스킬 파일

Epidemiological Model Parametrization Guide

Name: Epidemiological Model Parametrization Guide
Author: guillaumechabotcouture

Guide for researching epidemiological parameters, identifying calibration targets, and determining model structure for spatial disease transmission models. Use when the user needs to find R0, infectious/latent periods, birth/death rates, vaccination coverage, observed data for calibration, or guidance on model component selection (compartment type, seasonal forcing, spatial coupling, vaccination strategy). Trigger phrases include "research parameters", "what parameters do I need", "find R0 for", "epidemiological data sources", "disease parameters", "parametrize model", "calibration targets", "model structure for", "parameter space", "what data do I need".

guillaumechabotcouture0 스타2026. 2. 20.

직업
카테고리: 머신러닝

스킬 내용

Three Outputs

This skill produces three structured outputs for downstream use:

Parameter Space → calabaria ParameterSpace + ConfigurationSpace
- Uncertain parameters with ranges (for calibration)
- Known parameters with point estimates (fixed)
Calibration Targets → observed data + loss function design
- What observable data exists for this disease/region?
- How to structure it as a calabaria loss function
- Alignment strategy (temporal, spatial, age-stratified)
Model Structure Guidance → LASER component selection
- Which compartment model? (SEIR, SIR, SEIRS, etc.)
- Which seasonal forcing? (school-term, monsoon, none)
- Which spatial coupling? (gravity, radiation, none)
- Which vaccination? (routine, campaign, both, age bands)
- Which importation? (endemic seeding, stochastic, none)

Output 1: Parameter Space

Epidemiological Model Parametrization Guide

guillaumechabotcouture0 스타2026. 2. 20.

직업
카테고리: 머신러닝

스킬 내용

Three Outputs

This skill produces three structured outputs for downstream use:

Parameter Space → calabaria ParameterSpace + ConfigurationSpace
- Uncertain parameters with ranges (for calibration)
- Known parameters with point estimates (fixed)
Calibration Targets → observed data + loss function design
- What observable data exists for this disease/region?
- How to structure it as a calabaria loss function
- Alignment strategy (temporal, spatial, age-stratified)
Model Structure Guidance → LASER component selection
- Which compartment model? (SEIR, SIR, SEIRS, etc.)
- Which seasonal forcing? (school-term, monsoon, none)
- Which spatial coupling? (gravity, radiation, none)
- Which vaccination? (routine, campaign, both, age bands)
- Which importation? (endemic seeding, stochastic, none)

Output 1: Parameter Space

관련 스킬

Parameter	Typical Range	Red Flags
R0	1.5–18 depending on disease	<1 means no sustained transmission
Latent period	2–21 days	>30 days unusual for acute infections
Infectious period	3–14 days	>30 days → chronic, different model needed
CBR	8–50 per 1000/yr	<1 or >60 likely unit error
CDR	3–20 per 1000/yr	Same unit check as CBR
Vaccination coverage	0.0–1.0	>1.0 is a unit error (not percentage)
Generation interval	5–25 days	Should ≈ latent + infectious/2 for SEIR
Infection-to-case ratio	1:1–1:2000	Disease-specific; >1:100 common for enteric pathogens

from calabaria.parameters import ParameterSpace, ParameterSpec
from calabaria.parameters import ConfigurationSpace, ConfigSpec

# Calibration parameters (uncertain → ranges for Optuna)
PARAMS = ParameterSpace([
    ParameterSpec("beta", lower=<lower>, upper=<upper>, kind="float",
                  doc="<source: Author et al. YYYY, range justification>"),
    ParameterSpec("gravity_k", lower=<lower>, upper=<upper>, kind="float",
                  doc="<source>"),
    # ... additional uncertain parameters
])

# Fixed parameters (well-known → point estimates)
CONFIG = ConfigurationSpace([
    ConfigSpec("latent_period_mean", default=<value>,
               doc="<source: WHO/PubMed systematic review>"),
    ConfigSpec("infectious_period_mean", default=<value>,
               doc="<source>"),
    ConfigSpec("cbr", default=<value>,
               doc="<source: UN World Population Prospects YYYY>"),
    # ... additional fixed parameters
])

Data Type	Example	Loss Function	Alignment
Case incidence time series	Weekly reported cases by district	MSE on log-transformed counts	Temporal (week) × spatial (district)
Annual/total case counts	Total cases per province per year	Poisson or negative binomial likelihood	Spatial × annual
Seroprevalence	Fraction immune by age group	Beta-binomial likelihood	Age × spatial
Extinction/persistence	Fraction of weeks with zero cases	Logistic curve fit (CCS similarity)	Spatial (by population size)
Spatial spread patterns	Phase lags in wavelet analysis	Phase difference similarity	Spatial (distance from epicenter)
Vaccination impact	Pre/post-campaign case reduction	Relative reduction ratio	Temporal (before/after)

Source	What It Provides	Format	URL/Access
WHO GHO	National incidence, mortality by year	CSV/API	gho.who.int
GPEI / POLIS	Subnational polio case counts, AFP surveillance	CSV	polioeradication.org
DHS	Seroprevalence, vaccination coverage surveys	Stata/CSV	dhsprogram.com
UN WPP	Population, CBR, CDR, age pyramids by country	CSV/Excel	population.un.org
GBD / IHME	Disease burden estimates by country/year	CSV	ghdx.healthdata.org
Ministry of Health reports	Subnational case data, outbreak reports	PDF/tables	Country-specific
Published studies	Age-stratified seroprevalence, outbreak curves	Extract from papers	PubMed

import polars as pl

# Observed data as polars DataFrame
# Schema should match what your @model_output methods produce
observed = pl.DataFrame({
    "year": [2015, 2015, 2016, 2016, ...],
    "patch": [0, 1, 0, 1, ...],
    "cases": [45, 12, 38, 8, ...],
})

# Loss function for calibration loop
def compute_loss(model_output_df: pl.DataFrame, observed_df: pl.DataFrame) -> float:
    """Compare model weekly_incidence output to observed data.

