MODEL_DESCRIPTION.md

AMR Simulation — Model Description

Contents

1. Overview
2. Population and Demographics
3. Infection Acquisition
4. Clinical Progression
5. Diagnostic Testing
6. Antibiotic Treatment
7. Resistance Dynamics
8. Microbiome and Carriage
9. Horizontal Gene Transfer
10. Mortality
11. Policy Evaluation

• Appendix A — Bacteria, Drugs, Mechanisms and Enums
• Appendix B — Parameter Reference
• Appendix C — Output Specification

---

1. Overview

This model simulates the emergence, transmission, and dynamics of antimicrobial resistance (AMR) across a synthetic human population from 1930 to 2035. It is an individual-based (agent-based) model in which each person can acquire bacterial infections, receive antibiotic treatment, develop resistance through de novo mutation or horizontal gene transfer, and carry resistant organisms in their microbiome.

The model tracks 42 bacterial species, 58 antibiotics (grouped into 18 drug classes), and 35 resistance mechanisms. The population is distributed across 6 world regions (North America, Europe, Asia, Oceania, South America, Africa), each with distinct epidemiological and healthcare profiles.

Time advances in discrete daily steps. On each day, every individual in the population is processed through a sequence of 21 rules covering ageing, infection acquisition, clinical progression, treatment, resistance dynamics, and death.

Scope

The model is designed for:

• Estimating the global burden of AMR over time
• Evaluating the potential impact of antibiotic stewardship policies
• Exploring "what if" counterfactual scenarios (e.g., a world without resistance)
• Understanding how resistance mechanisms spread across bacterial species and regions

Limitations

• Drug levels are modelled as abstract potency units rather than pharmacokinetic concentrations
• Bacteria-bacteria competition within the microbiome is represented implicitly through resistance promotion and decay rather than explicit strain dynamics
• The model does not capture within-host spatial heterogeneity (e.g., biofilm vs planktonic)
• Vaccine effects are not explicitly modelled (though their population-level impact is partially captured through incidence parameters)
• Region definitions are broad continental groupings

---

2. Population and Demographics

2.1 Initialisation

The population is created at day 0 (calendar year 1930). Each individual is assigned:

• Age: drawn from a demographic distribution that encodes both living individuals and future births (negative age values represent individuals not yet born)
• Sex: male or female with equal probability
• Region: sampled from demographic weights reflecting the global population distribution

Age bands are specified in 4,000-day (~11-year) intervals. The demographic distribution is the product of per-region weights:

Variable name	Description
demo_{region}_age_{start}_{end}	Probability weight for an individual in a given region and age band

The six regions and their approximate population shares:

Region	Share
Asia	55%
Europe	15%
Africa	12%
North America	9%
South America	6%
Oceania	3%

2.2 Ageing

Each day, every individual's age increments by one day. Age categories are recalculated:

Age category	Age range	Variable suffix
Infant	0–1 year	infant
Preschool	1–5 years	preschool
School age	5–18 years	school
Young adult	18–50 years	young_adult
Middle age	50–70 years	middle_age
Elderly	70+ years	elderly

An additional classification is used for sepsis and mortality:

Category	Age range	Variable suffix
Neonatal	0–28 days	neonatal
Pediatric	28 days–18 years	pediatric
Young adult	18–50 years	young_adult
Elderly	50+ years	elderly

2.3 Immunodeficiency

Individuals can be immunosuppressed, increasing their risk of infection acquisition, sepsis, and death. Two types exist:

Type	Onset rate	Recovery rate
Temporary	temporary_immunosuppression_onset_rate_per_day = 0.00005	temporary_immunosuppression_recovery_rate_per_day = 0.01
Chronic	chronic_immunosuppression_onset_rate_per_day = 0.00006	chronic_immunosuppression_recovery_rate_per_day = 0.0012

Chronic immunodeficiency probability at birth varies by age category:

Variable	Value
chronic_immunodeficiency_probability_age_0_1	0.3
chronic_immunodeficiency_probability_age_1_18	0.2
chronic_immunodeficiency_probability_age_18_65	0.4
chronic_immunodeficiency_probability_age_65_plus	0.6

2.4 Hospitalisation

Hospital admission is modelled as a logistic function of clinical state:

$$P(\text{admission}) = \frac{1}{1 + e^{-\text{log_odds}}}$$

where:

$$\text{log_odds} = \text{base} + \text{age_effect} + \text{sepsis_effect} + \text{infection_effect} + \text{region_effect}$$

Variable	Default	Description
hospitalization_base_log_odds	−10.4	Baseline (≈0.003%)
hospitalization_log_odds_per_age_year	0.02	~2% increase per year of age
hospitalization_log_odds_sepsis	4.4	ln(80), strong driver
hospitalization_log_odds_symptomatic_infection	2.5	~12× for symptomatic cases
hospitalization_symptomatic_infection_level_threshold	3.0	Minimum infection level
hospitalization_recovery_rate_per_day	0.28	~3.6-day average stay
hospitalization_max_days	30.0	Maximum stay
hospitalization_prevent_discharge_with_sepsis	1.0	Block discharge during sepsis

Regional hospital admission log-odds:

Region	Variable	Value
North America	north_america_hospitalization_log_odds	0.5
Europe	europe_hospitalization_log_odds	0.6
Oceania	oceania_hospitalization_log_odds	0.4
Asia	asia_hospitalization_log_odds	0.0
South America	south_america_hospitalization_log_odds	−0.2
Africa	africa_hospitalization_log_odds	−0.5

2.5 Travel

Individuals may travel between regions, acquiring different drug availability and resistance profiles:

Variable	Default	Description
travel_probability_per_day	0.00005	Base daily travel probability
north_america_travel_multiplier	3.0	Relative travel frequency
europe_travel_multiplier	3.5
oceania_travel_multiplier	2.5
asia_travel_multiplier	1.5
south_america_travel_multiplier	0.8
africa_travel_multiplier	0.3

---

3. Infection Acquisition

3.1 Community Acquisition

Each day, each non-infected individual has a probability of acquiring each of the 42 bacterial species. The probability is computed via a logistic model combining:

• Base acquisition rate for the bacteria
• Regional modifier
• Age-dependent risk (via templates and overrides)
• Immunodeficiency modifier
• Seasonal variation (sinusoidal for respiratory pathogens)
• Calendar-era effects (temporal multiplier)
• Population-level resistance prevalence (majority_r) reducing the probability that the bacteria the individual acquires is resistant

Variable pattern	Description
bacteria_{name}_acquisition_log_odds	Baseline acquisition log-odds for each species
{region}_bacteria_{name}_acquisition_log_odds	Regional override
bacteria_{name}_log_odds_{age_category}	Age-specific override
{bacteria}_{region}_log_odds_{age_category}	Bacteria × region × age interaction

Age risk templates

Each bacteria is assigned a risk template that defines its age distribution:

