Appendix A — Dataset Codebook

Dataset Codebook

This appendix describes all datasets used throughout the course. Each dataset is stored as a CSV file in the data/ directory.

A.1 Framingham Heart Study (Teaching Dataset)

File: data/framingham.csv Source: NHLBI Biologic Specimen and Data Repository (BioLINCC) teaching dataset. Also available via the R package riskCommunicator. Description: Prospective cohort study data from Framingham, Massachusetts. This is the classic cardiovascular epidemiology dataset used to develop the Framingham Risk Score.

Variable Type Description
age Continuous Age at baseline examination (years)
sex Binary 0 = Female, 1 = Male
bmi Continuous Body mass index (kg/m2)
sbp Continuous Systolic blood pressure (mmHg)
dbp Continuous Diastolic blood pressure (mmHg)
totchol Continuous Total cholesterol (mg/dL)
hdl Continuous HDL cholesterol (mg/dL)
glucose Continuous Fasting blood glucose (mg/dL)
smoking Binary 0 = Non-smoker, 1 = Current smoker
diabetes Binary 0 = No, 1 = Yes
bp_meds Binary 0 = Not on BP meds, 1 = On BP meds
chd_10yr Binary 10-year incident CHD (0 = No, 1 = Yes)
time_chd Continuous Time to CHD event or censoring (days)

Used in: Chapters 3, 4, 5, 10, 11, 19 (Capstone 1)

A.2 Wisconsin Diagnostic Breast Cancer (WDBC)

File: data/wdbc.csv Source: UCI Machine Learning Repository. Also available via sklearn.datasets.load_breast_cancer() in Python and mlbench::BreastCancer in R. Description: Features computed from digitised images of fine needle aspirates (FNA) of breast masses. Each row represents one tumour sample.

Variable Type Description
id Integer Patient ID
diagnosis Binary M = Malignant, B = Benign
radius_mean Continuous Mean of distances from centre to perimeter
texture_mean Continuous Standard deviation of grey-scale values
perimeter_mean Continuous Mean tumour perimeter
area_mean Continuous Mean tumour area
smoothness_mean Continuous Local variation in radius lengths
compactness_mean Continuous Perimeter2 / area - 1.0
concavity_mean Continuous Severity of concave portions
concave_points_mean Continuous Number of concave portions
symmetry_mean Continuous Tumour symmetry
fractal_dimension_mean Continuous “Coastline approximation” - 1
Plus _se and _worst variants of each Standard error and worst (largest) value

Total variables: 32 (ID + diagnosis + 30 features) Observations: 569 (357 benign, 212 malignant) Used in: Chapters 5, 8, 9, 15

A.3 Primary Biliary Cholangitis (PBC)

File: data/pbc.csv Source: Mayo Clinic trial in primary biliary cholangitis (formerly primary biliary cirrhosis). Available via survival::pbc in R. Description: Data from 418 patients enrolled in a randomised trial of D-penicillamine vs placebo for PBC, a chronic liver disease.

Variable Type Description
id Integer Patient ID
time Continuous Days from registration to death, transplant, or censoring
status Integer 0 = Censored, 1 = Transplant, 2 = Dead
trt Binary 1 = D-penicillamine, 2 = Placebo
age Continuous Age (years)
sex Binary 0 = Male, 1 = Female
ascites Binary Presence of ascites
hepato Binary Presence of hepatomegaly
spiders Binary Presence of spider angiomata
edema Ordinal 0 = None, 0.5 = Untreated/resolved, 1 = Despite treatment
bili Continuous Serum bilirubin (mg/dL)
chol Continuous Serum cholesterol (mg/dL)
albumin Continuous Serum albumin (g/dL)
copper Continuous Urine copper (µg/day)
alk_phos Continuous Alkaline phosphatase (U/L)
ast Continuous Aspartate aminotransferase (U/L)
trig Continuous Triglycerides (mg/dL)
platelet Continuous Platelet count
protime Continuous Prothrombin time (seconds)
stage Ordinal Histologic stage (1–4)

Used in: Chapters 4, 6, 14 (Capstone 3)

A.4 NHANES Subset

File: data/nhanes_subset.csv Source: National Health and Nutrition Examination Survey (CDC). Extracted using the nhanesA R package from NHANES 2017–2020 cycle. Description: A curated subset of NHANES data with demographics, lab results, and health indicators for dimensionality reduction and clustering exercises.

Variable Type Description
seqn Integer Respondent sequence number
age Continuous Age (years)
sex Binary 1 = Male, 2 = Female
race_ethnicity Categorical Race/ethnicity category
education Ordinal Education level
bmi Continuous Body mass index (kg/m2)
sbp Continuous Systolic blood pressure (mmHg)
dbp Continuous Diastolic blood pressure (mmHg)
hba1c Continuous Glycated haemoglobin (%)
totchol Continuous Total cholesterol (mg/dL)
hdl Continuous HDL cholesterol (mg/dL)
ldl Continuous LDL cholesterol (mg/dL)
triglycerides Continuous Triglycerides (mg/dL)
creatinine Continuous Serum creatinine (mg/dL)
egfr Continuous Estimated GFR (mL/min/1.73m2)
albumin Continuous Serum albumin (g/dL)
diabetes Binary Self-reported diabetes
hypertension Binary Hypertension (SBP ≥ 140 or DBP ≥ 90 or on meds)
smoking Categorical Never / Former / Current
physical_activity Continuous Minutes of moderate-vigorous activity per week

Used in: Chapters 2, 10, 15, 16, 19 (Capstone 2)

A.5 Simulated Meningitis Data

File: data/meningitis_sim.csv Source: Simulated dataset inspired by the case study in Lopez-Ayala et al. (BMJ 2025). Variable distributions are based on published summary statistics from the Duke University Medical Center meningitis cohort. Description: Simulated data on acute meningitis patients for demonstrating spline modelling of non-linear associations.

Variable Type Description
id Integer Patient ID
age Continuous Age (years)
sex Binary 0 = Female, 1 = Male
csf_glucose Continuous CSF glucose (mg/dL)
csf_leuk Continuous CSF leucocyte count (cells/mm3)
csf_protein Continuous CSF protein (mg/dL)
blood_glucose Continuous Blood glucose (mg/dL)
bacterial Binary 1 = Acute bacterial meningitis, 0 = Viral

Observations: 501 Used in: Chapter 4 (primary), Chapter 3

A.6 Downloading the data

All datasets are included in the course repository. If you have cloned the repo, they are in the data/ directory.

To load them:

Code 
library(readr)
framingham <- read_csv("data/framingham.csv")
wdbc <- read_csv("data/wdbc.csv")
pbc <- read_csv("data/pbc.csv")
nhanes <- read_csv("data/nhanes_subset.csv")
meningitis <- read_csv("data/meningitis_sim.csv")
Code 
import pandas as pd
framingham = pd.read_csv("data/framingham.csv")
wdbc = pd.read_csv("data/wdbc.csv")
pbc = pd.read_csv("data/pbc.csv")
nhanes = pd.read_csv("data/nhanes_subset.csv")
meningitis = pd.read_csv("data/meningitis_sim.csv")