    Returns single float loss for TrialResult.
    """
    joined = model_output_df.join(observed_df, on=["year", "patch"], suffix="_obs")
    log_model = (joined["cases"] + 1).log()
    log_obs = (joined["cases_obs"] + 1).log()
    return ((log_model - log_obs) ** 2).mean()

Does infection confer lasting immunity?
├── Yes
│   ├── Is there a latent (non-infectious) period?
│   │   ├── Yes → SEIR
│   │   └── No  → SIR
│   └── Does immunity wane on simulation timescale?
│       ├── Yes → SEIRS or SIRS
│       └── No  → SEIR or SIR
└── No → SIS

Driver	Profile Shape	Examples
School terms	Biweekly step function (Bjornstad)	Measles, influenza (temperate)
Climate/monsoon	Cosine peaking in wet/warm season	Cholera, dengue, malaria
Behavioral	Holiday/pilgrimage calendar	Meningitis belt (dry season gatherings)
None apparent	Flat (`ValuesMap.from_scalar(1.0, ...)`)	Some chronic infections
Unknown	Include `seasonal_amplitude` in PARAMS	Let calibration decide

Model	When to Use	LASER Function
Gravity	Default for human diseases; well-studied	`gravity()`
Radiation	When intervening populations create barriers	`radiation()`
Competing destinations	When destinations compete for travelers	`competing_destinations()`
None	Single-patch or well-mixed	Set `gravity_k = 0`

Strategy	Component	Key Parameters
Routine only	`RoutineImmunizationEx`	Age at vaccination, coverage
Campaign only	`VaccinationCampaign`	Period, coverage, age band
Both	Both components	Consider correlated missedness
None	Omit vaccination components	Pre-vaccine era modeling

Pattern	When to Use	Configuration
Endemic corridor	Continuous cross-border transmission	`end_tick = nticks`, moderate `count`
Stochastic reintroduction	Sustain sub-CCS patches	Moderate `period`, low `count`
Initial seeding only	No ongoing importation	Seed I > 0 in scenario GeoDataFrame only
None	Fully closed system	Omit importation component

Parameter Type	Primary Sources	Search Terms
R0, generation interval	PubMed systematic reviews	"{disease} R0 systematic review"
Latent/infectious period	WHO disease factsheets, PubMed	"{disease} incubation period", "{disease} serial interval"
Demographics (CBR, CDR)	UN World Population Prospects	"{country} crude birth rate"
Vaccination coverage	DHS, MICS, WHO/UNICEF estimates	"{country} {disease} vaccine coverage subnational"
Population by patch	National census, WorldPop	"{country} district population"
Seasonal forcing	Published modeling studies	"{disease} seasonal transmission {region}"

Parameter	Value/Range	Unit	Source	Uncertainty
R0	12-18	dimensionless	Systematic review (Author YYYY)	Calibrate
Latent period	8-13 days	days	WHO factsheet	Fixed at 10
CBR	29.2	per 1000/yr	UN WPP 2024	Fixed
...	...	...	...	...

Disease	Typical Detection Ratio	Source
Measles	1:3–1:10 (varies by surveillance quality)	WHO
Polio (WPV1)	1:200 (paralysis:infection)	CDC
Cholera	1:4–1:25	WHO
Influenza	1:10–1:100	CDC

Epidemiological Model Parametrization Guide

Three Outputs

Output 1: Parameter Space

Epidemiological Model Parametrization Guide

Three Outputs

Output 1: Parameter Space

Parameter Checklist by Model Type

SEIR (Susceptible-Exposed-Infectious-Recovered)

SIR (no latent period)

SIRS / SEIRS (waning immunity)

SIS (no lasting immunity)

Parameter Validation Ranges

Output Format

Output 2: Calibration Targets

What to Calibrate Against

Target Types by Data Availability

Data Sources for Calibration Targets

Structuring for calabaria

Accounting for Under-Ascertainment

Output 3: Model Structure Guidance

Decision Tree for Component Selection

1. Compartment Model

2. Seasonal Forcing

3. Spatial Coupling

4. Vaccination Strategy

5. Importation / Seeding

Research Workflow

Step 1: Scope the Problem

Step 2: Research Parameters (Output 1)

Step 3: Identify Calibration Targets (Output 2)

Step 4: Determine Model Structure (Output 3)

Step 5: Validate and Document

Disease-Specific Quick References

Measles

Polio (WPV1)

Cholera

Influenza (Seasonal)

Handoff to Downstream Skills

Continuous Learning V2

Continuous Learning V2

Continuous Learning V2

Continuous Learning

Continuous Learning

Pytorch Patterns