation were more similar to the cycling of able-bodied subjects than during freewheel pedaling, because most work was generated by knee extensors in the power phase during the former pedaling mode.

Conclusions: The superiority of forced smooth cycling to free cycling, as regards functional output distance, is based on less energy expenditure (less jerk loss and muscle tension) and on more efficient production of energy (more efficient timing and joint location of work production). Some energetic mechanisms that are advantageous for fixed-gear cycling act predominantly in unsteady phases; others work continuously during all phases of cycling.

1Center for Sensorimotor Research, Department of Neurology, Ludwig-Maximillians University, Munich, GERMANY; and 2Clinics of Neurology, Bad-Aibling, GERMANY

Address for correspondence: Johann Szecsi, M.D., Center for Sensorimotor Research, Dept. of Neurology, Ludwig-Maximillians University, Marchioninistrasse 23, 81377 Munich, GERMANY; E-mail:

Submitted for publication December 2005.

Accepted for publication December 2006.

It is common knowledge that stationary ergometric cycling of spinal cord-injured (SCI) persons by means of functional neuromuscular electrical stimulation (FES) has positive therapeutic effects such as muscle and cardiovascular training, and it also helps prevent insulin resistance, pressure sores, and osteoporosis (6,13). The general therapeutic usefulness of this technology, however, has remained limited because of the small amount (10-25 W) of generated power (5,19).

Mobile cycling (16,20) will be used for recreation, mobility, or fitness only if it permits the coverage of useful distances (e.g., at least 10 km (7)) at a speed of at least 6-7 km·h−1. These requirements have been achieved only in exceptional cases, despite the considerable effort invested in technical improvements of FES cycling in recent years (1,5,12). This topic will attract increasing interest in the future as mobile cycling systems for paraplegics continue to proliferate.

Few studies have explicitly discussed the fact that the majority of SCI persons perform FES cycling in fatigue mode rather than in steady-state mode (25), because their main concern is short-term generation of maximal functional output. In the 1970s, Harrison (11) discovered a tie dependency of maximal human output to suitable kinematic constraints of motion cycle. His maximal power-time data collected for short-duration rowing were based on relative power measurements produced by the same able-bodied individuals using different motions (free or forced) and mechanisms that could force the ends of the rower's stroke and conserve the kinetic energy of the moving masses. The essential difference between forced and free motion is that the kinetic energy of the limbs can be fed back into the mechanical system during forced motion, whereas much of this energy must be absorbed by the limbs during free motion. Moreover, linkages impose a harmonic velocity of movement in forced motion, whereas the motion is transmitted through a one-way clutch in free motion. Harrison's main finding was that considerably longer delivery times for a certain amount of power could be achieved during forced rowing than during normal (free) rowing. Forced motion in bicycle pedaling corresponds to a fixed-gear transmission, that is, one without a freewheel. The harmonic limb velocity of the forced motion corresponds in cycling to uniform crank-rotating velocity (smooth pedaling). Whereas fixed-gear pedaling seems to play no role in steady-state cycling of able-bodied persons (29), it is apparently important in fatigue-mode FES cycling (25) of SCI subjects. The freewheel decouples when the contact torque exerted on the rear wheel goes below about 0.5 N·m (10), an occurrence that is likely in SCI subjects, who have less than 10% of the torque produced by able-bodied persons (17). Although some of today's stationary mechanically braked or motorized ergometers realize fixed-gear pedaling, the inertial loads of their flywheels are usually less than 5.2 kg·m−2. In contrast, the corresponding inertial loads of outdoor bicycles and tricycles range from 10 to 160 kg·m−2 (9,10). One study (9) has pointed out that the application of high inertial flywheels (corresponding to the inertia of outdoor bicycles) to a fixed-gear ergometer led to unusually easy pedaling against a resistive load for a poststroke hemiparetic subject. Consequently, the combination of high inertial load with fixed gearing seems to enhance power potential in rehabilitation. This subject has not been previously investigated in the relevant literature.

Whether pedaling is forced smooth (with fixed gearing) or free (jerky with freewheeling) may have an impact on the effectiveness of work, particularly through the following:

I. The force-velocity relationship of muscle influences the efficiency of nonsmooth moving (positive work efficiency is strongly dependent on contraction velocity; at 0 or maximum velocity, the power is 0 (28)).

II. The alternating coupling and decoupling of the freewheel and rear wheel generate peak contact forces, tensions, and, therefore, more fatigue (22).

III. Kinetic energy is lost because of jerk (11).

IV. In jerky rather than smooth movement, more inefficiency occurs because of generation of energy at one joint and its absorption at another (30).

V. Accelerating and decelerating inertial moment (10) with nonsmooth movement affects the small energy reserves of SCI subjects.

The principal aim of our study was to elucidate the influence of forced smooth pedaling on the functional output of outdoor FES cycling of SCI subjects. We compared functional output measures such as distance, duration (endurance) to exhaustion, and the output power during outdoor FES cycling of SCI subjects riding the same tricycle twice: once using the normal freewheel-clutch mechanism, and once with fixed gear (crank hard-coupled with the flywheel).

The next goal was to establish what causes the different functional performance during cycling, and when it occurs. Therefore, cycling phase-related differences in simple mechanical parameters like smoothness (root mean square of cadence and torque) and freewheeling (crank angle residual) between cycling conditions were investigated. A further step was to analyze energetic mechanisms potentially responsible for the differences manifested in endurance behavior of the muscle, in smoothness of pedaling, and in freewheeling behavior, while comparing the fixed-gear condition with the freewheel condition. To do this, energy exchange between crank and external load (external work), and energy flow that moves the leg segments (internal work), were compared under both cycling conditions.

Back to Top | Article Outline



Twelve subjects (nine males and three females; mean age 39 ± 7 yr) with complete spastic SCI (ASIA-A, T4-T12) and limited spastics (modified Ashworth scale 0-1) were recruited. The subjects were experienced with FES cycling but had rather limited previous FES cycling training in the outpatient clinic (on average, 6-28 months, with 0.6 training sessions per week). These training sessions consisted, on average, of 1.5 h of FES-propelled pedaling on a motor-braked ergometer. The two inclusion criteria for the study were (i) ability to perform FES cycling outdoors for a distance of at least 400 m (preliminary work had shown that weak patients covered distances < 400 m, which did not allow us to distinguish the three-phase structure; see further), and (ii) a modified Ashworth spasticity scale ≤ 1 to avoid the strong interference of daily varying spasticity during forced smooth or free cycling.

The study was approved by the local ethics committees, and subjects gave their informed, written consent.

Back to Top | Article Outline

Study Design

Each subject underwent two separate experimental sessions: one while forced smooth pedaling a tricycle equipped with a fixed gear, and a second session while free pedaling the same tricycle equipped with a freewheel mechanism. Two sessions were used because the functional output of FES cycling is predominantly determined by the initial fatigue situation (25). The forced smooth cycling tests of all subjects were done within a block of 12 sessions corresponding to 14 d, using the tricycle with the hard coupled crank/freewheel. Subsequently, the fixed hub of the tricycle was replaced with a speedhub that allowed freewheeling. The free cycling sessions of the 12 subjects were run in a second block of 14 d. The time between tests ranged from 7 to 28 d. Because preliminary experiments had shown a tendency for higher functional output in forced smooth pedaling, the forced smooth cycling experiments were deliberately performed first and the free cycling experiments were performed later, to avoid any possible bias of functional output in the second experiment resulting from training performed in the first experiment.

