Chapter 1 Introduction

package-overview UKBAnalytica is a high-performance R package for processing UK Biobank (UKB) Research Analysis Platform (RAP) data exports. It provides efficient extraction of diagnosis records from multiple sources and generates Cox regression-ready survival datasets.

1.1 Key features

Built on data.table for efficient processing of large-scale biobank data.
Supports ICD-10, ICD-9, self-reported illness, and death registry data.
Flexible dual-source case definitions for main and sensitivity analyses.
Generates survival datasets with proper prevalent/incident case handling.
Includes predefined definitions for common cardiovascular and metabolic diseases.
Provides standardized preprocessing for common UKB baseline variables.
Advanced analysis modules: subgroup analysis, propensity score methods, mediation analysis, and multiple imputation pooling.

1.2 Installation

Install the development version from GitHub:

# install.packages("devtools")
devtools::install_github("Hinna0818/UKBAnalytica")

1.3 Package overview

The package covers the following workflow:

Data download – Python scripts for downloading data from the RAP platform.
Variable preprocessing – Standardized cleaning and recoding of baseline characteristics.
Disease phenotyping – Predefined and custom disease definitions from multiple coding systems.
Survival analysis – Build Cox-ready datasets with proper prevalent/incident classification.
Baseline tables – Create Table 1 summaries stratified by case/control status.
Multiple imputation – Impute missing covariates and merge results back to the full dataset.
Main analysis – Correlation analysis and regression models (linear, logistic, Cox).
Advanced analysis – Subgroup analysis, propensity score, mediation, MI pooling.

The remaining chapters walk through each step with code examples.

1.4 Core functions

Function	Description
`build_survival_dataset()`	Build wide-format survival dataset
`extract_cases_by_source()`	Extract cases from specified data sources
`get_predefined_diseases()`	Get predefined disease definitions
`create_disease_definition()`	Create a custom disease definition
`preprocess_baseline()`	Preprocess baseline characteristics
`create_baseline_table()`	Create a Table 1 (via `tableone`)
`run_imputation()`	Multiple imputation with `mice`

1.5 Advanced analysis functions

Function	Description
`run_subgroup_analysis()`	Stratified analysis with interaction p-values
`run_multi_subgroup()`	Batch subgroup analysis across multiple variables
`estimate_propensity_score()`	Estimate propensity scores
`match_propensity()`	Propensity score matching
`calculate_weights()`	IPTW weight calculation
`run_mediation()`	Causal mediation analysis
`pool_mi_models()`	Pool results from multiple imputation
`plot_forest()`	Forest plot visualization
`plot_km_curve()`	Kaplan-Meier curves

1.6 Machine learning functions

Function	Description
`ukb_ml_model()`	Train ML models (RF, XGBoost, GLMNet, SVM, NNet)
`ukb_ml_predict()`	Generate predictions from ML model
`ukb_ml_cv()`	Cross-validation for ML models
`ukb_ml_compare()`	Compare multiple ML models
`ukb_ml_metrics()`	Calculate performance metrics
`ukb_ml_roc()`	ROC curve analysis
`ukb_ml_calibration()`	Calibration curve analysis
`ukb_ml_confusion()`	Confusion matrix
`ukb_ml_importance()`	Variable importance
`ukb_shap()`	Compute SHAP values
`ukb_ml_survival()`	Survival ML models (RSF, GBM, CoxNet)
`plot_shap_summary()`	SHAP summary plot
`plot_shap_dependence()`	SHAP dependence plot

UKBAnalytica: A Guide to UK Biobank Data Analysis