Modelling Components Introduction

This page provides an overview of the modelling components in the NCD module.

Overall Structure

The general structure is as follows:

• All models have a demographic component
• There are four disease models built around this demographic component
• Disease models have clinical interventions
• Disease models may have one or many risk factors
• Cardiovascular disease also has a "CVD Prevention" component

How Disease Models Work

Disease models are systems dynamic models that simulate population health over time. They track people moving between health states according to transition probabilities that can be modified by interventions.

Key Concepts

• States: The model organizes people into health states (e.g., healthy, diseased, deceased)
• Transitions: People move between states based on rates (incidence, remission, mortality)
• Population Structure: Each state holds arrays by age and sex (Males aged 0-100, Females aged 0-100 = 202 arrays)
• Weights: Transition probabilities that can be age and sex specific (e.g., 2% chance of moving from Healthy to Asthma state)
• Dynamic Changes: Weights can be altered over time (e.g., if tobacco prevalence increases, transition rates may increase)
• Interventions: Clinical treatments that modify transition rates through coverage and impact parameters

The models are built using the Botech Protocol. Further documentation can be found here (opens in a new tab)

Health Metrics Calculated

• Prevalence: People with the disease in a given year
• Incidence: New cases during the year
• Mortality: Disease-specific and background deaths
• YLD (Years Lived with Disability): Prevalence × disability weight
• YLL (Years of Life Lost): Deaths × remaining life expectancy
• DALYs (Disability-Adjusted Life Years): YLD + YLL

Common Order of Operations

All disease models (asthma, COPD, CVD, and diabetes) follow a similar order of operations, which is simplified below.

Steps 1-3: Initialization

1. Generate population using UNDP World Population Prospects data
2. Load constants for calculation components e.g. the impact for of a clinical intervention on the disability weight
3. Calculate current coverage change as Current Coverage - Starting Coverage

Steps 4-7: Treatment Effect Calculations (Disease-Specific)

4. Calculate treatment effects e.g. PIN × Impact × Coverage
5. Aggregate effects e.g. summing the total effects on incidence
6. Calculate disease-specific disability weights using a mix of GBD disability weights for healthy populations, and constants
7. Initialise incidence, mortality and other rates

Steps 8-11: Main Population Movements

8. Record birth calculations for fertile women
9. Execute state transitions between disease states
10. Record disability by multiplying populations × disability weights
11. Apply migration using UNDP population projections

Steps 12-17: Year-End Processing

12. Age populations by 1 year (remove age 100+)
13. Apply mortality using background and disease-specific rates
14. Women give birth from previously calculated values
15. Clear temporary values to prevent carryover
16. Increment time step to next year
17. End simulation at year 2119

Disease Model Variations

Each disease model page documents only the specific differences from this common process, particularly:

• Treatment-specific calculations
• Unique state transitions
• Model-specific mortality/disability rates

This approach eliminates redundancy while preserving the technical detail needed for each disease model.