Back to Top | Article Outline

Stimulation, FES Tricycle Settings, and Instrumentation

Leg-propelled cycling used electrical stimulation of the quadriceps, hamstrings, and glutei muscle groups of the SCI subjects who rode a commercially available tricycle (Fig. 1A OVG, Munich, Germany). The ankle joint was immobilized, and leg movement was restricted to the sagittal plane using shank/foot orthoses. The straps of these orthoses fixed the feet and the shanks to the pedals. Twelve auto-adhesive electrodes, 5 × 9 cm2 in size, were used (Krauth and Timmermann, Hamburg, Germany). The Microstim8 constant-current eight-channel stimulator (Krauth and Timmermann, Hamburg, Germany) provided a rectangular biphasic stimulation current with a frequency of 20 Hz, covering an intensity range of 0-99 mA and maintaining a pulse width of 500 μs. Stimulation intensity (participant controlled by a handgrip-integrated potentiometer) was gradually increased from the start of cycling (< 5% of cycling duration) up to the maximal value of 99 mA.

An eight-bit incremental encoder (Fig. 1A) (accuracy 1.4°), synchronized to turn with the crank shaft, determined the actual position of the crank. The 0 reference point of the crank angle was defined by the backward-pointing left crank arm (top dead point at +23°). Stimulation of the muscles was switched on or off depending on the crank angle (Appendix 1). The angle of compensation for the electromechanical delay was fixed at 28° (corresponding to 140 ms at 35 rpm).

The tangential and radial forces applied by the rider to the crank arms were collected simultaneously from the left and right sides using instrumented crank arms (o-tec GmbH, Bensheim, Germany). Both left and right force-measurement units (Fig. 1A) consisted, respectively, of two Hall sensors allowing the acquisition of tangential and radial force components. Calibration of the force sensors was performed with 20.6-, 31.7-, and 63.4-N weights. Nominal force amounted to 1 kN, and accuracy was 1 N, corresponding to crank torques of 0.15 N·m (using 0.15-m crank arms).

Crank angle and crank forces measured during the drive by the position decoder and the force sensors, respectively, were recorded at a sample rate of 20 Hz on a laptop with a NI-6024 data-acquisition card (National Instruments, Austin TX), which was attached to the tricycle. Data were averaged over a number of consecutive revolutions within time intervals of 5 s. Therefore, at the minimal cadence of 30 rpm, 2.5 cycles were averaged; thus, 2.5 × 20 = 50 data points were collected per sample interval.

The kinematic chain of the OVG tricycle (Fig. 1A) consisted of four cog wheels and two chains. The propelled left rear wheel was equipped with a Sachs Spectro S7 (ZF Sachs AG, Schweinfurt, Germany) speedhub in the freewheeling tricycle (Fig. 1C). This gave a transmission ratio of 0.56 in the immobilized first gear and allowed freewheeling. Thus, the total transmission ratio amounted to 42/28 × 24/19 × 0.56 = 1.06. In the fixed-gear tricycle, the left rear wheel was propelled through a solid hub (Fig. 1B, no freewheeling). Some of the cogwheels were changed accordingly to fit the transmission ratio. Therefore, the total transmission ratio in fixed-gear cycling was 24/28 × 22/18 = 1.06. Differences in energy loss attributable to the use of different gearing mechanisms to achieve the same overall transmission ratio were estimated to be less than 5% (Appendix 2).

The participants sat in the same position (seat height and crank hip distance) in both trials, with maximal extension angles of the knees of 150-160° (in the bottom dead point of pedaling). The chain(s) of the fixed-gear tricycle were completely taut without any visible sign of sagging.

Back to Top | Article Outline

Cycling Experiments

To preserve a constant driving resistance, cycling experiments were performed in a 528-m-long tunnel (consisting of 11 marked segments, each 48 m long) located under the clinic and used for deliveries. The drive resistance was determined for all subjects in supplementary sessions, applying a towing test with a spring balance (accuracy 1:100 corresponding to 0.49 N) and the freewheel tricycle with passively sitting subject (no stimulation). By performing the test at uniform cadences of 40 rpm (checked by observing the signal of the wheel-position decoder on the laptop display), we obtained an average consumed work per revolution against drive resistance of 10.5 ± 1.12 J (Wroll, Appendix 3).

Cycling experiments (free and forced smooth) began with short (2-3 min) warm-up stimulation periods performed on the jacked-up tricycle. Subsequently, several cycling trials, propelled by stimulation until exhaustion of the musculature (standstill of the tricycle), were conducted. Subjects turned around at the end of the tunnel by means of full stimulation intensity, without any breaks or external support. After a short 4- to 5-min break, the next trial was started, until distances of at least two segments (96 m) could no longer be achieved. In this case, exhaustion was considered complete for this day, and the session was ended.

Back to Top | Article Outline

Data Reduction

Filtering and averaging.

Cadence n[rpm] computed as the finite difference of angle values was digitally filtered with a second-order Butterworth filter with a cutoff frequency of 4 Hz. The four pedal-force components were forward-backward filtered offline with 5-Hz cutoff frequency. Low-pass filtering performed for noise reduction captured 97-93% of the crank torque signal power. Every 5 s of a trial, average values of cadences and force components over cycles covered by this time interval were computed. Thus, a variable number of cycles was averaged, but the minimum number of cycles was 2.5 for a cadence of 30 rpm.

Back to Top | Article Outline

Kinematic parameters.

Covered distances s[m] were obtained from the filtered cadence, according to the formula

where rw = 0.24 m is the radius of the propelled wheel, and γ = 1.06 is the transmission ratio of the tricycle in both cases. For each session, the distance achieved in the first trial and the total sum of distances achieved during all trials in the session were computed and recorded. Preliminary experiments had shown that the trials could be divided into three phases in both free and forced smooth cycling conditions (see Fig. 2A, B). The first phase P1 corresponded to a power (cadence) valley (5), the second and usually longest phase P2 showed a power (cadence) plateau, and the third phase P3 represented the final power (cadence) decay (Appendix 4). Nonsmoothness was characterized by the root mean square (RMS) of cadence, and crank torque by the RMS of torque, using 2-s windows. For the description of freewheeling, the crank angle residual was used (Appendix 5).

Back to Top | Article Outline

External work balance.

The resultant two-legged crank torque Mcrank was calculated by summing the left and right tangential crank forces and multiplying the sum by the crank arm length rc = 0.15 m. Crank work per revolution (4) cycle was computed as Wcrank = cycle Mcrankdα, with α crank position. Therefore, crank work contains not only the active muscle work, but also a passive part attributable to the potential and the kinetic energy of legs, orthoses, pedals, and crank arms.

