Survival Analysis Package

A Python package for analyzing survival data with a focus on body composition assessment. It was designed to utilize the results obtained by the BOA - Body and Organ Analysis workflow. In this repository we provide tools to reorganize the result of this algorithm to merge it to the patient table, add tools for data cleaning and a lifelines wrapper for automatical explorative anaylsis on survival outcomes given the Body-Composition results.

Features

• Survival Analysis: Cox proportional hazards regression and Kaplan-Meier survival curves
• Body Composition Analysis: Tools for processing and analyzing BCA data
• BOA Extractor: Command-line tool for extracting measurements from BOA data
• Data Preprocessing: Utilities for cleaning and preparing survival data
• CLI Tools: Command-line utilities for data merging, format conversion, and PDF encryption

Installation

pip install bca-survival

Usage

Basic Survival Analysis

from bca_survival.analyzer import BCASurvivalAnalyzer

# Load your data, sharing the same identifiers
df_main = pd.read_csv('clinical_data.csv')
df_measurements = pd.read_csv('bca_measurements.csv')

# Initialize the analyzer
analyzer = BCASurvivalAnalyzer(
    df_main, df_measurements,
    main_id_col='patient_id', measurement_id_col='id',
    start_date_col='diagnosis_date', event_date_col='event_date', event_col='event_status'
)

# Perform univariate analysis
columns = ['l5::WL::imat::mean_ml', 'l5::WL::tat::mean_ml', 'age', 'gender']
results = analyzer.univariate_cox_regression(columns)

# Generate Kaplan-Meier plot
analyzer.kaplan_meier_plot('l5::WL::imat::mean_ml', split_strategy='median')

# Perform multivariate analysis
model = analyzer.multivariate_cox_regression(columns)

Command-Line Tools

The package includes several command-line tools for common data processing tasks:

BOA Extractor

Extract measurements from BOA (Body Composition Assessment) data:

boa-extract /path/to/data /path/to/output

Purpose: Processes BOA data files and extracts relevant measurements for survival analysis.

Arguments:

• data_path: Path to the directory containing BOA data files
• output_path: Path where extracted measurements will be saved

---

BCA Merger

Merge two Excel files based on ID columns:

bca-merge <first_file> <second_file> <id_column_name>

Purpose: Combines clinical data with body composition measurements by matching on ID columns.

Arguments:

• first_file: Path to the first Excel file (e.g., clinical data)
• second_file: Path to the second Excel file (e.g., BCA measurements)
• id_column_name: Name of the ID column in the first file to match with 'StudyID' in the second file

Example:

bca-merge clinical_data.xlsx bca_measurements.xlsx patient_id

Output: Creates a file named {first_file}_merged.xlsx with:

• All rows from both files (outer join)
• Matched records combined into single rows
• Date columns formatted as DD.MM.YYYY
• No duplicate StudyID columns

Notes:

• The second file must have a column named 'StudyID'
• Uses outer merge to preserve all data from both files
• Automatically removes duplicate ID columns

---

Survival Result Converter

Convert Excel files to multiple formats (PDF, CSV, TXT):

survival-result-converter [directory]

Purpose: Batch converts Excel files to multiple formats for reporting and data sharing.

Arguments:

• directory: Directory to scan for Excel files (default: current directory)

Example:

# Convert all Excel files in current directory
survival-result-converter

# Convert Excel files in specific directory
survival-result-converter /path/to/results

Output Structure:

directory/
├── PDF/
│   ├── file1.pdf
│   └── file2.pdf
├── CSV/
│   ├── file1.csv
│   ├── file2_sheet1.csv
│   └── file2_sheet2.csv
└── TXT/
    ├── file1.txt
    └── file2.txt

Features:

• Recursively processes all .xlsx files in the directory tree
• Creates separate output folders (PDF, CSV, TXT)
• For multi-sheet Excel files:
  • PDF: All sheets in single file
  • CSV: Separate file per sheet
  • TXT: All sheets in single file with separators
• PDF generation supports two methods:
  • Windows: Uses COM automation for high-quality output
  • Cross-platform: Uses fpdf library with automatic column sizing

PDF Features:

• Landscape orientation for better table visibility
• Automatic column width adjustment
• Fits tables to page width
• Handles large tables (up to 1000 rows per sheet)
• Text wrapping for long content

---

PDF Report Extractor

Encrypt and organize PDF files from a directory tree:

pdf-report-extractor <input_path> <output_path> <password>

Purpose: Finds PDF files in a directory structure, copies them with standardized names, and encrypts them for secure distribution.

Arguments:

• input_path: Root directory to search for PDF files
• output_path: Destination directory for encrypted PDFs
• password: Password to encrypt the PDFs with

Example:

pdf-report-extractor /data/patient_reports /encrypted_reports MySecureP@ss123

Behavior:

• Recursively searches for all .pdf files
• Copies files to destination with naming pattern: encrypted_{parent_folder_name}.pdf
• Encrypts each file using user password protection
• Requires pdftk to be installed

Check pdftk Installation:

pdf-report-extractor --check-pdftk

Installing pdftk:

• Ubuntu/Debian: sudo apt-get install pdftk
• macOS: brew install pdftk-java
• Windows: Download from PDFtk website

Output Summary:

Processing: /data/patient_reports/folder1/report.pdf
  -> /encrypted_reports/encrypted_folder1.pdf
  -> Encrypted successfully

Processing complete:
  - Files processed successfully: 15
  - Errors: 0

Notes:

• Original files remain unchanged
• If encryption fails, the unencrypted copy is removed from destination
• Parent folder name is used for output filename (one level up from the PDF)

---

Documentation

Refer to the documentation in the docs/ directory for detailed information:

1. Install the package with documentation dependencies:

   pip install -e ".[docs]"

2. Build the documentation on Windows:

   cd docs
   make.bat html

Or on Linux/macOS:

   cd docs
   make html

3. Open docs/build/html/index.html in your browser

Development

Clone the repository and install in development mode:

git clone https://gitlab.com/your-group/survival-analysis.git
cd survival-analysis
pip install -e ".[dev]"

Requirements

Core Dependencies

• pandas
• openpyxl (for Excel file handling)
• lifelines (for survival analysis)

Optional Dependencies

• For PDF conversion (survival-result-converter):
  • Windows: pywin32
  • Cross-platform: fpdf, openpyxl
• For PDF encryption (pdf-report-extractor):
  • pdftk (external dependency)

Common Workflows

Workflow 1: Complete Data Processing Pipeline

# 1. Merge clinical and BCA data
bca-merge clinical.xlsx measurements.xlsx PatientID

# 2. Perform survival analysis (Python)
# ... (use BCASurvivalAnalyzer)

# 3. Convert results to multiple formats
survival-result-converter ./results

# 4. Encrypt PDF reports for distribution
pdf-report-extractor ./results/PDF ./encrypted_reports SecurePassword123

Workflow 2: Quick Data Conversion

# Convert a directory of Excel results to PDF
survival-result-converter /path/to/results

# PDFs are created in /path/to/results/PDF/