---
title: India TB Missed Cases Analysis & Living Model (2025)
license: cc0-1.0
task_categories:
- tabular-regression
- tabular-classification
tags:
- medical
- epidemiology
- india
- tuberculosis
- bayesian
- mcmc
- causal-inference
- pca
pretty_name: India TB Missed Cases Analysis (2025)
size_categories:
- n<1K
---

# India TB Missed Cases Analysis & Living Model (2025)

![TB Analysis Framework](https://huggingface.co/datasets/hssling/india-tb-missed-cases-analysis/resolve/main/dag_causal_tb_analysis.png)

## 🌟 Project Overview
This repository hosts a comprehensive, multi-method analytical framework designed to estimate and understand the "missing" millions of Tuberculosis (TB) cases in India. By integrating Bayesian statistics, Dimensionality Reduction (PCA), and Causal Inference (DAG), this project provides a high-resolution view of TB detection determinants across Indian states.

### Core Analytical Pillars:
1. **Bayesian MCMC Estimation**: Hierarchical probabilistic modeling to quantify national and state-level missed cases with 95% Credible Intervals.
2. **Principal Component Analysis (PCA)**: Construction of data-driven 'System Strength' and 'Risk Burden' indices, explaining 2.8x more variance than traditional indicators.
3. **Causal Directed Acyclic Graphs (DAG)**: Structural mapping of 36 causal pathways across 26 variables to identify intervention priorities.

---

## 🚀 The Living Model (`living_tb_analysis_model.ipynb`)
The centerpiece of this project is a **"Living" Bayesian Notebook**. Unlike static reports, this model is designed for continuous refinement:
- **Autonomous Updates**: Fetches latest data from WHO Global Health Observatory APIs and Ni-kshay reports.
- **Incremental Learning**: Uses the project's 2023 estimates as *priors*, fine-tuning the posterior as 2024/2025 data becomes available.
- **Policy Priority Matrix**: Automatically regenerates state-specific intervention strategies based on the latest data.

---

## 📊 Dataset Contents
### 1. Primary Analysis Data
- `pca_integrated_analysis.csv`: The final merged dataset with PCA components.
- `integrated_mcmc_system_risk.csv`: Pre-calculated indices for system-risk correlation analysis.
- `state_missed_cases_latest.csv`: Final state-level estimates for 2023.

### 2. Raw Harmonized Sources
- `nfhs5_state_agg.csv`: Aggregated nutrition and risk behavior indicators from India's NFHS-5 survey.
- `who_india_ts.csv`: Historical TB incidence and notification trends for India (2000-2023).

### 3. Model Results (JSON)
- `mcmc_bayesian_results.json`: Summary of posterior distributions.
- `mcmc_missed_cases_sensitivity_results.json`: Sensitivity analysis findings.

---

## 🛠️ Usage Instructions
### Academic Researchers
Load the CSV files into Python/R to study the relationship between healthcare infrastructure (`System_PC1`) and nutritional vulnerabilities (`Risk_PC1`).
```python
import pandas as pd
df = pd.read_csv('pca_integrated_analysis.csv')
# Analyze correlation between System PC1 and Missed Cases
print(df[['System_PC1', 'missed_cases_mean']].corr())
```

### Policy Makers
Refer to the **`living_tb_analysis_model.ipynb`** output (specifically the State Prioritization Section) to identify whether a state requires "System Strengthening" or "Risk Management" focus.

---

## 📖 Citation
```bibtex
@misc{siddalingaiah2025tb,
  author = {Siddalingaiah, H. S.},
  title = {Advanced Multi-Method Analysis of Missed Tuberculosis Cases in India},
  year = {2025},
  note = {MCMC-PCA-DAG Framework}
}
```

## ⚖️ License
This work is released under **CC0: Public Domain**. Use, modify, and distribute freely to support the global mission to **End TB**.