Wcrank has to additionally overcome the drive resistance Wroll, also a part of work Wadd responsible for both the losses attributable to the moving parts (friction in chains and gears; impact losses) and for the energy absorbed or generated by linearly accelerating or decelerating the rider/tricycle system:

Because impact losses increase at higher speed (29), we also expected higher Wadd during jerky and nonsmooth pedaling than during forced smooth pedaling. If impact losses attributable to jerk and nonsmoothness occur, these will belong to the additional work. In fact, low-pass filtering of the pedal force components reduces noise and also attenuates impact oscillations (Appendix 6). Although low-pass filtering reduced the signal power of crank torque by 3-7%, the crank work per revolution varied less than 0.5-1.5%, and, therefore, additional work (7-35% of Wroll; see results) remained unbiased.

Back to Top | Article Outline

Internal work balance.

The lower limb was modeled as a bilateral, planar, two-segment, rigid body system with an external reaction force at the pedal spindle. Cadence, tangential, radial pedal forces, and body segment parameter data (Appendix 7) served as inputs to the inverse dynamic analysis (using Simmechanics Toolbox of Matlab 6.1, by MathWorks, Natick, MA) to calculate knee- and hip-joint generalized muscle moments (GMM). Muscle power was calculated for each joint by multiplying the GMM at an individual joint and the angular velocity of the two segments constituting that joint.

Muscle power profiles were timebase adjusted to account for normalization and were integrated across each revolution to calculate work at each joint. Positive (concentric) work was defined as integration of all values of muscle power above the zero line, and negative (eccentric) work was defined as the integration of all values of muscle power below the zero line. Net muscle work relative to joint (leg) was calculated as the difference of concentric and eccentric work relative to the respective joint (leg). In the per-joint (leg)-based approach, the bilateral combined averages for total concentric, eccentric, and net muscle work per joint (leg) were obtained. The direction of the GMM in a particular joint revealed the extensor and flexor work.

Muscle work Wmuscle produced by both legs was obtained by adding the net muscle work occurring in the four joints.

External and internal balance are coupled by the equation:

whereas passive work per revolution occurs because of potential and kinetic energy changes of the legs. If passive work is negligible, this means a steady-state condition from the viewpoint of internal balance and by no means external balance (the tricycle may accelerate or decelerate during the same revolution cycle).

Back to Top | Article Outline

Statistical Analyses

After the data were collected, the time courses of cadence and torque production were plotted, and phases P1, P2, and P3 were delimited (Appendix 4). Subsequently, descriptive statistics were derived for distances (total, first trial, and phase related), cycling durations (first trial), and output power (torque × cadence, related to phase P2) during freewheeling and fixed-gear cycling. Means of distances, durations, and powers were compared using paired, one-sided t-tests.

Further, for every subject, nonsmoothness and freewheeling parameters, crank, and muscle work data (with its leg- and joint-related extensor, flexor, concentric, and eccentric components and their bilaterally combined averages) were averaged and were assigned to the three phases P1, P2, and P3. Descriptive group statistics were derived and depicted graphically for all these parameters.

To perform interconditional (free vs fixed-gear pedaling) and phase course-related comparisons on phase-related work components, repeated-measures (N = 12) ANOVA with two independent factors (factor 1 = {free/fixed-gear cycling}, factor 2 = {phases P1, P2, P3}) and with Scheffe's multiple-contrast post hoc tests was used. Descriptive statistics are represented as mean ± standard deviation. Statistical tests were considered significant if P < 0.05, unless otherwise mentioned. All analyses were performed with the Statistics Toolbox of Matlab 6.1. (Mathworks, Natick, MA).

Back to Top | Article Outline


Back to Top | Article Outline

Kinematic and kinetic parameters (subject 1).

Examples of the cadence and crank-torque data collected during the first trial are given in Figures 2A and 3A and in Figures 2B and 3B for subject 1 during free cycling and forced smooth cycling, respectively. Both cadence curves clearly show the three-phase structure; the first phase (P1) corresponds to a power valley, the second and longest phase (P2) shows a power plateau, and the third phase (P3) represents the final power decrease. The distances covered by subject 1 during the three phases (P1, P2, and P3) of the free cycling trial were 121.9, 374.4, and 220.2 m, respectively; a total of 716.5 m was covered during the whole first trial. The corresponding distances during the forced smooth cycling trial were 317.5, 1244.9, and 478.8 m, respectively; a total of 2041.2 m was covered during the first trial. Nonsmoothness for the same subject characterized by cadence RMS increased in the free cycling trial (Fig. 2A) from an initial 5.6 rpm to a maximum of 13.6 rpm in P1, decreased to mean 5.4 rpm computed over the entire P2, and increased again to mean 7.5 rpm in the final P3. In the forced smooth cycling trial (Fig. 2B), the RMS of cadence increased less, from an initial 5.5 rpm to a maximal 6.9 rpm in the power valley of P1, further decreasing to mean 3.5 rpm computed over the entire P2, and finally increasing again to a mean of 6.5 rpm in the final P3.

Crank torques in both free and forced smooth pedaling are depicted in Figure 3A and B, respectively. Crank torques were mostly (Appendix 6) monophasic (positive according to cycling direction) in the free pedaling trial and biphasic (positive according to and negative, opposite to cycling direction) in the forced smooth pedaling trial (Fig. 3B), whereas pronounced negative torques occurred during phases P1 and P3 of forced smooth pedaling. Therefore, considerable power exchange took place over the crank in both directions in the nonsteady phases of cycling during fixed-gear cycling (cadence is always positive).

Nevertheless, in the P2 phase of the forced smooth trial, crank torque was almost monophasic (it only had a small negative part in the vicinity of the top dead point). The cyclist made almost no use of his possibility to have energy exchange with the cycle in the steady phase of forced smooth cycling. The largely monophasic crank torque in free pedaling shows RMS magnitudes of maximum 15.1 N·m, mean 5.2 N·m, and 9.1 N·m in P1, P2, and P3, respectively. The biphasic crank torque in forced smooth pedaling shows somewhat lower RMS magnitudes (P1, P2, and P3: 11.5, 5.1, and 5.5 N·m, respectively). Thus, nonsmoothness of cadence and crank-torque magnitudes were higher in the nonsteady P1 and P3 than in the steady P2 phases, during both free and forced smooth pedaling.

Inspection of the amount of freewheeling in terms of crank angle residuals during both cycling conditions (Fig. 4) shows that freewheeling occurs in both conditions, but a-3s expected, it is much stronger during "freewheeling" (P1: mean 27.4°; P2: mean 18.1°; P3: mean 24.5°) than during fixed-gear cycling (P1: mean 12.1°; P2: mean 7.9°; P3: mean 12.5°). The drive transmission system containing four cogs and two chains caused a certain amount of freewheeling during fixed-gear cycling as well. The amount of freewheeling was large during the nonsteady phases P1 and P3 and was less so in the steady phase P2, irrespective of cycling mode.

Back to Top | Article Outline

Work balance (subject 1).

Figure 5A and B depicts the time courses of crank work per revolution during the free and forced smooth pedaling trials, respectively. Evidently, crank work in conditions of freewheeling shows a more valleylike course than in conditions of fixed-gear cycling. In conditions of freewheeling cycling, more crank work was produced in unsteady phases P1 and P3 than in the steady phase P2 (14.84, 11.0, and 15.2 J, respectively). In conditions of fixed gear, crank work generated in unsteady phases P1 and P3 exceeded work produced in the steady phase P2 less, or did not exceed it at all (11.8, 12.0, and 12.1 J, respectively) compared with conditions of freewheeling.

