lcrpm

🚀 Example Usage: Lung Cancer Risk Profiling

This example demonstrates how to use the lcrpm package to extract and analyze real-world healthcare data for lung cancer risk modeling using an OMOP Common Data Model (CDM) instance.

📦 Prerequisites

Ensure the package is installed and you have a properly configured .env file containing your database connection details.

🧰 Step 1: Initialization

import lcrpm
from sqlalchemy import inspect

# Load database configuration and create SQLAlchemy engine
config = lcrpm.load_config(".env")
engine = lcrpm.create_engine_from_config(config)

# Extract the time-to-event dataset
df = lcrpm.extract_tte_dataset(engine)
print(f"Dataset shape: {df.shape}")
print(df.head())

# Initialize the OMOP analyzer
analyzer = lcrpm.OMOPAnalyzer(config)

📈 Step 2: Kaplan–Meier Survival Plot

# Disease metadata
disease_name = "lung_cancer"
DB_NAME_TITLE = "Lung Cancer"
FIGURES_DIR = "./FIGURES"

# ICD codes for lung cancer
LC_ICD_Codes = [
    "162.0", "162.2", "162.3", "162.4", "162.5", "162.8", "162.9",
    "C33", "C34.0", "C34.00", "C34.01", "C34.02", "C34.1", "C34.10",
    "C34.11", "C34.12", "C34.2", "C34.3", "C34.30", "C34.31", "C34.32",
    "C34.8", "C34.80", "C34.81", "C34.82", "C34.9", "C34.90", "C34.91", "C34.92"
]

# Generate Kaplan–Meier plot
lcrpm.plot_km_simple(
    df,
    subgroup_label=DB_NAME_TITLE,
    title=f"Kaplan–Meier Survival Curves – {DB_NAME_TITLE}",
    output_png=f"{FIGURES_DIR}/kaplan_meier_professional.png",
    output_pdf=None,
    show=True,
    save=True
)

📊 Step 3: Comprehensive Data Analysis

# Inspect OMOP schema
inspector = inspect(engine)
schema_check = analyzer.inspect_omop_schema(inspector, engine, config['db_schema'])

# Perform descriptive analyses
df_age = analyzer.get_age_distribution(engine, config['federated_node_id'], save=True)
df_summary = analyzer.get_summary_statistics(engine, config['federated_node_id'], save=True)
df_demographics = analyzer.get_demographics_summary(engine, config['federated_node_id'], save=True)
df_locations = analyzer.get_location_distribution(engine, config['federated_node_id'], save=True)
df_visits = analyzer.get_visit_distribution(engine, config['federated_node_id'], save=True)
df_conditions = analyzer.get_condition_distribution(engine, config['federated_node_id'], save=True)
df_drugs = analyzer.get_drug_distribution(engine, config['federated_node_id'], save=True)

# Lung cancer-specific analyses
df_lc_conditions = analyzer.get_lc_condition_distribution(engine, config['federated_node_id'], LC_ICD_Codes, save=True)
df_incident = analyzer.get_incident_rates(engine, config['federated_node_id'], LC_ICD_Codes, save=True)
df_km = analyzer.get_kaplan_meier_data(engine, config['federated_node_id'], LC_ICD_Codes, save=True)

# Smoking status analysis
smoking_codes = [1568177, 45538046, 45595884, 45562025, 45542821, 45591117, 45600719, 1568178, 43021779]
df_smoking = analyzer.get_smoking_status_analysis(engine, config['federated_node_id'], smoking_codes, save=True)

📌 Step 4: Summary Report

print("\n===================== ANALYSIS SUMMARY =====================")
print(f"Schema Inspection: {'Passed' if schema_check else 'Failed'}")
print(f"Age Distribution: {len(df_age)} records")
print(f"Summary Statistics: {len(df_summary)} entries")
print(f"Demographics Summary: {len(df_demographics)} records")
print(f"Location Distribution: {len(df_locations)} locations")
print(f"Visit Distribution: {len(df_visits)} types")
print(f"Condition Distribution: {len(df_conditions)} codes")
print(f"Drug Distribution: {len(df_drugs)} records")
print(f"Lung Cancer Conditions: {len(df_lc_conditions)} codes")
print(f"Incident Rates: {len(df_incident)} time points")
print(f"Kaplan–Meier Data: {len(df_km)} rows")
print(f"Smoking Status Analysis: {len(df_smoking)} groups")
print("============================================================")

📁 Output

All analysis outputs (plots, tables, and figures) are saved in the local directory and can be used for further reporting or modeling.