Template	Multipliers [0–1y, 1–5y, 5–18y, 18–50y, 50–70y, 70+y]
respiratory	[3.0, 1.8, 0.8, 1.0, 1.3, 2.5]
gastrointestinal	[2.5, 2.0, 1.2, 1.0, 1.1, 1.8]
urogenital	[1.2, 0.8, 0.9, 1.0, 1.4, 2.2]
skin_soft_tissue	[1.5, 1.3, 1.1, 1.0, 1.2, 1.8]
bloodstream	[4.0, 2.0, 0.7, 1.0, 1.5, 3.0]
sexually_transmitted	[0.1, 0.2, 0.8, 1.0, 0.8, 0.3]
flat	[1.0, 1.0, 1.0, 1.0, 1.0, 1.0]

Template assignments:

Variable	Default	Selected overrides
{bacteria}_age_risk_template	"respiratory"
salm_typhi_age_risk_template	"gastrointestinal"
esch_coli_age_risk_template	"urogenital"
pseud_aerug_age_risk_template	"bloodstream"
staph_aureus_age_risk_template	"skin_soft_tissue"
n_gonorrhoeae_age_risk_template	"sexually_transmitted"

Age-specific overrides (selected examples)

STIs — C. trachomatis, N. gonorrhoeae, T. pallidum: peak in young adults (log-odds 0.4–2.0), very low in children (−2.5 to −4.3).

Respiratory — S. pneumoniae, H. influenzae, M. catarrhalis: U-shaped, high in infants (1.7–2.5) and elderly (1.0–1.2), low in young adults (−0.4 to −0.8).

Enteric — S. Typhi, Shigella, C. jejuni: highest in preschool/infants (1.2–2.0). H. pylori accumulates with age (elderly 1.8).

Healthcare-associated — A. baumannii peaks in elderly (1.5), C. difficile extreme elderly peak (2.0).

Invasive — N. meningitidis bimodal: infant (1.8) + young adult (1.3). L. monocytogenes: neonatal (1.8) + elderly (1.5).

3.2 Hospital Acquisition

Hospitalised individuals have elevated acquisition rates controlled by:

Variable pattern	Description
bacteria_{name}_hospital_acquisition_log_odds	Hospital-specific baseline

3.3 Carrier-Derived Infection

Individuals carrying bacteria in their microbiome can develop active infection from their carriage flora. This pathway is probability-gated:

Variable	Default	Description
carrier_resistance_inheritance_probability	0.50	Probability the new infection inherits resistance from microbiome
infection_from_microbiome_dampening	0.70	Dampening factor applied to carrier-to-infection conversion

3.4 Resistance at Acquisition

When a new infection is acquired from the community, its initial resistance profile is derived from:

1. Population-level prevalence (majority_r): the rolling average of resistance observed across all individuals for that bacteria–drug combination
2. Community resistance dilution: controlled by community_resistance_dilution_factor (default 0.50), which scales down the community resistance signal to avoid over-estimating resistance in newly acquired infections
3. Mechanism profile sampling: resistance mechanisms are not assigned independently — instead, a correlated mechanism profile is sampled from a reservoir of up to 200 profiles observed in currently infected individuals (MechanismProfileCache), preserving realistic co-resistance patterns

---

4. Clinical Progression

4.1 Syndrome Assignment

Each new infection is assigned a clinical syndrome (site of infection), which affects drug penetration, treatment selection, and progression:

Syndrome	Index	Examples
UTI	1	Urinary tract infection
Skin/soft tissue	2	Cellulitis, wound infection
Respiratory	3	Pneumonia, bronchitis
Bloodstream	4	Bacteraemia
Intra-abdominal	5	Peritonitis, abscess
CNS	6	Meningitis
Gastrointestinal	7	Gastroenteritis
Genital/pelvic	8	PID, STI
Bone/joint	9	Osteomyelitis
Other	10	Device-related, other

Syndrome probabilities are bacteria-specific. For example, E. coli infections are predominantly UTIs and bloodstream, while S. pneumoniae infections are predominantly respiratory.

Syndrome initiation multipliers

Variable	Syndrome	Value	Effect
syndrome_3_initiation_multiplier	Respiratory	2.5	Higher treatment-seeking
syndrome_7_initiation_multiplier	GI	0.5	Lower treatment-seeking
syndrome_8_initiation_multiplier	Genital	0.25	Lower treatment-seeking
Others	—	1.0	Default

Syndrome bacteria growth multipliers

Variable	Syndrome	Value
syndrome_1_bacteria_growth_multiplier	UTI	0.7
syndrome_2_bacteria_growth_multiplier	Skin	0.8
syndrome_3_bacteria_growth_multiplier	Respiratory	1.0
syndrome_4_bacteria_growth_multiplier	Bloodstream	1.3
syndrome_5_bacteria_growth_multiplier	Intra-abdominal	1.1
syndrome_6_bacteria_growth_multiplier	CNS	1.2
syndrome_7_bacteria_growth_multiplier	GI	0.9
syndrome_8_bacteria_growth_multiplier	Genital	0.6
syndrome_9_bacteria_growth_multiplier	Bone/joint	0.8
syndrome_10_bacteria_growth_multiplier	Other	1.0

4.2 Symptom Onset

Infection progression from asymptomatic to symptomatic is driven by bacterial load exceeding a threshold, modified by:

• Bacteria-specific symptom onset rates
• Immune status
• Duration of infection

4.3 Sepsis

Sepsis onset is a logistic function of multiple risk factors:

$$\text{log_odds}_{\text{sepsis}} = \text{baseline} + \text{infection_level} + \text{duration} + \text{age} + \text{region} + \text{syndrome} + \text{immune} + \text{hospital} + \text{care_status}$$

Variable	Default	Description
sepsis_baseline_log_odds	−14.0	Very low baseline
log_odds_sepsis_infection_level	0.8	Per unit of bacterial load
log_odds_sepsis_infection_duration	0.005	Per day of infection
log_odds_sepsis_onset_immunosuppressed	0.7	~2× for immunosuppressed
log_odds_sepsis_onset_hospitalized	0.5	~1.6× for hospitalised
log_odds_sepsis_onset_not_under_care	1.0	~2.7× if not under care

Bacteria-specific sepsis baseline log-odds

Bacteria	Value	Comment
E. coli	−21.0	Very common, rarely septic
N. gonorrhoeae	−21.0	Extremely rare sepsis
C. trachomatis	−19.0
C. jejuni	−20.0
Shigella spp.	−12.0
C. difficile	−11.0
M. catarrhalis	−11.0
T. pallidum	−12.0
B. pertussis	−11.0
S. pneumoniae	−9.0
H. influenzae	−9.0
V. cholerae	−9.0
N. meningitidis	−9.0
H. pylori	−250.0	Effectively never septic
MDR M. tuberculosis	−38.0	Very rare sepsis
Most other bacteria	−8.0	Default

Age-dependent sepsis log-odds

Variable	Value	Effect
sepsis_age_log_odds_neonatal	1.10	~3× (ln 3.0)
sepsis_age_log_odds_pediatric	0.18	~1.2× (ln 1.2)
sepsis_age_log_odds_young_adult	0.0	Reference
sepsis_age_log_odds_elderly	0.69	~2× (ln 2.0)

Per-bacteria-age sepsis interactions

Neonatal: S. agalactiae +0.981, E. coli +0.511, L. monocytogenes +0.693, S. aureus +0.288.