Cadence profiles of the left leg during the revolution cycle taken in the "deepest" point of the power break-in valley in unsteady phase P1 during free and fixed-gear cycling are shown in Figure 6 (corresponding to t = 75-80 s and t = 170-175 s in Fig. 7A and B). In freewheeling conditions, cadence presented oscillations with large ranges (17-57 rpm), with two deep valleys (16.5 and 17.5 rpm at the top and bottom dead points: 23 and 203° of this patient) and two moderately deep valleys. Cadence oscillations were far less in the fixed-gear condition (32-57 rpm), with two moderately deep valleys, which are all shifted with respect to the same dead points as in the freewheeling condition (the geometrical sitting position was the same). Corresponding power profiles occurring in the left knee and hip joints are shown in Figure 7A and B. The shape and peaks of power (and also force profiles) differed in the two conditions, because power (and force) profiles are not only dependent on isometric torque profiles (24) that have to be the same but also on differing cadence profiles (Fig. 6). Under fixed-gear conditions, in the minimum point of the power break-in, the left leg muscle work (8.6 J = 100%) consisted of three thrusts occurring in the power, early, and late recovery phase of the revolution cycle (Figs. 7B, 8B), produced by the concentric knee extensor and the eccentric hip flexor GMM (11.24 J, 131%), the concentric knee flexor and eccentric hip extensor (0.48 J = 6%), and the concentric hip flexor and eccentric knee extensor GMM couples (−3.12 J = −37%), respectively. Therefore, in the fixed-gear condition, almost all left-side work was generated in the power phase by the concentric knee extensor and the eccentric hip flexor couple. In freewheel conditions, the left leg muscle work (18.32 J = 100%) at the minimum point of the power break-in (Figs. 7A, 8A) was produced by the same three thrusts in the power, early, and late recovery phases of the cycle as in the fixed-gear condition; nevertheless, there was a completely different percentual distribution across these thrusts; 4.8 J = 26%, 3.12 J = 17%, and 10.4 J = 57%, respectively. Thus, all three thrusts produced considerable parts of the total work, whereas the bulk of the work was generated in late recovery by the concentric hip flexor and the eccentric knee extensor.

Time courses of total concentric and eccentric muscle work of the left and right legs under freewheeling (Fig. 8A) and fixed-gear (Fig. 8B) conditions showed larger absolute magnitudes of concentric work in unsteady phases P1 and P3 than in steady phase P2, irrespective of the condition. Thus, for example, in the freewheel condition, the (bilaterally averaged) concentric work and the net muscle work (difference between concentric and eccentric work) decreased and increased again during P1, P2, and P3 (Fig. 8A). In net muscle work, interphase differences comparing steady and nonsteady phases during the fixed-gear condition were smaller (Fig. 8B). Variation of passive mechanical energy of the legs (Fig. 8A and B) was approximately negligible during steady and nonsteady phases. Therefore, crank work can be considered identical to bilateral muscle work.

Joint-based separation of concentric/eccentric extensor/flexor work under the freewheeling condition showed unsteady behavior in the P1 and P3 phases and roughly steady behavior in the P2 phase (Fig. 9A). During the fixed-gear cycling condition, joint-based separation of concentric/eccentric extensor/flexor work showed (Fig. 9B) that concentric and eccentric knee extensor work decreased, whereas concentric and eccentric hip flexor increased during cycling. Negligible activity of the concentric knee flexor and the eccentric hip extensor was recorded.

Back to Top | Article Outline

Covered distances, times to exhaustion, and power.

Figure 10 shows that group-averaged distances achieved in the first forced smooth cycling trial (1279 ± 488 m) exceeded distances achieved in the first free cycling trial (761 ± 270 m) by 68% (P < 0.001). The group-averaged total distances achieved in the former trial (2347 ± 538 m) exceeded the total distances achieved in the latter trial (1160 ± 482 m) by 103% (P < 10−5). A further analysisof the distances achieved in the P1, P2, and P3 phases of the first trials showed that the advantage of the first forced smooth trial over the first free trial was predominantly attributable to the P2 phase (587 ± 277 vs 363 ± 176 m, P < 0.05). The distances of P1 and P3 in the forcedsmooth trial also significantly exceeded the corresponding distances in the free trial (P < 0.005 and P < 0.05, respectively).

First trial cycling times were significantly prolonged for forced smooth cycling in comparison with free cycling (1279 ± 428 s for forced smooth vs 686 ± 278 s for free cycling, P < 0.01).

In contrast, a comparison of mean respective peak output power achieved in phase P2 found no significant differences between free and forced smooth cycling (Appendix 8).

Back to Top | Article Outline

Nonsmoothness of pedaling and freewheeling.

Nonsmoothness (as defined by cadence and crank-torque RMS) was significantly higher in the freewheeling than in the fixed-gear condition, in the unsteady phases P1 and P3 (Appendix 9). Moreover, freewheeling, defined as residual crank angle, was always significantly higher in all phases during the freewheel than during the fixed-gear cycling condition (Appendix 10).

Back to Top | Article Outline

External work balance.

Descriptive statistics of the crank work per revolution are graphically depicted in Figure 11.

During fixed-gear pedaling, more crank work (including losses attributable to moving parts and also acceleration or deceleration energy) was generated in unsteady phases P1 and P3 than in steady-state phase P2 (Fig. 11). Subtraction of individually determined work consumed by drive resistance Wroll (group average 10.5 ± 1.12 J) led to additional work (representing 18, 7, and 13% of Wroll) in phases P1, P2, and P3 respectively. Likewise, during free pedaling, additional work (representing 41, 9, and 35% of Wroll) occurred in P1, P2, and P3, respectively. Generally, significantly more additional work occurred in unsteady (P1, P3) than in steady phases (P2), except for P3 during fixed-gear pedaling. Comparing free versus fixed-gear pedaling conditions, significantly more additional work occurred during free than during fixed-gear pedaling in the unsteady phases P1 and P3 (Fig. 11). In the steady phase, there was no significant intercondition difference in additional work.

Back to Top | Article Outline

Leg-based internal work balance.

Descriptive statistics of the work generated (concentric), absorbed (eccentric), and the net muscle work per revolution occurring in the leg (bilateral average) during free and fixed-gear pedaling are graphically depicted in Figure 12.

Figure 12 shows that in both pedaling conditions, muscle work was generated as a result of considerable concentric and eccentric work, despite the fact that pronounced bidirectional energy exchange took place through the crank only in the fixed-gear pedaling condition. During free pedaling, concentric, eccentric, and net work showed valleylike courses during P1, P2, and P3 phases, respectively. Significantly more concentric, eccentric, and net work occurred in the unsteady phases P1 and P3 than in the steady phase P2 (Fig. 12). During fixed-gear pedaling, significantly more concentric and eccentric work occurred in the unsteady phase P1 than in the plateau phaseP2 (Fig. 12). Similarly, eccentric work in unsteady phase P3 was significantly higher than in the plateau phase. Netmuscle work did not show any significant phase-dependency in the fixed-gear pedaling condition.

A comparison of free versus fixed-gear pedaling conditions showed that concentric work occurring in unsteady phases P1 and P3 was significantly higher in free pedaling than in the forced smooth condition (Fig. 12). In the steady phase P2, there was no significant intercondition difference in concentric work. Eccentric work showed a significant intercondition difference only in phase P1. Net work produced in the unsteady phases P1 and P3 during the free condition significantly exceeded the net work produced in the fixed-gear cycling condition.

