List Size–
Mean Age–
Deaths/Yr0
Cumul. Deaths0
Mean QRISK–
Month–
🦠 COVID WAVE
Ethnicity Mix  |  Top Conditions
–
ℹ️ About This Model — methodology, assumptions & caveats (click to expand)

What this is

A synthetic GP population health simulator designed for educational and research use. It models a 10,000-patient list with demographic characteristics representative of a Harrow, NW London practice — one of the most ethnically diverse boroughs in England.


The simulator allows exploration of how population-level interventions affect mortality and disease prevalence over time, using a monthly tick engine with probabilistic incident and death events.

Clinical foundations

  • Mortality rates: ONS England & Wales life tables, with comorbidity multipliers from published literature (UKPDS, DAPA-HF, EMPEROR, EMPA-REG)
  • Condition prevalence: Calibrated against QOF 2022/23 national averages, adjusted for Harrow ethnicity profile
  • NICE guidelines: NG136 (HTN), NG28 (T2DM), NG196 (AF), NG106 (HF), NG115 (COPD)
  • QRISK: Simplified scoring based on QRISK3 variables (age, sex, ethnicity, BMI, SBP, cholesterol, comorbidities, smoking)
  • CHA₂DS₂-VASc: Standard ESC scoring for AF anticoagulation decisions
  • Ethnicity risk: South Asian CVD multipliers from Bhopal et al.; Pakistani T2DM 3× risk from Bradford/Leicester cohort data; Black HTN elevation from UK Biobank

Intervention evidence base

  • Polypill: TIPS/PolyIran trials — ~25% MI reduction, ~20% stroke reduction for statin + antihypertensive combinations
  • Intensive DM: DAPA-HF (dapagliflozin 26% HF reduction), EMPEROR-Reduced (empagliflozin 25% HF reduction), UKPDS (intensive glucose control 24% MI reduction)
  • Smoking cessation: ASH/NICE: 3% monthly quit rate with programme support; Doll/Hill cohort — ex-smoker mortality trends
  • COVID wave: ONS COVID-19 mortality by age — age 70+ hazard ratio ~3× baseline; Long COVID prevalence ~10% exposed (REACT-2)

Limitations

  • Synthetic patients — not real individuals
  • Simplified QRISK (not validated algorithm)
  • No socioeconomic gradient within ethnicities
  • Drug side-effects and discontinuation not modelled
  • Multi-morbidity interaction effects simplified
⚠️ Educational & research use only. This simulator uses synthetic data and approximated clinical parameters. It is not validated for clinical decision-making, service planning, or commissioning. Results should not be used to guide treatment of individual patients or to make resource allocation decisions without reference to primary evidence and local epidemiological data.
🏥 Developed at a Harrow, NW London GP practice using Claude AI (Anthropic). Built to support population health education, public health training, and scenario planning discussions among clinical colleagues. The model is open-source and freely shareable — if you adapt or extend it, please retain this methodology note. Feedback and suggestions welcome.
Speed: 4× Space1mo  Y1yr  Aauto  Escmodal
Preparing simulations…
Shaded band on mortality chart shows 10th–90th percentile range across 20 random seeds with current interventions active.
🧑‍🤝‍🧑 Harrow Demographics & Ethnicity Profile — click to adjust
40%
30%
20%
5%
35%

⚠ Changes take effect on Reset or Rebuild Population

3
13%
+0

Deprivation and lifestyle factors affect baseline comorbidity prevalence and incident rates.

🔬 Interventions — any combination active simultaneously
📊 Scenario Comparison
Saved curves overlay on the mortality chart →
Population Pyramid
Cumulative Deaths (scenario comparison)
Condition Prevalence (%)
⚖️ Health Inequality Gap Cumulative deaths per 1,000 patients — most deprived quintile (score 4–5) vs least deprived (score 1–2)

Each patient has an individual deprivation score distributed around the practice mean. Adjust interventions and uptake sliders to watch the gap narrow — or widen. The shaded red area is the inequality gap. Source: Marmot Review 2010; PHE Health Equity reports.

Patient List (10,000)
▶ Simulation running — pause to browse patients
Age: –
# Name Age Sex Ethnicity Conditions QRISK Meds
Harrow GP Population Health Simulator · Open source · Built with Claude AI · GitHub · 📝 Share feedback / how you used this tool
⚠️ Educational and research use only. Not validated for clinical decision-making or commissioning.