Pediatric: S. pneumoniae +0.916, H. influenzae +0.734, N. meningitidis +0.511, S. aureus +0.511.

Elderly: S. pneumoniae +0.693, E. coli +0.223, K. pneumoniae +0.405, P. aeruginosa +0.560, A. baumannii +0.405, E. faecium +0.336, S. aureus +0.470.

Young adult: N. meningitidis +0.336, S. aureus +0.588.

Regional sepsis onset log-odds

Region	Variable	Value
North America	north_america_sepsis_onset_log_odds	−0.5
Europe	europe_sepsis_onset_log_odds	−0.6
Oceania	oceania_sepsis_onset_log_odds	−0.5
Asia	asia_sepsis_onset_log_odds	−0.1
South America	south_america_sepsis_onset_log_odds	0.0
Africa	africa_sepsis_onset_log_odds	0.1

Syndrome-specific sepsis log-odds

Syndrome	Value	Comment
UTI (1)	−2.0	Low sepsis risk
Skin (2)	−1.0	Low
Respiratory (3)	0.0	Moderate
Bloodstream (4)	1.5	High
Intra-abdominal (5)	0.8	Moderate-high
CNS (6)	1.2	High
GI (7)	−0.5	Low-moderate
Genital (8)	−1.5	Low
Bone/joint (9)	0.5	Moderate

4.4 Natural Clearance

Infections can resolve naturally through immune-mediated clearance, subject to:

• Bacteria-specific clearance probabilities
• Antibiotic activity (enhances clearance)
• Duration of infection (longer infections are harder to clear)
• Immune status

Variable	Default	Description
default_microbiome_clearance_probability_per_day	0.01	Base clearance rate
microbiome_clearance_probability_on_drug_treatment	0.8	Enhanced clearance on antibiotics
antibiotic_clearance_log_odds_per_unit_activity	0.5	Clearance boost per unit of active drug
carriage_duration_log_odds_coefficient	−0.01	Longer carriage → harder to clear
carriage_duration_max_log_odds_effect	−2.0	Maximum reduction (~7.4× harder)

---

5. Diagnostic Testing

5.1 Testing Availability

Bacterial identification and resistance testing become available at historical time points:

Variable	Default	Calendar year
bacterial_testing_available_from_day	5,478	~1945
resistance_testing_available_from_day	9,131	~1955

5.2 Legacy Testing Parameters

Variable	Default	Description
test_delay_days	3.0	Days between sample and result
test_rate_per_day	0.2	Base probability of ordering a test
prob_test_r_done	0.95	Probability AST is performed given culture
test_r_error_probability	0.02	False result rate

5.3 Modern Testing Model

Variable	Default	Description
bacterial_testing_base_rate_per_day	0.15	Base culture rate
resistance_testing_base_rate_per_day	0.95	Conditional on positive culture
bacterial_testing_hospital_multiplier	8.0	Hospital increases testing
resistance_testing_hospital_multiplier	5.0
testing_immunosuppressed_multiplier	2.5
testing_sepsis_multiplier	4.0

Regional testing multipliers

Region	Value
North America	1.1
Europe	1.2
Asia	0.7
South America	0.6
Oceania	0.8
Africa	0.3

Temporal adoption (S-curve)

Variable	Default
bacterial_testing_initial_adoption_rate	0.1
bacterial_testing_max_temporal_multiplier	1.0
resistance_testing_initial_adoption_rate	0.05
resistance_testing_max_temporal_multiplier	1.0

---

6. Antibiotic Treatment

6.1 Treatment Initiation

The decision to start antibiotics is modelled as a logistic function:

$$P(\text{initiation}) = \frac{1}{1 + e^{-\text{log_odds}}}$$

Variable	Default	Description
antibiotic_initiation_base_log_odds	−6.5	Baseline (~0.1%)
antibiotic_initiation_log_odds_symptomatic_infection	6.5	Lifts to ~26%
antibiotic_initiation_log_odds_sepsis	6.0	Very strong driver
antibiotic_initiation_log_odds_test_identified	0.92	ln(2.5)
antibiotic_initiation_log_odds_already_on_drug	0.18	ln(1.2)
antibiotic_initiation_log_odds_immunodeficiency	2.08	ln(8.0)
antibiotic_initiation_log_odds_no_indication	−1.05	ln(0.35), dampens inappropriate use

Regional antibiotic initiation log-odds

Region	Value	Approximate effect
North America	0.0	Reference
Europe	0.0	Same
Oceania	0.0	Same
Asia	−0.5	~38% reduction
South America	−0.8	~55% reduction
Africa	−1.4	~75% reduction

6.2 Drug Selection

Drug selection follows a two-stage scoring system:

1. Empiric therapy (no test result available): drugs are scored based on a syndrome-specific template
2. Targeted therapy (test result available): drugs are scored based on known susceptibility

Empiric scoring parameters

Variable	Default	Description
empiric_therapy_broad_spectrum_bonus	0.85	Bonus for broader coverage
empiric_therapy_ineffective_drug_penalty	0.001	Penalty for likely-ineffective drugs

Targeted scoring parameters

Variable	Default	Description
targeted_therapy_narrow_spectrum_bonus	5.0	Strong preference for narrow-spectrum
targeted_therapy_broad_spectrum_penalty	0.1	Penalty for broad-spectrum when narrow available
targeted_therapy_ineffective_drug_penalty	0.001	Penalty for ineffective drugs

Regional resistance surveillance penalties

Drugs with high population-level resistance in a region receive selection penalties:

Variable	Default	Description
regional_resistance_penalty_very_high	0.3	Score multiplier when resistance >60%
regional_resistance_penalty_high	0.5	When resistance >45%
regional_resistance_penalty_moderate	0.8	When resistance >10%
regional_resistance_threshold_very_high	0.6
regional_resistance_threshold_high	0.45
regional_resistance_threshold_moderate	0.1

Syndrome-specific empiric scoring templates

Each syndrome has a pre-defined scoring table ranking drugs by appropriateness. A higher score means that drug is more likely to be selected empirically for that syndrome.

