A Shape-Based Functional Index for Objective Assessment of Pediatric Motor Function

Functional data analysis of arm curl and knock movement data using advanced statistical methods.

📋 Table of Contents

• Overview
• Features
• Installation
• Quick Start
• Data Structure
• Usage
• Results

🔍 Overview

This repository contains the complete analysis pipeline for our research on functional data analysis of movement patterns. The project focuses on:

• Movement analysis across multiple participants
• Functional data registration and alignment techniques
• Statistical modeling of temporal movement patterns
• Comparative analysis between different cohorts and time points

✨ Features

• 📊 Complete functional data analysis pipeline
• 📈 Advanced visualization with publication-ready plots
• 🔬 Statistical modeling using Partial Least Squares Analysis
• 📱 Jupyter notebook interface for interactive analysis
• 🔄 Reproducible research workflow
• 📋 Support for both arm curl and knock movement data

🚀 Installation

Prerequisites

• Python 3.7 or higher
• Jupyter Notebook or JupyterLab

Install Dependencies

# Clone the repository
git clone https://github.com/yourusername/u-extend.git
cd u-extend

# Install required packages
pip install fdasrsf seaborn statsmodels numpy pandas matplotlib scipy

Conda Environment (Recommended)

# Create a new conda environment
conda create -n uextend python=3.8
conda activate uextend

# Install packages
conda install -c conda-forge fdasrsf seaborn statsmodels numpy pandas matplotlib scipy jupyter

⚡ Quick Start

1. Launch Jupyter Notebook

   jupyter notebook paper_codes.ipynb

2. Run the Analysis

   • Execute all cells to reproduce arm curl results
   • Modify data paths for knock data analysis
   • View generated plots and statistical outputs
   • Full data is available upon request

Data Format Details

File	Description	Format	Example
ys_*	Time series gyroscope measurements	Numeric array	Movement trajectories
pids_*	Participant identifiers	Numeric array	[1, 2, 3, 4, ...]
visits_*	Visit dates	Datetime array	['7/30/24', '6/3/24', ...]
cohorts_*	Group assignments	String array	['Control', 'DMD', ...]

📖 Usage

Basic Analysis

import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
from fdasrsf import *

# Load data
ys = np.loadtxt('ys_curl_wodelta.txt')
pids = np.loadtxt('pids_curl_wodelta.txt')
visits = np.loadtxt('visits_curl_wodelta.txt', dtype=str)
cohorts = np.loadtxt('cohorts_new.txt', dtype=str)

# Run analysis pipeline
# [Analysis code follows...]

Reproducing Paper Results

1. Arm Curl Analysis: Run paper_codes.ipynb as-is
2. Knock Data Analysis:
   • Update file paths to knock data files
   • Follow same analysis pipeline
   • Compare results across movement types

📊 Results

The analysis generates:

• 📈 Functional trajectories showing movement patterns
• 📉 Mean functions for different groups and visits
• 🔍 Statistical comparisons between cohorts and time points
• 📋 Model summaries with parameter estimates and significance tests
• 🎨 Publication-ready visualizations using seaborn styling