Meta Analysis Data Medic

Raw clinical trial data rarely comes ready for statistical synthesis. It requires robust cleaning, transformation, feature engineering, and network modeling before yielding evidence-based clinical insights. This project builds a reproducible data pipeline for conducting Network Meta-Analyses (NMA) across multiple clinical datasets.

ETL & Statistical Preprocessing

Using SQL for querying and extracting records from database schemas, the pipeline processes heterogeneous datasets from clinical trials. Python libraries (Pandas and NumPy) are utilized to handle missing value imputation, outlier filtering, and data restructuring, transforming raw study-level records into consistent, analysis-ready tabular data.

Network Meta-Analysis Engine

The core statistical analysis is performed using R and MetaInsight. Rather than traditional pairwise comparisons, the Network Meta-Analysis (NMA) engine evaluates multiple antidepressant treatments simultaneously. It models network geometry, performs consistency and heterogeneity checks, and ranks treatment efficacy through probability distribution metrics.

Architecture

Deliverables & Impact

The pipeline outputs publication-ready tables and figures, including forest plots, funnel plots for publication bias, and treatment ranking curves. By automating data ingestion, preprocessing, and model execution, the pipeline reduces time-to-insight for systematic literature reviews (SLR) and ensures full reproducibility of clinical trial synthesis.