Data specification — Demo dataset

This repository contains a small demo dataset (CSV) and a specification document describing the fields and provenance. The purpose of this demo is to illustrate how a data package for diabetes-related primary care metrics might be structured for sharing and documentation.

Files included

• data/sample_patients.csv — one row per patient with demographic fields.
• data/measurements.csv — repeated measures per patient with clinical measurements.
• data/derived_patient_status.csv — one row per patient with computed BMI, weight category, blood pressure category, and assumed heights (derived output).

Both CSV files are intentionally synthetic and contain dummy values for demonstration only.

Derived outputs

• data/derived_patient_status.csv — patient-level summary generated by scripts/derive_patient_status.py. Contains one row per patient with computed BMI, weight category, blood pressure category, and assumed heights. See the Scripts section below for details.

• Processing workflow — describes the steps and script used to produce the derived dataset.

See also (downloadable):

• scripts/derive_patient_status.py
• scripts/requirements.txt

How the data were obtained

This is simulated data created for demonstration. No real patient data were used. Values were chosen to mimic plausible distributions and column names commonly used in primary care diabetes datasets.

Data dictionary

• patient_id: Unique patient identifier (synthetic).
• age: Integer age in years.
• sex: ‘M’ or ‘F’.
• ethnicity: Broad ethnicity group.
• imd_quintile: Index of Multiple Deprivation quintile (1 = most deprived, 5 = least deprived).
• date: ISO date of measurement (YYYY-MM-DD).
• hba1c_mmol_per_mol: HbA1c value in mmol/mol.
• bp_systolic: Systolic blood pressure in mmHg.
• bp_diastolic: Diastolic blood pressure in mmHg.
• weight_kg: Weight in kilograms (one decimal allowed).

Data dictionary: data/derived_patient_status.csv

The derived file contains the following columns (generated by the script):

• patient_id: Unique patient identifier (matches input patients).
• sex: ‘M’ or ‘F’ (from input patients).
• ethnicity: Broad ethnicity group (from input patients).
• measurement_date: ISO date of the latest measurement for this patient (YYYY-MM-DD).
• bp_systolic: Systolic blood pressure from the latest measurement (mmHg).
• bp_diastolic: Diastolic blood pressure from the latest measurement (mmHg).
• bp_category: Derived blood pressure classification: ‘high’ (systolic ≥ 140 or diastolic ≥ 90), ‘low’ (systolic < 90 or diastolic < 60), or ‘normal’.
• weight_kg: Weight from the latest measurement (kg).
• height_cm: Assumed average height in centimetres based on sex and ethnicity (demo-only assumptions).
• height_source: String indicating whether the height was assumed using ‘assumed_by_sex_ethnicity’ or ‘assumed_by_sex’ (fallback).
• bmi: Computed Body Mass Index (weight_kg / height_m²), rounded to 1 decimal place.
• weight_category: Derived weight classification from BMI: ‘underweight’ (BMI < 18.5), ‘normal’ (18.5 ≤ BMI < 25), ‘overweight’ (25 ≤ BMI < 30), or ‘obese’ (BMI ≥ 30).

Provenance and processing notes

• Synthetic generation: values were hand-crafted to be plausible and do not correspond to any real individual.
• No identifying information is present.

How to access the data from the published site

When this repository’s data specification is published to GitHub Pages (via the manual workflow Publish data specification), a static site is created containing this document and the data/ folder. Links on the published page point to the CSV files here:

• sample_patients.csv
• measurements.csv
• derived_patient_status.csv

You can click these links on the published site to download or view the raw CSV files.

License and reuse

This demo dataset is provided for educational/demonstration purposes. You may reuse the data and documentation freely.

Contact

Repository owner: Ricky Olivier (ricky-lv426).