Syndrome 1 — UTI

Drug	Score
nitrofurantoin	15.0
trim_sulf	14.0
ciprofloxacin	12.0
levofloxacin	11.0
cephalexin	10.0
amoxicillin_clavulanate	9.0
cefuroxime	8.0
ampicillin	6.0
amoxicillin	6.0
ceftriaxone	5.0
gentamicin	4.0
meropenem	3.0

Syndrome 2 — Skin/Soft Tissue

Drug	Score
penicillin_g	13.0
amoxicillin_clavulanate	12.0
cephalexin	12.0
clindamycin	11.0
doxycycline	10.0
trim_sulf	9.0
cefazolin	8.0
vancomycin	7.0
linezolid	6.0
ciprofloxacin	5.0
azithromycin	5.0
metronidazole	4.0

Syndrome 3 — Respiratory

Drug	Score
amoxicillin_clavulanate	20.0
amoxicillin	12.0
azithromycin	12.0
doxycycline	11.0
levofloxacin	10.0
ceftriaxone	9.0
cefuroxime	8.0
trim_sulf	7.0
moxifloxacin	6.0
penicillin_g	6.0
clarithromycin	5.0
ampicillin	4.0
ciprofloxacin	3.0
erythromycin	3.0

Syndrome 4 — Bloodstream

Drug	Score
piperacillin_tazobactam	18.0
meropenem	14.0
vancomycin	13.0
ceftriaxone	12.0
cefepime	11.0
ampicillin_sulbactam	10.0
ciprofloxacin	9.0
levofloxacin	9.0
amoxicillin_clavulanate	8.0
linezolid	7.0
metronidazole	5.0
trim_sulf	5.0
azithromycin	4.0
gentamicin	1.0
tobramycin	1.0
amikacin	1.0

Syndrome 5 — Intra-abdominal

Drug	Score
piperacillin_tazobactam	14.0
meropenem	13.0
ampicillin_sulbactam	11.0
metronidazole	11.0
ceftriaxone	10.0
ciprofloxacin	9.0
levofloxacin	9.0
cefepime	8.0
amoxicillin_clavulanate	7.0
clindamycin	6.0
vancomycin	4.0
azithromycin	3.0

Syndrome 6 — CNS

Drug	Score
ceftriaxone	15.0
vancomycin	14.0
ampicillin	12.0
meropenem	11.0
penicillin_g	10.0
linezolid	8.0
chloramphenicol	7.0
rifampicin	6.0
ciprofloxacin	5.0
trim_sulf	5.0
metronidazole	4.0
doxycycline	3.0

Syndrome 7 — Gastrointestinal

Drug	Score
ciprofloxacin	12.0
furazolidone	11.0
azithromycin	11.0
metronidazole	10.0
doxycycline	9.0
trim_sulf	8.0
ceftriaxone	7.0
amoxicillin_clavulanate	7.0

Syndrome 8 — Genital/Pelvic

Drug	Score
azithromycin	13.0
ceftriaxone	13.0
doxycycline	12.0
penicillin_g	12.0
amoxicillin_clavulanate	9.5
amoxicillin	9.0
cefuroxime	9.0
clindamycin	9.0
ciprofloxacin	7.0
levofloxacin	6.5
trim_sulf	5.0
rifampicin	4.0
metronidazole	2.5

Syndrome 9 — Bone/Joint

Drug	Score
penicillin_g	14.0
cefazolin	13.0
ampicillin	12.0
vancomycin	12.0
linezolid	11.0
cephalexin	11.0
ceftriaxone	11.0
tedizolid	10.0
dalbavancin	10.0
clindamycin	10.0
ciprofloxacin	9.0
levofloxacin	9.0
rifampicin	9.0
trim_sulf	8.0
meropenem	7.0
piperacillin_tazobactam	6.5

Syndrome 10 — Other/Device-Related

Drug	Score
piperacillin_tazobactam	8.0
cefepime	8.0
ceftriaxone	8.0
meropenem	8.0
imipenem_c	8.0
vancomycin	8.0
linezolid	7.0
ciprofloxacin	7.0
azithromycin	6.0

6.3 Drug Pharmacokinetics

Each drug has a decay half-life determining how quickly its level falls after administration:

Variable pattern	Default	Description
drug_{name}_half_life_days	Drug-specific	PK half-life in days
drug_{name}_initial_level	10.0	Level at administration
drug_{name}_double_dose_multiplier	2.0	Level multiplier for double dose
drug_{name}_spectrum_breadth	3.0	Microbiome disruption potential (higher = broader)

Drug half-lives (selected)

Drug	Half-life (days)
sulfanilamide	0.29
penicillin_g	0.042
ampicillin	0.063
amoxicillin	0.063
ceftriaxone	0.33
azithromycin	2.92
doxycycline	0.75
vancomycin	0.25
meropenem	0.042
ciprofloxacin	0.17
linezolid	0.21
colistin	0.21
dalbavancin	14.0
cefiderocol	0.10

Spectrum breadth overrides

Drug	Breadth	Classification
penicillin_g	2.0	Narrow
vancomycin	2.5	Narrow-medium
linezolid	2.0	Narrow
trim_sulf	3.5	Medium-broad
azithromycin	4.0	Broad
colistin	4.0	Broad
ciprofloxacin	4.5	Very broad
ceftriaxone	4.0	Broad
cefepime	4.0	Broad
meropenem	5.0	Very broad

6.4 Drug Penetration by Syndrome

Drug efficacy at different infection sites varies by drug class. Penetration values range from 0.0 (no effective concentration) to 1.0 (full systemic levels):

Syndrome	Best penetration	Worst penetration
UTI (1)	FQ, TMP-SMX, nitrofurantoin, fosfomycin (1.0)	Macrolides (0.4), clindamycin (0.3), daptomycin (0.1)
Skin (2)	Daptomycin (0.95), FQ (0.9), oxazolidinones (0.9)	Nitrofurantoin (0.2)
Respiratory (3)	Macrolides (0.95), FQ (0.95), oxazolidinones (0.9)	Daptomycin (0.0), AG (0.4)
Bloodstream (4)	All 1.0 (reference compartment)	—
Intra-abdominal (5)	Metronidazole (0.9), FQ (0.75), carbapenems (0.75)	AG (0.3)
CNS (6)	Metronidazole (0.80), oxazolidinones (0.70), chloramphenicol (0.70)	AG (0.05), colistin (0.05), daptomycin (0.05)
GI (7)	Fidaxomicin (1.0), metronidazole (0.95), oral vancomycin (0.90)	Glycopeptides IV (0.35)
Genital (8)	FQ (0.9), metronidazole (0.8), TMP-SMX (0.8)	AG (0.35)
Bone/joint (9)	Rifampicin (0.80), oxazolidinones (0.75), FQ (0.70)	AG (0.25), colistin (0.2)

6.5 Drug Potency Matrix

Intrinsic drug activity against each bacterium (when no resistance is present) is defined in a 42×52 potency matrix. Values range from 0.0 (no activity) to 1.0 (maximum activity).

The potency matrix is embedded as POTENCY_EMBEDDED_DATA in config.rs and generates parameters with the key pattern:

drug_{drug}_for_bacteria_{bacteria}_potency_when_no_r

For example, meropenem has potency 0.95 against E. coli but 0.0 against MRSA (when mecA is present). Penicillin G has potency 0.90 against S. pneumoniae but 0.0 against P. aeruginosa (intrinsically resistant).

