⚡ electrical engineering

Sensitivity of Modeled L5–S1 Joint Reaction to Virtual Hand-Wrench Translation During a Squat: A Single-Trial OpenSim Study

This single-trial OpenSim study quantifies the near-linear sensitivity of modeled L5–S1 joint reaction forces to virtual translations of hand-applied external wrenches during a squat, while explicitly noting that the findings are limited to internal model consistency and do not generalize to physical load relocation, injury risk, or broader populations.

Original author: Jone Maycon Monteiro¹

Published 2026-11-01

📖 4 min read☕ Coffee break read

CC BY 4.0

Original author: Jone Maycon Monteiro¹

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the author. For technical accuracy, refer to the original paper. Read full disclaimer

When a person lifts a heavy object, their body is not just a static pillar holding up weight; it is a complex machine of bones, muscles, and joints working together to keep balance. The lower back, specifically the junction where the spine meets the pelvis, bears a tremendous amount of this burden. Scientists who study human movement use powerful computer programs to simulate these actions, creating digital avatars that can predict how much force is pressing on the spine during a lift. These simulations are valuable because they allow researchers to test "what if" scenarios that would be impossible or unsafe to try on real people. By changing one small detail in the computer model while keeping everything else exactly the same, they can isolate exactly how that single change affects the forces inside the body. This kind of precise, controlled testing helps separate the effects of a specific movement from the many other factors that change when a person actually lifts a box.

In a recent technical study, a researcher used this approach to investigate how the position of a load in a person's hands changes the force on the lower back during a squat. The study focused on a single, recorded trial of a person lifting a seven-kilogram box with both hands. Using a detailed computer model of the human body, the researcher did not ask the person to move differently or lift a different weight. Instead, the researcher performed a virtual experiment. They kept the person's actual body movements and the forces their feet pushed against the floor exactly as they were recorded. The only thing that changed was a mathematical adjustment to the "twist" or rotational force applied by the hands holding the box. The researcher virtually shifted the point where the hands applied this force forward, backward, left, and right by specific distances, calculating how the computer model's predicted back force would respond to these imaginary shifts.

The results showed a clear and predictable pattern. When the virtual force was shifted fifty millimeters forward, the peak force on the lower back joint increased by about eighty-eight newtons. When it was shifted fifty millimeters backward, the force decreased by almost the exact same amount. Shifting the force by one hundred millimeters doubled these effects, increasing or decreasing the peak force by roughly one hundred seventy-seven newtons. The relationship was so consistent that if you plotted the five different shifts on a graph, they formed a nearly perfect straight line. The study also checked if moving just one hand produced half the effect of moving both, and it did. These findings confirmed that within this specific computer model, the system reacted to the virtual shifts with high internal consistency and mathematical precision.

However, the study is careful to define what these numbers mean and, just as importantly, what they do not mean. The researcher explicitly states that this is a simulation of a single trial, not a measurement of what happens to a real person's spine in the real world. The computer model predicted a peak force of about 5,643 newtons for the baseline lift, which is roughly seven times the person's body weight. While the changes caused by the virtual shifts were clear and proportional, the study does not prove that shifting a real box in real life would change the actual pressure on a person's spinal discs in exactly this way. The computer model had to rely on some mathematical "residual" forces to stay balanced, which suggests the simulation was not a perfect representation of biological reality. Therefore, the results serve as a benchmark for the reliability of the computer code itself, showing that the software behaves consistently when fed specific inputs, rather than providing a definitive medical answer about injury risk or the best way to lift a box.

Ultimately, this work is a demonstration of computational rigor. It proves that for this specific digital model, a small, virtual change in how a load is held results in a proportional and predictable change in the calculated force on the lower back. It does not, however, offer a new rule for how humans should lift, nor does it confirm that the spine experiences these exact forces in reality. The study stands as a transparent, reproducible check on the tools scientists use to understand human movement, ensuring that when they do ask complex questions about lifting and injury, the computer models they rely on are responding to changes in a logical and consistent manner.

Technical Summary: Sensitivity of Modeled L5–S1 Joint Reaction to Virtual Hand-Wrench Translation

Problem Statement
In musculoskeletal modeling, isolating the specific effect of a change in external load placement is difficult because experimental comparisons of squat techniques inherently mix changes in applied loads with adaptive changes in kinematics, posture, and muscle recruitment. While simulations allow for controlled computational perturbations, there is a need to distinguish between modeled joint reactions and actual internal tissue loading. This study addresses the methodological challenge of quantifying how a precisely specified virtual translation of hand-applied external wrenches affects the modeled L5–S1 joint reaction, while strictly holding measured kinematics and ground reaction forces constant.

Methodology
The study employed a deterministic, single-trial technical approach using OpenSim 4.4.1 and the FATLS APP1 full-body model. The source data consisted of a single two-handed, 7-kg box-squat trial (Participant P2).