Comparison of muscle work with crank work in both cycling conditions revealed no significant differences in any phase or condition (Appendix 11). Also, comparisons of 5-s sample intervals gave maximal deviations of less than 4.8% of muscle work.

Back to Top | Article Outline

Joint-based internal work balance.

Descriptive statistics of extensor and flexor work per revolution, generated (concentric) and absorbed (eccentric) in the knee and hip joints, are graphically depicted in Figure 13 (free pedaling) and Figure 13B (fixed-gear pedaling). The convention adopted in these figures revealed that distribution patterns characteristic of pedaling conditions of work produced by GMM occurred for all cycling phases P1, P2, and P3. Regarding joint-referred net work, knee extensors and flexors predominantly generate work irrespective of pedaling conditions (Fig. 13A and B), whereas hip extensors and flexors predominantly absorb work. Comparison of Fig. 7A and B allowed us to define GMM groups (diagonally located in Figs. 13 A and B) producing work mainly simultaneously. In the power phase of pedaling, knee extensors generated work, and hip flexors absorbed work. In the early recovery phase, only knee flexors generated energy. Finally, in the late recovery phase, hip flexors generated work and knee extensors absorbed work.

Figure 14A and Brelate the three described group mechanisms to revolution cycle phases. While free pedaling, the knee extensor work generated in the power phase during the unsteady phase P1 significantly exceeded (P < 0.002) that generated in phase P2. Likewise, during fixed-gear pedaling, work generated by the knee extensors in the power phase during P1 significantly exceeded that generated during P2 (P < 0.003). Knee extensors generated a significantly larger amount of work during the fixed-gear pedaling power phase in all cycling phases (P1, P2, and P3) than during free cycling (P < 0.001).

Hip flexors absorbed work in the power phase during free cycling in all phases P1, P2, and P3. Because they act in the power phase along with their counterpart, the concentric working knee extensor, the net work generated by this group in the power phase was small (Fig. 14A). The hip flexors absorbed less energy during the power phase of fixed-gear pedaling than in the corresponding cycling phases during free pedaling. Therefore, the hip flexors did not abolish the effect of knee extensors in the power phase during fixed-gear pedaling (Fig. 14B).

The hip flexors absorbed significantly more energy during the power phase of the unsteady P1 phase than in the steady P2 phase (free pedaling: P < 0.001, fixed-gear pedaling: P < 0.002); the concentric/eccentric values begin to differ more from each other in the unsteady phase P1.

Regarding the early recovery phase, only the knee flexors in the P1 phase of free cycling generated significant nonzero work (P < 0.001); during all other phases in both free and fixed-gear cycling conditions, no significant work was generated by knee flexors or absorbed by hip extensors.

Considerable amounts of work were generated in the late recovery phase during free pedaling by the hip flexor in the P1, P2, and P3 phases, respectively. Although this work was partially absorbed by the knee extensors, 96% of total net work was generated in the late recovery phase (averaged over P1, P2, and P3; Fig. 14A). In contrast, the amount of work generated by the hip flexors during fixed-gear pedaling in late recovery phase was completely abolished by the eccentrically working knee extensors in all phases P1, P2, and P3. Therefore, during fixed-gear pedaling, net work was generated entirely in the power phase (Fig. 14B).

Corresponding to the different work-producing profiles, significant interconditional (free vs fixed-gear cycling) differences existed between net work components during the power and late recovery phases (P < 0.001).

Back to Top | Article Outline


Back to Top | Article Outline

General comments.

The aim of Harrison's experiments (11) was to maximize human power over a short time by means of forced smooth rowing of able-bodied subjects (like the sprint phase of racing rowers). Our goal was to analyze the influence of forced smooth pedaling on the functional output measured as covered distance during outdoor FES cycling by SCI persons. The mean and peak power did not significantly vary between conditions in the steady P2 phase in our experiments, but first trial distance, total distance, and cycling time during the first trial were all significantly longer for forced smooth cycling than for free cycling. Therefore, if covered distance and cycling time are considered measures of endurance (18), in the setup of outdoor cycling, the forced smooth movement enhances endurance rather than power.

Because the treatment effect (freewheel/fixed-gear mechanism) applied in the present study was not randomly assigned to each subject, interference effects, mainly fatigue and carryover (training effects), cannot be completely excluded (14). The literature (17,18) and also preliminary work have shown that to achieve significant power increase, FES ergometrical training of SCI subjects has to be done daily or at least three times a week. Complete recovery of muscle condition typically can be expected after 1 d. Although no training or fatigue effect can occur after 7-28 d of rest between experiments, to diminish the learning effects technical improvements like "hip pushing" during cycling, the forced smooth cycling experiments that had higher output in preliminary experiments were performed first, and the free cycling experiments were performed later. Therefore, this experimental setup diminishes the studied effect rather than increases it.

A recent study (27) shows that the introduction of an additional degree of freedom (by releasing the ankle joint) hardly increases FES cycling power. By making it impossible to decouple the legs from high inertial load (Appendix 12), we actually removed the additional degree of freedom of the rider-tricycle system, which occurs during freewheeling. The results of our study fit the theoretical notion underlying the study mentioned: if a particular behavior (freewheeling) is considered undesirable, it is best to make it mechanically impossible.

Back to Top | Article Outline

External work balance.

The external balance analysis operates in this paper according to the notion of additional work, thus encompassing both losses attributable to moving parts and energy absorbed or generated by linear acceleration or deceleration (in the nonsteady case). In both cycling modi and during the steady-state phase P2, no significant difference in terms of the additional work was detected (Fig. 11), although the steady state phase P2 made the decisive contribution to prolonging distance in the first trial (Fig. 2 A,B). This indicates that from the viewpoint of external energy analysis, the mechanism, which accounts for the major advantage of forced smooth cycling over free cycling, occurs earlier in the initial (P1) phase of cycling. As a matter of fact, significantly higher additional energy occurred in phase P1 than in P2, under both cycling conditions. To decide whether the moving part losses increased (impact losses attributable to jerk) or simply the inertial part increased the additional work, we considered the P3 phase in both conditions (Fig. 4A, B). Additional work would be expected to decrease as a result of deceleration of the tricycle (equation 1). Nevertheless, because increasing additional work occurred, losses attributable to moving parts might have increased. We propose calling this nonmuscle increase that is technical in origin "losses attributable to jerk and nonsmoothness."

Therefore, from the viewpoint of external work balance, because the initial phase P1 during forced smooth cycling is less fatiguing than during free cycling, the FES cyclist using the forced smooth cycling modus is able to produce a superior functional output in the subsequent phase P2. Because phase-related ratios of free distance per forced smooth distance do not differ significantly from each other (P > 0.5, in one-way ANOVA with the factor phase (P1/P2/P3), the fatigue-ameliorating advantage achieved in phase P1 might work continuously during the rest of the trial. Nevertheless, significant interconditional differences in nonsmoothness and freewheeling found in steady phase P2 suggest the existence of internal balance related factors that act continuously in P2 and contribute to superior functional output during fixed-gear cycling.

Back to Top | Article Outline

Leg-based work balance.

