Force Analysis – Gimbal Experiment

Author: Federico Tessari, PhD
Affiliation: Mechanical Engineering Department, MIT
Contact: ftessari@mit.edu
Last Update: October 16, 2025

---

Overview

This repository contains the full MATLAB analysis pipeline for the Gimbal Experiment, designed to study maximum voluntary force (MVF) and force coordination under different arm configurations and task conditions.
The script reproduces all figures and statistics reported in the related manuscript (Figs. 7–10 and Supplementary Figs. 12–20).

The analysis is fully automated: simply run the main .m file to import, process, and visualize data for all subjects.
All the required data are already included in this repository.

---

What the Script Does

The script executes the following stages:

1. Setup and Formatting
   Sets MATLAB’s figure, font, and interpreter defaults for publication-quality plots.

2. Data Loading

   • Imports per-subject MVF data (S#_MVF_i.txt, S#_MVF_f.txt)
   • Imports all trial data for both conditions (CN, UCN) and all tasks (FLAT, LCK, FEX, FEXURD, ULCK)
   • Automatically trims the last 3 s of each trial for steady-state analysis

3. Computation of Force Metrics

   • Calculates instantaneous and median total force magnitude ‖F‖
   • Computes per-trial, per-subject, and pooled values across repetitions

4. Statistical Analysis

   • Performs 2-way ANOVA (Condition × Task) per subject with Bonferroni-corrected post-hoc tests
   • Computes Cohen’s d effect sizes for relevant comparisons
   • Conducts paired t-tests between Pre- and Post-MVF conditions

5. Visualization

   • Single-Subject Analysis (Figs. 7, S12–S20): time-series plots for total and component forces (Fx, Fy, Fz)
   • Per-Subject Boxplots (Fig. 8): distributions across tasks and conditions
   • Histogram Summaries (Fig. 9): median differences between configurations
   • Pre vs Post Comparisons (Fig. 10): boxplots aligned by subject with significance reporting

---

Folder Structure

├── Data/ # Folder containing all raw text files │ ├── S1_MVF_i.txt │ ├── S1_MVF_f.txt │ ├── S1_CN_FLAT_T1.txt │ └── ... ├── ForceAnalysis_Gimbal.m # Main analysis script └── README.md # This file

---

Input File Naming Convention

Type	Example	Description
MVF (Initial)	S1_MVF_i.txt	Pre-experiment maximum voluntary force
MVF (Final)	S1_MVF_f.txt	Post-experiment maximum voluntary force
Trial	S1_CN_LCK_T3.txt	Subject 1, Condition CN (Undesired), Task LCK, Trial 3

Each file must contain time-series data with at least the first three columns being Fx, Fy, Fz [N].

---

Key Variables

Variable	Meaning
condition	{'CN','UCN'} — Undesired / Preferred arm configurations
task	{'FLAT','LCK','FEX','FEXURD','ULCK'}
fs	Sampling frequency (default 50 Hz)
idx_min, idx_max	Index range used for 3 s steady-state window
MVF_tot_i, MVF_tot_f	Subject-level MVF stats (mean / std of ‖F‖)
f_tot_med	Median force magnitude per trial
medianCondDiff	CN − UCN difference in median force
T_summary	Table summarizing Pre / Post MVF results

---

Output and Figures

When run, the script automatically generates:

• Per-Subject Figures – time-series plots (Undesired vs Preferred)
• Boxplots – task-wise distributions across conditions
• Histograms – median differences in absolute / percentage terms
• Summary Table – Pre vs Post MVF differences with t-test p-values
• ANOVA Post-hoc Tables – stored as variables (resultsTable{subj}) with effect sizes

---

Dependencies

• MATLAB R2022a or newer
• Statistics and Machine Learning Toolbox (for anovan, multcompare, ttest2)
• No particular hardware required
• Tested configuration: Windows 11, Intel i7 11th Gen CPU, 16 GB RAM

---

Performance and Setup

Parameter	Detail
Typical running time	~120 seconds (≈ 2 minutes)
Installation time	None (after MATLAB is running)
Hardware requirements	None beyond a standard laptop/desktop
Operating system tested	Windows 11
Data availability	All data files included in Data/

---

How to Run

1. Ensure MATLAB is installed and running.
2. Place all data files inside a folder named Data in the same directory as the script.
3. Open ForceAnalysis_Gimbal.m in MATLAB.
4. Press Run, or execute:

run('ForceAnalysis_Gimbal.m')