6.6 Drug Availability by Region and Era

Drug availability varies by region and becomes available at historical introduction dates:

Variable pattern	Range	Description
{region}_drug_{drug}_availability	0.0–1.0	Regional availability (1.0 = fully available)

Regional availability patterns:

Region	Pattern
North America	All drugs 1.0
Europe	All drugs 1.0
Asia	Most 1.0; tedizolid/ceftaroline 0.3, teicoplanin 0.7
Oceania	Most 1.0; tedizolid/ceftaroline 0.5
South America	Limited newer drugs (tedizolid 0.1, linezolid 0.5, carbapenems 0.6–0.7)
Africa	Basic antibiotics 0.8–1.0; ceftriaxone 0.6; vancomycin 0.3; carbapenems 0.1–0.2; most newer drugs 0.0–0.1

Drug introduction dates

Each antibiotic becomes available at a specific time step:

Drug	Time step	~Year
sulfanilamide	2,555	1937
penicillin_g	3,555	1942
chloramphenicol	6,935	1949
tetracycline	6,575	1948
colistin	8,020	1952
erythromycin	8,025	1952
nitrofurantoin	8,395	1953
furazolidone	9,125	1955
vancomycin	10,215	1958
fosfomycin	10,590	1959
metronidazole	10,965	1960
ampicillin	11,315	1961
fusidic_a	11,680	1962
gentamicin	12,045	1963
rifampicin	13,140	1966
doxycycline	13,505	1967
clindamycin	13,870	1968
trim_sulf	13,870	1968
amoxicillin	13,780	1972
cephalexin	14,605	1970
minocycline	14,965	1971
cefazolin	15,700	1973
tobramycin	16,325	1975
amikacin	16,690	1976
ticarcillin	14,600	1977
cefuroxime	17,525	1978
piperacillin	16,065	1981
ceftriaxone	19,715	1984
piperacillin_tazobactam	19,715	1984
ceftazidime	20,080	1985
imipenem_c	20,080	1985
amoxicillin_clavulanate	16,425	1985
aztreonam	20,445	1986
ciprofloxacin	20,805	1987
teicoplanin	21,170	1988
ampicillin_sulbactam	18,250	1990
ticarcillin_clavulanate	18,250	1990
clarithromycin	21,895	1990
ofloxacin	21,895	1990
azithromycin	22,260	1991
cefepime	24,195	1996
meropenem	24,195	1996
levofloxacin	24,195	1996
moxifloxacin	25,290	1999
quinu_dalfo	25,290	1999
linezolid	25,550	2000
ertapenem	25,920	2001
daptomycin	27,375	2005
ceftazidime_avibactam	27,740	2006
tigecycline	28,040	2007
retapamulin	28,405	2007
ceftaroline	29,305	2010
fidaxomicin	29,565	2011
tedizolid	30,660	2014
dalbavancin	30,660	2014
ceftolozane_tazobactam	30,295	2014
meropenem_vaborbactam	32,045	2018
cefiderocol	33,510	2019

Special case — Colistin: Withdrawn from routine use between ~1970 and ~1995 (availability drops to 5% during that window), reflecting historical concerns about nephrotoxicity before its re-adoption for MDR Gram-negative infections.

6.7 Drug Toxicity

Drugs accumulate a toxicity reservoir, which can trigger sub-lethal discontinuation or lethal toxicity death:

Variable	Default	Description
default_drug_toxicity_death_hazard_per_unit_level	0.0	Per-drug hazard contribution (most drugs = 0; high for colistin, aminoglycosides)
default_toxicity_reservoir_half_life_days	1.5	Decay rate of accumulated toxicity

Toxicity discontinuation (sub-lethal)

When accumulated toxicity exceeds a threshold, the treating clinician may discontinue the drug:

Variable	Default	Description
toxicity_discontinuation_base_log_odds	−3.0	Baseline discontinuation probability
toxicity_discontinuation_log_odds_per_reservoir_unit	1.5	Per unit of toxicity reservoir
toxicity_discontinuation_log_odds_sepsis	−1.5	Clinicians tolerate more toxicity during sepsis
toxicity_avoidance_penalty_multiplier	0.05	Recently-stopped drugs are avoided in reselection
toxicity_avoidance_window_days	14.0	Duration of avoidance penalty

Lethal toxicity death

Variable	Default	Description
toxicity_death_base_log_odds	−8.0	Very low baseline
toxicity_death_log_odds_per_reservoir_unit	2.0	Per unit of toxicity
toxicity_death_log_odds_age_infant	0.6
toxicity_death_log_odds_age_child	0.2
toxicity_death_log_odds_age_adult	0.0	Reference
toxicity_death_log_odds_age_elderly	0.8
toxicity_death_log_odds_immunosuppressed	0.9
toxicity_death_log_odds_hospitalized	0.25	Slight increase (monitoring enables detection but may indicate frailty)

6.8 Antibiotic infection prevention

Variable	Default
antibiotic_infection_prevention_efficacy	0.7

---

7. Resistance Dynamics

7.1 Resistance Mechanisms

The model tracks 35 distinct resistance mechanisms, each representing a biological pathway for antibiotic resistance:

Mechanism	Variable name	Description
ESBL CTX-M	esbl_ctx_m	Extended-spectrum β-lactamase (most common ESBL globally)
ESBL TEM	esbl_tem	Extended-spectrum β-lactamase (TEM-type)
ESBL SHV	esbl_shv	Extended-spectrum β-lactamase (SHV-type)
AmpC CMY	ampc_cmy	Plasmid-mediated AmpC β-lactamase
AmpC DHA	ampc_dha	Plasmid-mediated AmpC β-lactamase
KPC	kpc	Klebsiella pneumoniae carbapenemase
NDM/VIM	ndm_vim	Metallo-β-lactamases
OXA-48	oxa_48	Oxacillinase-type carbapenemase
PBP2a/MecA	pbp2a_meca	Penicillin-binding protein alteration (MRSA)
VanA	vana	High-level vancomycin resistance
VanB	vanb	Variable-level vancomycin resistance
GyrA (primary)	gyra_primary	DNA gyrase mutation (fluoroquinolone resistance, step 1)
GyrA + ParC	gyra_parc	Additional topoisomerase mutation (high-level FQ resistance)
ErmB	ermb	Erythromycin ribosome methylase (macrolide resistance)
Cfr	cfr	23S rRNA methyltransferase (multi-class resistance)
16S rRMT	16s_rrmt	16S rRNA methyltransferase (aminoglycoside resistance)
CAT	cat	Chloramphenicol acetyltransferase
Qnr	qnr	Quinolone resistance protein
MCR-1	mcr_1	Mobilised colistin resistance
AcrAB-TolC	acrab_tolc	Gram-negative efflux pump
MexXY-OprM	mexxy_oprm	Pseudomonas-specific efflux pump
OmpK35/36	ompk35_36	Outer membrane porin loss (Klebsiella)
OprD	oprd	Outer membrane porin loss (Pseudomonas)
Global efflux	global_efflux	Non-specific efflux upregulation
Global porin loss	global_porin_loss	Non-specific porin downregulation
Folate pathway	folate_pathway	Altered dihydrofolate reductase (trimethoprim resistance)
Nitroreductase	nitroreductase	Nitroreductase loss (nitrofurantoin resistance)
FosA	fosa	Fosfomycin-modifying enzyme
MprF	mprf	Membrane charge modification (daptomycin resistance)
RpoB	rpob	RNA polymerase mutation (rifampicin resistance)
FusB	fusb	Fusidic acid resistance determinant
As-yet-unknown 1–4	as_yet_unknown_{1..4}	Placeholder for future/novel mechanisms

7.2 Mechanism–Drug-Class Enhancement Multipliers

Each mechanism reduces the efficacy of specific drug classes. The enhancement multiplier (0.0–1.0) determines how much a mechanism reduces drug activity: 0.0 = no effect, 1.0 = complete resistance.