During both pedaling conditions, the muscle work related to one leg results from the occurrence of considerable concentric (positive) and eccentric (negative) work (Fig. 12). The eccentric work/concentric work ratios are in both conditions and all cycling phases by far higher (averaged over all phases and conditions 72%) than in the case of able-bodied cyclists (8.5%) (4). If this is caused by work transfer between knee and hip joint by biarticular muscles (e.g., the rectus femoris or long head of biceps femoris muscles), increased tension and more fatigue of these muscles would result. Moreover, the increasing disparity between concentric/eccentric values becomes greater in the unsteady phases P1 and P3 (larger amounts of concentric and eccentric work) than in the steady P2 phase during the free pedaling condition. Thus, this seems to be a mechanism that compensates for the increased demand of crank work under this condition.

Another theoretically possible source of energy to be transferred to the crank besides muscle work is the net variation of passive (kinetic and potential) energy of the legs during one revolution (equation 2).

Nevertheless, no significant variations of passive energy were detected in this study, neither in freewheeling nor in fixed-gear conditions. Therefore, crank power has to be produced at every instant by muscle power, irrespective of the condition.

Back to Top | Article Outline

Joint-based work balance.

We showed in the present study that the production of cycling power during both conditions is based on the occurrence of three synergistic GMM groups: concentric knee extensor and eccentric hip flexor (power phase), concentric knee flexor (early recovery phase), and concentric hip flexor and eccentric knee extensor (late recovery phase). Figure 7A, B suggest that power phase and early recovery phase GMMs occur because of to quadriceps and hamstrings/gluteus stimulation, respectively.

The occurrence of GMMs in the late recovery phase needs to be interpreted in more detail. We suggest that it appears at least partly because of the extreme slowing of cadence in this phase, and therefore the quadriceps stimulation occurs too early (Figs. 6, 7A). This would be a consequence of imperfect stimulation, and it would be supported by the fact that in parallel with the deeper cadence break-in, significantly more work is produced by the late recovery phase during free pedaling, than during fixed-gear pedaling (Fig. 14A, B). Analogous to this imperfect stimulation hypothesis for FES cycling of SCI subjects, a report in the literature (9) claims that lower inertial loads (corresponding to freewheeling) are also harder for untrained healthy riders (imperfect neural drive) to control than are high inertial loads.

Two different patterns of producing cycling power, corresponding to the two cycling conditions, were identified in this study:

1. During free cycling, work per revolution (100%) was distributed percentually as −2, 6, and 96% (averaged over P1, P2, and P3) in power, early, and late recovery phases, respectively.

2. During fixed-gear pedaling, work per revolution (100%) was distributed percentually as 137, 0, and −37% (averaged over P1, P2, and P3) in power, early, and late recovery phases, respectively.

The distribution of power production during free cycling found in the present study is slightly similar to that observed during FES pedaling performed on an Ergys ergometer (8). These authors found net power absorption in the knee joint during FES cycling, and they hypothesized the existence of a dominating power generating source in the hip joint, on analogy to the hip flexor moment that produces the largest thrust (in terms of work per revolution cycle) in free cycling (Figs. 7A, 14A). It is not known whether the ergometer used allowed freewheeling or fixed gearing, but a different stimulation protocol and presumably less inertial load were used.

During fixed-gear pedaling, most of work was produced in the power phase by the quadriceps (Figs. 7B, 14B), and the positive part of this work was generated in the knee joint (this is analogous to the cycling of able-bodied subjects, (4)). Although a considerable negative part of work was absorbed by the hip joint during fixed-gear pedaling (in the case of able-bodied, the hip also generates work in power phase), we suspect that the more natural loading of the quadriceps in the power phase is an advantage for fixed-gear pedaling over free pedaling in terms of functional output.

Back to Top | Article Outline

Influence of body segment estimates on internal work balance.

Following recommendations given for the assessment of joint kinetics by inverse dynamics solutions during gait (21), we prefer anthropometric models, which provide body segment parameters closer to in vivo measurement (VM) (3,31), over cadaver-based models (CM) (30). Additionally, formulas given in VM (31) approximate shank inertia more adequately in persons with substantial atrophy (23), as is often found in SCI, than overestimating formulas provided by CM (30). Nevertheless, this may be partially offset by orthoses (23) that fix shank and foot to the pedals in FES cycling. Therefore to verify the robustness of differences found between freewheel and fixed-gear cycling, we investigated the influence of both modeling approaches, VM (31) and CM (30), on the internal work balance (Appendix 13).

Whereas the model effect on inverse dynamics during gait of healthy persons is reported (21) to be joint related (no effect on the knee and a significant effect on the hip), we found that only concentric hip flexor and eccentric knee extensor work (and moment) in FES cycling of SCI subjects were considerably influenced by the model, irrespective of fixed-gear (52-53% percentage of variation) or freewheel conditions (36-37% percentage of variation).

Although late-recovery phase net work was lower in CM (30) than in VM (31) based modeling, two different patterns of producing cycling power, similar to patterns found using the VM (31) model, were deduced during CM (30) based modeling as well. Therefore all conclusions derived above that refer to internal work balance differences between fixed-gear and freewheel pedaling remained qualitatively valid irrespective of the modeling approach.

Back to Top | Article Outline

Cadence break-in (valley) during the initial phase.

The cause of the cadence (power) break-in, which occurs in the initial phase (P1) of FES cycling (Fig. 2A, B), is not known. However, explanations such as (i) a temporal effect of spasticity (5) or (ii) of transient metabolical fatigue could be considered (26). The investigation of a strongly spastic (modified Ashworth scale 3-4) SCI subject not included in this study revealed an extremely pronounced cadence break-in in the initial (P1) phase of cycling. Therefore, we also favor the spastic origin hypothesis for the power valley.

However, the data of the present study show that the amplitude of cadence break-in amounts in freewheel and fixed-gear cycling do not significantly differ (23.5 ± 5.4 rpm and 19.3 ± 6.1, respectively, P > 0.1), but the duration of the initial phases (P1) do (P < 0.001). Therefore, the cycling mode (freewheel or fixed gear) influences in some way the shape of the power break-in. The spasticity explanation of the power break-in leads to the conclusion that fixed-gear cycling not only increases the effectiveness of movement (as in able-bodied subjects investigated by Harrison (11)) but also ameliorates spasticity-caused power break-in in SCI persons, allowing them to achieve longer distances.

Future work must definitively decide the origin of the power break-in, for instance, by using stimulation frequencies with strong antispastic effect in FES cycling.

Back to Top | Article Outline

Forced and/or smooth pedaling.

The approach adopted in this study is analogous to that of Harrison's (11) in that forced smooth movement is realized by kinematic constraints. We cannot decide on the basis of our experiments whether the superior outcome of fixed-gear pedaling is attributable to the fact that movement is forced or smooth. Fixed-gear pedaling is conditioned by the method, which is simultaneously smooth and forced. Nonmechanical (forced) smooth pedaling would be theoretically possible, for instance, by modulating the muscle-stimulation intensity, optimized to smoothness of cadence in an open loop (24) or feedback manner (2). Nevertheless, the rapid progress of muscle fatigue (25) would quickly necessitate stimulation with maximal current in FES cycling, thus making the application of such methods unfeasible.