Perturbation Strategy: Instead of physically moving a load or altering kinematics, the researchers applied a "virtual hand-wrench translation." The measured global hand-force vectors ( $F$ ) and application points were preserved. For each specified virtual offset ( $r$ ), a free moment ( $\Delta M = r \times F$ ) was added to the hand wrench.
Conditions: Fourteen accepted model conditions were evaluated:
- Bilateral Global-X: Offsets of $\pm 50$ mm and $\pm 100$ mm.
- Bilateral Global-Z: Offsets of $\pm 50$ mm.
- Anatomical Frame: A pelvis-axis offset of $+50$ mm (anterior).
- Unilateral: Left and right hand offsets of $-50$ mm.
- Reserve Sensitivity: Baseline and $X-50$ conditions tested at $0.5\times$ and $2\times$ reserve-actuator optimal-force scales.
Primary Outcome: The Euclidean resultant force ( $|F|$ ) of the L5–S1 joint reaction (specifically the reaction on lumbar5 expressed in the lumbar5 coordinate system).
Validation Checks: The study included checks for input identity, additivity (comparing unilateral sum to bilateral effects), symmetry, and robustness to reserve-actuator scaling. Residual forces and moments were compared against contextual reference thresholds (5% of external force and 1% of center-of-mass height times external force).

Key Results

Baseline: The unmodified model produced a peak L5–S1 resultant of 5,643.59 N (approx. 7.14 body weights) at $t = 2.0483$ s.
Global-X Sensitivity: Bilateral global-X translations demonstrated a nearly linear, dose-proportional response:
- $X -50$ mm: $+88.38$ N change ( $+1.57\%$ ).
- $X +50$ mm: $-88.39$ N change ( $-1.57\%$ ).
- $X -100$ mm: $+176.79$ N change.
- $X +100$ mm: $-177.11$ N change.
- The five-point slope was $-1,769.13$ N/m with a descriptive $R^2$ of $0.999999$.
Additivity and Symmetry: Unilateral $X-50$ mm offsets (Left: $+44.08$ N; Right: $+44.29$ N) summed to reproduce the bilateral change with an additivity index of $5.90 \times 10^{-5}$ . Opposite-sign symmetry indices were extremely low ( $4.3 \times 10^{-5}$ and $9.2 \times 10^{-4}$ ), indicating high implementation consistency.
Other Directions: Global-Z offsets ( $\pm 50$ mm) resulted in minimal changes ($-2.30$ N and $+2.72$ N), though these slightly exceeded the pre-declared expectation of $<1$ N. The pelvis-axis $+50$ mm offset yielded a $+87.97$ N change.
Reserve Scaling: Changing reserve-actuator scales ( $0.5\times$ to $2\times$ ) altered the absolute baseline peak significantly but maintained a stable sensitivity slope relative to the default (ratios of $1.042$ and $0.970$), falling within the predefined robustness convention of $0.90–1.10$ .
Residuals: The baseline residual-moment ratio exceeded the adopted reference thresholds (2.55 on the interior grid and 3.08 on the full grid), indicating the solver relied on residual moments beyond the chosen model-quality context.

Key Contributions
The primary contribution of this work is a bounded, auditable sensitivity benchmark for a specific musculoskeletal model instance. It demonstrates:

Internal Consistency: The model responds linearly and symmetrically to virtual wrench translations in the global-X direction.
Implementation Verification: The study validates the additivity of unilateral perturbations and the stability of the sensitivity metric across different reserve-actuator scalings.
Methodological Distinction: It clarifies that such computational perturbations isolate the model's response to external moment changes without simulating physical load relocation, posture changes, or altered muscle recruitment strategies found in real-world technique variations.

Significance and Claims
The paper explicitly frames its findings as a reproducible model-sensitivity benchmark rather than a discovery of new physical lifting mechanics or an estimate of injury risk.

Scope of Validity: The results characterize a single model/trial instance. They do not provide evidence regarding disc/contact forces, injury risk, other squat techniques, or population-level effects.
Limitations: The author emphasizes that the elevated residual-moment demand restricts the attribution of joint-reaction changes uniquely to muscles or internal structures. The study does not validate the absolute magnitude of the baseline reaction as a true anatomical tissue load.
Conclusion: The analysis confirms that for a fixed-input model instance, a virtual translation of hand wrenches generates a dose-proportional, directionally opposed change in the modeled L5–S1 reaction. This serves as a transparent technical reference for future computational studies but should not be transferred to physical exercises (e.g., Zercher squats) without specific kinematic and force validation.

Technical Summary: Sensitivity of Modeled L5–S1 Joint Reaction to Virtual Hand-Wrench Translation

More like this