These are defined per mechanism × drug class (35 × 18 = 630 values), with variable pattern:

mech_{mechanism}_enhancement_{drug_class}

Global (legacy) enhancement multipliers (used when no per-class override exists):

Mechanism	Global multiplier
ESBL CTX-M	0.85
ESBL TEM	0.80
ESBL SHV	0.75
AmpC CMY/DHA	0.70
KPC	0.90
NDM/VIM	0.95
OXA-48	0.80
PBP2a/MecA	0.90
VanA	0.95
VanB	0.85
GyrA primary	0.70
GyrA + ParC	0.85
ErmB	0.80
Cfr	0.75
16S rRMT	0.85
CAT	0.70
Qnr	0.50
MCR-1	0.60
AcrAB-TolC	0.40
MexXY-OprM	0.45
OmpK35/36	0.50
OprD	0.55
Global efflux	0.35
Global porin loss	0.45
Folate pathway	0.70
Nitroreductase	0.60
FosA	0.65
MprF	0.55
RpoB	0.80
FusB	0.70
As-yet-unknown 1–4	0.50 each

7.3 Resistance Emergence

De novo resistance emergence occurs when an individual is under antibiotic pressure. The probability is a function of:

• Base emergence rate for the mechanism
• Bacteria-specific incidence band multiplier
• Whether the bacteria is biologically capable of acquiring that mechanism

Incidence band multipliers

Bacteria are classified into incidence bands reflecting their frequency in the population. More common bacteria have lower per-capita emergence rates to avoid unrealistically fast resistance evolution:

Band	Multiplier	Example bacteria
High incidence	×0.1	E. coli, Shigella, N. gonorrhoeae, C. jejuni, H. pylori, S. aureus, S. pneumoniae, M. pneumoniae, C. trachomatis
Moderate incidence	×1.0	K. pneumoniae, S. Typhi, S. pyogenes, B. pertussis, M. genitalium, T. pallidum
Low incidence	×3.0	Enterobacter, E. cloacae, Proteus, iNTS, P. aeruginosa, V. cholerae, M. catarrhalis, H. influenzae, S. epidermidis, S. agalactiae, E. faecalis, C. difficile, B. fragilis
Very low incidence	×10.0	Citrobacter, Morganella, Serratia, P. stuartii, A. baumannii, S. maltophilia, B. cepacia, N. meningitidis, L. pneumophila, E. faecium, L. monocytogenes, MDR TB, Y. enterocolitica

Base emergence rates are mechanism-specific, ranging from ~1×10⁻⁹ (rare chromosomal mutations) to ~1×10⁻⁶ (common plasmid-mediated acquisitions).

Mechanism assignment probability: mechanism_assignment_probability = 0.8 — when a resistance event occurs, this is the probability the specific mechanism responsible is tracked (versus being attributed to a generic mechanism).

7.4 Resistance Reversion

Resistance mechanisms that impose a fitness cost on the bacteria can be lost over time when antibiotic pressure is removed. Reversion is:

• Drug-gated: reversion only occurs when the individual is NOT currently exposed to any drug in the relevant drug class
• Mechanism-specific: each mechanism has its own reversion rate

Mechanism	Reversion rate (per day)
ESBL CTX-M, TEM, SHV	0.0005
AmpC CMY, DHA	0.0003
KPC, NDM/VIM	0.0002
OXA-48	0.0004
PBP2a/MecA	0.0003
VanA	0.002
VanB	0.001
GyrA primary, GyrA + ParC	0.0001
ErmB	0.0008
Cfr	0.001
16S rRMT	0.0005
CAT	0.001
Qnr	0.001
MCR-1	0.001
AcrAB-TolC, MexXY-OprM	0.0005
OmpK35/36, OprD	0.0003
Global efflux, Global porin loss	0.0005
Folate pathway	0.0008
Nitroreductase	0.001
FosA	0.001
MprF	0.0008
RpoB	0.002
FusB	0.001
As-yet-unknown 1–4	0.0005

7.5 Resistance Floors

For clinically important bacteria, the model enforces minimum resistance levels that ramp up after the introduction of each drug class. This prevents unrealistically low resistance for well-established resistance phenotypes:

Variable pattern	Default	Description
resistance_floor_feature_enabled	>0.5 to enable	Global toggle
bacteria_{name}_resistance_floor_enabled	>0.5 to enable	Per-bacteria toggle
bacteria_{name}_resistance_floor_ramp_years	10.0	Years from drug class introduction to full floor
bacteria_{name}_{drug_class}_resistance_floor	0.0	Target minimum resistance level

The floor ramps linearly from 0 to the target level over the ramp period, starting from the introduction date of the earliest drug in the relevant class.

7.6 Cross-Resistance Groups

For each bacteria, drugs that share resistance mechanisms are grouped together. Acquiring resistance to one drug in a cross-resistance group confers resistance to all drugs in that group. Examples:

Bacteria	Group	Drugs sharing resistance
E. coli	ESBL	Penicillins + C1-3G + BL/BLI
E. coli	FQ	Ciprofloxacin, levofloxacin, moxifloxacin, ofloxacin
E. coli	AG	Gentamicin, tobramycin, amikacin
S. aureus	β-lactamase	Penicillins
S. aureus	MRSA	All cephalosporins
S. aureus	MLS	Macrolides + clindamycin
A. baumannii	Carbapenemase	All carbapenems + BL/BLI
P. aeruginosa	β-lactamase	Piperacillin + cephalosporins + BL/BLI
S. pneumoniae	PBP	Penicillin + ampicillin + amoxicillin + BL/BLI
E. faecium	Glycopeptide	VanA/VanB (vancomycin + teicoplanin)

---

8. Microbiome and Carriage

8.1 Overview

Each individual maintains a per-bacteria microbiome resistance state. Bacteria can colonise the microbiome (carriage) without causing active infection. Carriage is the primary reservoir for resistance transmission.

8.2 Carriage Compartments

Each bacterial species is assigned to a carriage compartment reflecting its natural ecological niche:

Compartment	Example bacteria
Gut	E. coli, K. pneumoniae, Enterococcus spp., Shigella, Salmonella, C. difficile
Respiratory	S. pneumoniae, H. influenzae, P. aeruginosa, A. baumannii, M. catarrhalis, M. tuberculosis
Skin/Soft Tissue	S. aureus, S. epidermidis
Genitourinary	N. gonorrhoeae, C. trachomatis, M. genitalium, T. pallidum, S. agalactiae
Systemic	(Reserved; not currently assigned to any bacteria)

8.3 Resistance in the Microbiome

The microbiome resistance level (microbiome_r) is tracked per bacteria per individual as a value from 0.0 to 1.0.