Back to Top | Article Outline

Ergonomic and technical relevance.

The performance of forced smooth pedaling by SCI subjects on a fixed-gear tricycle offers other advantages than an energy-efficient cycling mode. Patients perceive the smooth, steady pedaling as comfortable and enjoy riding the fixed-gear tricycle. In contrast, a feeling of vulnerability may arise during the jerky and bucking (freewheel) cycling of a weak SCI subject; this does not occur with fixed-gear cycling. Furthermore, contrary to normal (freewheel) cycling, fixed-gear cycling offers the possibility of shunting backwards by stimulating the muscles in inverted sequence. Obviously starting from standstill (by manually pushing the leg in middle power phase position) is also easier on the fixed-gear cycle. One disadvantage, however, is the need for stronger brakes and more expensive speed gear systems.

Back to Top | Article Outline


Forced smooth pedaling is significantly superior to free pedaling regarding functional output (e.g., distance covered during the first trial and the total session increased, on average, by 68 and 103%, respectively).

The superiority of forced smooth cycling over free cycling can be explained by differences found in the external (load oriented) and internal (leg oriented) work balance of FES cycling.

Significantly more additional crank work is generated during nonsteady cycling phases to overcome increased losses attributable to moving parts (jerk losses) during freewheel than during fixed-gear pedaling. Generally, concentric and eccentric work increase significantly in unsteady phases, leading in free pedaling to actual net work increase to meet the rising crank work demand. During fixed-gear pedaling, the timing and joint location of muscle work generation are more similar to the cycling of able bodied subjects) than during freewheel pedaling, because during the former pedaling mode the bulk work is generated by the quadriceps in the knee joint during the power phase.

Certain energetic mechanisms that are advantageous for fixed-gear pedaling act predominantly in the unsteady cycling phases (occurrence of nonsmoothness losses and higher concentric work during freewheel cycling), and others work during all cycling phases (more advantageous joint localization and timing of power generation in case of fixed-gear pedaling).

This work was supported by the Else Kr¨ner-Fresenius Foundation. The authors would like to thank Ms. Judy Benson and Ms. Hedwig Fröhlich-Szecsi for copyediting the manuscript.

Back to Top | Article Outline


Stimulation intervals were individually determined by the method described in Perkins et al. (15). For a particular subject, the same stimulation intervals were used in freewheeling and fixed-gear condition tests. The stimulation intervals for patient 1 and for all study participants (mean ± SD) were as follows:

Table. No caption av...

These firing angle range limits contain the fix angular compensation of 28°. Hamstring and glutei muscle limits are set equal because of technical limitations of the stimulator.

Back to Top | Article Outline


Differences in transmission efficiency using different gearing mechanisms to achieve the same overall transmission ratio could occur (29) because of:

1. Efficiency was lower in fixed-gear condition, because smaller (less teeth) cogs and the taut chain entered and left the articulations under full tension (higher than in the freewheel case).

2. Efficiency was lower in freewheel condition because the speed hub used reduces efficiency by approximately 9-10% at low gearing.

Therefore, freewheeling was estimated to have approximately 5% more efficiency losses than fixed-gear pedaling.

Back to Top | Article Outline


The formula used for calculating average work done during one crank revolution against drive resistance:

where Fw = average measured towing force (6.63 ± 0.71 N), i = transmission ratio (1.06), rw = tricycle wheel (0.24 m), gave Wroll = 10.5 ± 1.12 J.

Back to Top | Article Outline


To delimit the plateau phase P2, a moving average over 5-s intervals was computed over the complete cadence course, and the maximum was found. Instances that corresponded to moving averaged cadences of 80% of the maximum were taken as begin and end of P2 (Fig. 2A, B).

Back to Top | Article Outline


Crank angle residual, defined as the variation in crank angle away from a linear function of time (9, was formed from the relationship

where θcrank is zero at the start of the cycle, t is the time from the start of the cycle, and tf is the period of the cycle.

Back to Top | Article Outline


Raw crank torque shows large positive and small negative (even in the freewheel case) peaks attributable to corresponding impact force oscillations and sensor noise (App. Fig. 1).

Back to Top | Article Outline


Apart from body weight (77.66 ± 13.53 kg; mean ± SD) and height (1.79 ± 0.08 m), further anthropometric data (segment length, mass, and moment of inertia) were calculated for the shank (0.41 ± 0.02 m, 5.86 ± 0.59 kg, and 0.1 ± 0.02 kg·m−2, respectively) and the thigh (0.54 ± 0.02 m, 10.87 ± 1.89 kg, and 0.21 ± 0.05 kg·m−2) of each participant using regression equations reported by VM (3,31).

Back to Top | Article Outline


Comparison of output power during the P2 phase in freewheel and fixed-gear cycling conditions.

Table. No caption av...
Back to Top | Article Outline


Comparison of nonsmoothness during the P2 phase in freewheel and fixed-gear cycling conditions.

Table. No caption av...
Back to Top | Article Outline


Comparison of freewheeling in freewheel and fixed-gear cycling conditions.

Table. No caption av...
Back to Top | Article Outline


Comparison of crank work and muscle work.

Table. No caption av...
Back to Top | Article Outline


Effective inertia was calculated in the case of patient 1 as follows:

where mB = mass of tricycle frame (35.5 kg), mC = mass of the rider (80 kg), mD = mass of tricycle wheel (1.5 kg), mF= mass of chainwheel, crank arms, pedals (2.1 kg), mG= mass of freewheel (0.3 kg), i = transmission ratio (1.06), ID = rotational inertia of tricycle wheel (0.18kg·m−2), IF = combined inertia of chainwheel, crank arms, pedals (0.032 kg·m−2), IG = inertial of freewheel (0.0003 kg·m−2), rw = radius of tricycle wheel (0.24 m).

Thus, we obtained Ieff = 8.3 kg·m−2. A further increase of the effective inertia by using higher transmission ratios was not possible, because this would cause effective drive resistance (seen from the crank) to also increase. Therefore, a part of the patients had to be excluded from the study (criterium: distance covered less than 400 m).

Back to Top | Article Outline


In line with the literature, thigh (21) and shank (23) inertia were found to be higher, respectively lower when estimated from living-based model (VM) (31) than from cadaver-based model (CM) (30):

Table. No caption av...

Regarding model effects on the inverse dynamics solution (App. Fig. 2), we found that only concentric hip flexor and eccentric knee extensor work (and moment) were considerably (> 10%) influenced by the model, irrespective of fixed-gear (53% variation) or freewheel conditions (36% variation). Thus, comparing VM (31) with CM (30): (i) the absolute magnitudes of net hip and knee work were decreased using VM (31) (App. Fig. 2), (ii) the concentric, eccentric, and net work produced by the late recovery GMM were increased using VM (31) (App. Fig. 3), and (iii) power and late recovery phase related net work preserved sign irrespective of modeling approach.

Although late recovery phase net work was lower in CM (30) than in VM (31), two different patterns of producing cycling power could be deduced during CM (30) as well. In this modeling approach, work per revolution during freewheeling cycling distributes percentually as 5, 28, and 67% and during fixed-gear pedaling, as 122, −2, and −20% in power, early, and late recovery phases, respectively.

Back to Top | Article Outline


1. Angeli, T., M. Gf¨hler, T. Eberharter, P. Lugner, L. Rinder, and H. Kern. Optimization of the pedal path for cycling powered by lower extremity muscles activated by functional electrical stimulation. In: Computer Methods in Biomechanics and Biomechanical Engineering-3, J. Middleton, M. L. Jones, N. G. Shrive, and G. N. Pande (Eds.). London, Gordon and Breach Science Publishers, pp. 263-268, 2001.
2. Chen, J. J., N. Y. Yu, D. G. Huang, B. T. Ann, and G. C. Chang. Applying fuzzy logic to control cycling movement induced by functional electrical stimulation. IEEE Trans. Rehabil. Eng. 5:158-169, 1997.
3. De Leva, P. Adjustments to Zatsiorsky-Seluyanov's segment inertia parameters. J. Biomech. 29:1223-1230, 1996.
4. Ericson, M. On the Biomechanics of Cycling. A Study of Joint and Muscle Load during Exercise on the Bicycle Ergometer [dissertation]. Karolinska Institute Stockholm, Department of Physical Medicine and Rehabilitation, 1986.
5. Eser, P. C., N. N. Donaldson, H. Knecht, and E. Stussi. Influence of different stimulation frequencies on power output and fatigue during FES-cycling in recently injured SCI people. IEEE Trans. Neural. Syst. Rehabil. Eng. 11:236-240, 2003.
6. Figoni, S. F., R. M. Glaser, M. M. Rodgers, et al. Acute hemodynamic responses of spinal cord injured individuals to functional neuromuscular stimulation-induced knee extension exercise. J. Rehabil. Res. Dev. 28:9-18, 1991.
7. Fitzwater, R. A personal user's view of functional electrical stimulation cycling. Artif. Organs 26:284-286, 2002.
8. Franco, J. C., K. L. Perell, R. J. Gregor, and A. M. Scremin. Knee kinetics during functional electrical stimulation induced cycling in subjects with spinal cord injury: a preliminary study. J.Rehabil. Res. Dev. 36:207-216, 1999.
9. Fregly, B. J. The Significance of Crank Load Dynamics to Steady-State Pedaling Biomechanics: An Experimental and Computer Modeling Study [dissertation]. Stanford University, Department of Mechanical Engineering, 1993.
10. Fregly, B. J., F. E. Zajac, and C. A. Dairaghi. Crank inertial load has little effect on steady-state pedaling coordination. J. Biomech. 29:1559-1567, 1996.
11. Harrison, J. Y. Maximizing human power output by suitable selection of motion cycle and load. Hum. Factors 12:315-329, 1970.
12. Hunt, K. J., B. Stone, N. O. Negard, et al. Control strategies for integration of electric motor assist and functional electrical stimulation in paraplegic cycling: utility for exercise testing and mobile cycling. IEEE Trans. Neural. Syst. Rehabil. Eng. 12:89-101, 2004.
13. Jacobs, P. L., and M. S. Nash. Modes, benefits, and risks of voluntary and electrically induced exercise in persons with spinal cord injury. J. Spinal Cord. Med. 24:10-18, 2001.
14. Neter, J., and W. Wassermann. Applied Linear Statistical Models, D. Richard (Ed.). Homewood, IL: Irvin, Inc., 1974.
15. Perkins, T. A., N. D. Donaldson, N. A. Hatcher, I. D. Swain, and D. E. Wood. Control of leg-powered paraplegic cycling using stimulation of the lumbo-sacral anterior spinal nerve roots. IEEE Trans. Neural Syst. Rehabil. Eng. 10:158-164, 2002.
16. Petrofsky, J. S., H. Heaton, and C. A. Phillips. Outdoor bicycle for exercise in paraplegics and quadriplegics. J. Biomed. Eng. 5:292-296, 1983.
17. Petrofsky, J. S., and M. Laymon. The effect of previous weight training and concurrent weight training on endurance for functional electrical stimulation cycle ergometry. Eur. J. Appl. Physiol. 91:392-398, 2004.
18. Petrofsky, J. S., and R. Stacy. The effect of training on endurance and the cardiovascular responses of individuals with paraplegia during dynamic exercise induced by functional electrical stimulation. Eur. J. Appl. Physiol. Occup. Physiol. 64:487-492, 1992.
19. Pollack, S. F., K. Axen, N. Spielholz, N. Levin, F. Haas, and K. T. Ragnarsson. Aerobic training effects of electrically induced lower extremity exercises in spinal cord injured people. Arch. Phys. Med. Rehabil. 70:214-219, 1989.
20. Pons, D. J., C. L. Vaughan, and G. G. Jaros. Cycling device powered by the electrically stimulated muscles of paraplegics. Med. Biol. Eng. Comput. 27:1-7, 1989.
21. Rao, G., D. Amarantini, E. Berton, and D. Favier. Influence of body segments' parameters estimation models on inverse dynamics solutions during gait. J. Biomech. 39:1531-1536, 2006.
22. Sinclair, P. J., G. M. Davis, R. M. Smith, B. S. Cheam, and J. R. Sutton. Pedal forces produced during neuromuscular electrical stimulation cycling in paraplegics. Clin. Biomech. 11:51-57, 1996.
23. Stein, R. B., E. P. Zehr, M. K. Lebiedowska, D. B. Popovic, A. Scheiner, and H. J. Chizeck. Estimating mechanical parameters of leg segments in individuals with and without physical disabilities. IEEE Trans. Rehabil. Eng. 4:201-211, 1996.
24. Szecsi, J., M. Fiegel, S. Krafczyk, and A. Straube. The smooth pedal stroke of the paraplegic cyclist-a computer simulation and experimental study. In: Proceedings of the 6th International Symposium on Computer Methods in Biomechanics and Biomedical Engineering [CD-ROM]. Madrid, Spain, 2004.
25. Szecsi, J., S. Krafczyk, A. Straube, and J. Quintern. Functional electrical stimulation propelled cycling of paraplegics-an occupational physiological approach. In: Proceedings of the 8th Vienna International Workshop on Functional Electrical Stimulaton. Vienna, Austria: Medical University of Vienna, 2004.
26. Theisen, D., C. Fornusek, J. Raymond, and G. M. Davis. External power output changes during prolonged cycling with electrical stimulation. J. Rehabil. Med. 34:171-175, 2002.
27. van Soest, A. J., M. Gf¨hler, and L. J. Casius. Consequences of ankle joint fixation on FES cycling power output: a simulation study. Med. Sci. Sports Exerc. 37:797-806, 2005.
28. White, D. C. S. Muscle mechanics. In: Mechanics and Energetics of Animal Locomotion, M. N. Alexander and G. Goldspink (Eds.). London, UK: Chapman and Hall, pp. 23-56, 1977.
29. Wilson, D. G. Bicycling Science. Cambridge, MA: MIT Press, 2004.
30. Winter, D. A. Biomechanics and Motor Control of Human Movement. New York, NY: John Wiley, 1990.
31. Zatsiorsky, V., and V. Seluyanov. The mass and inertial characteristics of main segments of the human body. In: Biomechanics VIII-B, H. Matsui and K. Kobayashi (Eds.). Champaign, IL: Human Kinetics, pp. 1152-1159, 1983.


©2007The American College of Sports Medicine