• Minority resistance (<0.5): resistant strains are present but not dominant
• Majority resistance (≥0.5): resistant strains dominate the carriage flora

The transition from minority to majority is driven by antibiotic exposure (selective pressure promoting resistant strains):

Variable	Default	Description
microbiome_majority_promotion_rate_per_day	0.02	Daily probability of minority→majority promotion under drug pressure

Without antibiotic pressure, resistance in the microbiome decays. This represents fitness cost and competitive displacement by susceptible strains.

8.4 Microbiome Disruption

Antibiotic use disrupts the normal microbiome, increasing susceptibility to colonisation by resistant organisms. The model employs a Microbiome Disruption Reservoir (microbiome_disruption_level) on each individual that accumulates positive disruption while actively taking antibiotics (scaled by the drug's spectrum_breadth). When antibiotics are discontinued, this ecological disruption acts as a trailing "hangover" penalty, decaying logarithmically over the specified half-life timeframe. This persistent disruption allows the model to accurately capture the prolonged colonization window for opportunistic pathogens (e.g. VRE, C. difficile) following rigorous, multi-drug therapies.

Variable	Default	Description
default_microbiome_disruption_log_odds	0.3	Base disruption accumulation rate per active unit
microbiome_resistance_multiplier_on_acquisition	0.50	Resistance level at acquisition
antibiotic_disruption_decay_half_life_days	30.0	Recovery time after stopping antibiotics

8.5 Majority Resistance Cache

Population-level resistance prevalence (majority_r) is computed from a rolling window:

Variable	Default	Description
majority_r_window_days	100	Window for computing rolling prevalence
majority_r_min_total_samples	10	Minimum samples before prevalence is computed
majority_r_freeze_at_last_positive	0.0	If >0.5, freeze prevalence when data is scarce

---

9. Horizontal Gene Transfer

9.1 Overview

Horizontal gene transfer (HGT) allows resistance mechanisms to spread between different bacterial species. The model implements mechanism-driven HGT: only mechanisms that are biologically capable of horizontal transfer (primarily plasmid-borne) can be transferred.

9.2 Transfer Probabilities

HGT probability depends on the taxonomic relationship between donor and recipient bacteria. Bacteria are assigned to plasmid pools:

Pool	Members
GramPositive	S. aureus, S. epidermidis, S. pneumoniae, S. pyogenes, S. agalactiae, Enterococcus spp., L. monocytogenes
EntericGramNegative	E. coli, Klebsiella, Enterobacter, Proteus, Salmonella, Shigella, Citrobacter, Serratia, Morganella, Y. enterocolitica
RespiratoryGramNegative	P. aeruginosa, A. baumannii, S. maltophilia, B. cepacia, M. catarrhalis, H. influenzae, B. pertussis, Legionella
Anaerobe	C. difficile, B. fragilis
None	T. pallidum, H. pylori, C. jejuni, M. tuberculosis, N. gonorrhoeae, N. meningitidis, Chlamydia, Mycoplasma (no HGT)

Transfer type	Probability
Same group	1×10⁻¹⁰
Cross-group (same Gram stain)	1×10⁻¹¹
Cross-Gram	0.0 (blocked)
Excluded organisms (None pool)	0.0

9.3 HGT Modifiers

Variable	Default	Description
hgt_base_probability	1×10⁻⁵	Global multiplier
hgt_co_infection_multiplier	10.0	Increased when both species are actively infecting
hgt_hospital_multiplier	5.0	Hospital environment premium
hgt_microbiome_multiplier	2.0	Carriage-to-carriage transfer
hgt_gut_compartment_multiplier	2.0	Additional multiplier for gut-compartment bacteria (dense microbial environment)

9.4 Mechanism Transferability

Only certain mechanisms are eligible for HGT. Transferable mechanisms include plasmid-borne determinants (ESBLs, carbapenemases, MCR-1, etc.). Chromosomal mutations (GyrA, RpoB, porin loss) are not transferable.

---

10. Mortality

10.1 Background Mortality

Non-infection-related death is modelled to maintain realistic population demographics:

$$\text{log_odds} = \text{base} + \text{age_linear} + \text{age_quadratic} + \text{era} + \text{region} + \text{sex} + \text{immune} + \text{hospital}$$

Variable	Default	Description
background_mortality_baseline_log_odds	−14.0	Very low daily baseline
log_odds_mortality_per_year_of_age	0.04	Linear age effect
log_odds_mortality_per_year_of_age_squared	0.05	Quadratic age effect (accelerating)
mortality_baseline_1930_multiplier	3.0	Higher mortality in 1930
mortality_baseline_2035_multiplier	1.0	Current baseline
mortality_improvement_half_life_years	35.0	Rate of improvement
log_odds_mortality_immunosuppressed	0.916	ln(2.5), ~2.5×
log_odds_mortality_hospitalized	0.262	ln(1.3), ~1.3×

Regional mortality log-odds

Region	Value
North America	0.0
Europe	−0.105
Oceania	0.0
Asia	0.18
South America	0.26
Africa	0.69

Sex effect

Sex	Value
Male	0.095
Female	−0.105

10.2 Sepsis Mortality

Sepsis death follows a logistic model evaluated daily for individuals in sepsis:

Variable	Default	Description
sepsis_death_base_log_odds	−5.0	Daily baseline during sepsis
sepsis_death_log_odds_age_infant	1.1
sepsis_death_log_odds_age_child	−0.7	Children more resilient
sepsis_death_log_odds_age_adult	0.0	Reference
sepsis_death_log_odds_age_elderly	0.9
sepsis_death_log_odds_immunosuppressed	1.5	~4.5×
sepsis_death_log_odds_bacteria_level	0.35	Per unit of bacterial load
sepsis_death_log_odds_duration	0.04	Per day of sepsis
sepsis_death_log_odds_early_phase	0.8	Higher in first 3 days
sepsis_death_early_phase_days	3.0
sepsis_death_log_odds_not_under_care	1.4	~4× if not under care

Regional sepsis mortality multipliers

Region	Value
North America	0.5
Europe	0.4
Oceania	0.5
Asia	0.9
South America	1.1
Africa	1.5

10.3 Sepsis Recovery

Variable	Default	Description
sepsis_base_log_odds_of_recovery_per_day	0.0	Base daily recovery probability
sepsis_log_odds_bacteria_level	−0.3	Higher load → harder to recover
sepsis_log_odds_in_hospital	0.8	Hospital care assists recovery
sepsis_log_odds_age_infant	−0.5
sepsis_log_odds_age_child	0.4	Children recover faster
sepsis_log_odds_age_adult	0.0	Reference
sepsis_log_odds_age_elderly	−0.7
sepsis_log_odds_immunosuppressed	−1.0
sepsis_minimum_duration_days	1.0

Regional sepsis recovery log-odds

Region	Value
North America	0.4
Europe	0.5
Oceania	0.3
Asia	0.0
South America	−0.3
Africa	−0.7

---

11. Policy Evaluation

11.1 Branching

At a configurable branch year (default 2027), the simulation saves the full population state and runs three independent scenarios forward to the end of the simulation:

Branch	Description
Baseline	No policy changes — business as usual
Stewardship	Antibiotic stewardship adjustments (e.g., narrower prescribing, enhanced testing)
Counterfactual	Hypothetical scenario (e.g., a world with no resistance)

11.2 Policy Parameters

Each branch can modify:

Parameter	Baseline	Stewardship	Counterfactual
drug_selection_temperature	—	×0.65 (more deterministic)	—
minimal_potency_threshold_for_drug_selection	—	—	—
bacterial_testing_rate_multiplier	—	×1.5	—
resistance_testing_rate_multiplier	—	×1.5	—
counterfactual_resistance_multiplier	—	—	0.0
clear_all_resistance_on_branch_start	false	false	true
reserve_drug_penalty_multiplier	—	×2.0	—
drug_initiation_rate_multiplier	—	×0.85	—
drug_cessation_rate_multiplier	—	×1.2	—

11.3 Key Constants

Constant	Value	Description
SIMULATION_START_YEAR	1930.0	Calendar year at day 0
POLICY_BRANCH_YEAR	2027.0	Year when policies diverge
INFECTION_EPS	0.001	Minimum meaningful infection level
MICROBIOME_MAJORITY_THRESHOLD	0.5	Threshold for majority resistance
MAX_MECHANISM_PROFILES	200	Reservoir sample size per bacteria for mechanism profile cache

---

Appendix A — Bacteria, Drugs, Mechanisms and Enums

A.1 Bacteria (42 species)

Index	Species	Group	Carriage compartment
0	Acinetobacter baumannii	NonFermenter	Respiratory
1	Citrobacter spp.	Enterobacterales	Gut
2	Enterobacter spp.	Enterobacterales	Gut
3	Enterococcus faecalis	GramPositive	Gut
4	Enterococcus faecium	GramPositive	Gut
5	Escherichia coli	Enterobacterales	Gut
6	Klebsiella pneumoniae	Enterobacterales	Gut
7	Morganella spp.	Enterobacterales	Gut
8	Proteus spp.	Enterobacterales	Gut
9	Serratia spp.	Enterobacterales	Gut
10	Providencia stuartii	Enterobacterales	Genitourinary
11	Pseudomonas aeruginosa	NonFermenter	Respiratory
12	Stenotrophomonas maltophilia	NonFermenter	Respiratory
13	Staphylococcus aureus	GramPositive	Skin/Soft Tissue
14	Staphylococcus epidermidis	GramPositive	Skin/Soft Tissue
15	Streptococcus pneumoniae	GramPositive	Respiratory
16	Salmonella enterica serovar Typhi	Enterobacterales	Gut
17	Salmonella enterica serovar Paratyphi A	Enterobacterales	Gut
18	Invasive non-typhoidal Salmonella spp.	Enterobacterales	Gut
19	Shigella spp.	Enterobacterales	Gut
20	Neisseria gonorrhoeae	Fastidious	Genitourinary
21	Streptococcus pyogenes	GramPositive	Respiratory
22	Streptococcus agalactiae	GramPositive	Genitourinary
23	Haemophilus influenzae	Fastidious	Respiratory
24	Chlamydia trachomatis	Fastidious	Genitourinary
25	Mycoplasma genitalium	Fastidious	Genitourinary
26	Vibrio cholerae	EntericPathogen	Gut
27	Neisseria meningitidis	Fastidious	Respiratory
28	Listeria monocytogenes	GramPositive	Gut
29	Clostridioides difficile	Anaerobe	Gut
30	Bacteroides fragilis	Anaerobe	Gut
31	Campylobacter jejuni	Helicobacter	Gut
32	Enterobacter cloacae	Enterobacterales	Gut
33	Yersinia enterocolitica	Enterobacterales	Gut
34	Moraxella catarrhalis	Fastidious	Respiratory
35	Treponema pallidum	Spirochete	Genitourinary
36	Bordetella pertussis	Fastidious	Respiratory
37	Helicobacter pylori	Helicobacter	Gut
38	MDR Mycobacterium tuberculosis	Mycobacteria	Respiratory
39	Mycoplasma pneumoniae	Fastidious	Respiratory
40	Legionella pneumophila	Fastidious	Respiratory
41	Burkholderia cepacia complex	NonFermenter	Respiratory

A.2 Antibiotics (58 drugs in 18 classes)

Drug	Class
sulfanilamide	Sulfonamides
penicillin_g	Penicillins
ampicillin	Penicillins
amoxicillin	Penicillins
piperacillin	Penicillins
ticarcillin	Penicillins
cephalexin	Cephalosporins 1–2G
cefazolin	Cephalosporins 1–2G
cefuroxime	Cephalosporins 1–2G
ceftriaxone	Cephalosporins 3G
ceftazidime	Cephalosporins 3G
cefepime	Cephalosporins 4–5G
ceftaroline	Cephalosporins 4–5G
ceftolozane_tazobactam	Cephalosporins 3G
cefiderocol	Cephalosporins 4–5G
meropenem	Carbapenems
imipenem_c	Carbapenems
ertapenem	Carbapenems
aztreonam	Monobactams
erythromycin	Macrolides
azithromycin	Macrolides
clarithromycin	Macrolides
clindamycin	Macrolides
gentamicin	Aminoglycosides
tobramycin	Aminoglycosides
amikacin	Aminoglycosides
ciprofloxacin	Fluoroquinolones
levofloxacin	Fluoroquinolones
moxifloxacin	Fluoroquinolones
ofloxacin	Fluoroquinolones
tetracycline	Tetracyclines
doxycycline	Tetracyclines
minocycline	Tetracyclines
tigecycline	Tetracyclines
vancomycin	Glycopeptides
teicoplanin	Glycopeptides
dalbavancin	Glycopeptides
linezolid	Oxazolidinones
tedizolid	Oxazolidinones
daptomycin	Other
quinu_dalfo	Other
trim_sulf	Sulfonamides
chloramphenicol	Chloramphenicol
nitrofurantoin	Other
fosfomycin	Other
retapamulin	Other
fusidic_a	Other
metronidazole	Other
fidaxomicin	Other
furazolidone	Other
rifampicin	Other
amoxicillin_clavulanate	BLI Combinations
piperacillin_tazobactam	BLI Combinations
ampicillin_sulbactam	BLI Combinations
ticarcillin_clavulanate	BLI Combinations
ceftazidime_avibactam	Novel BL/BLI
meropenem_vaborbactam	Novel BL/BLI
colistin	Polymyxins

A.3 Drug Classes (18)

Code	Full name	Drugs
pen	Penicillins	penicillin G, ampicillin, amoxicillin, piperacillin, ticarcillin
bli	BLI Combinations	amoxicillin-clavulanate, piperacillin-tazobactam, ampicillin-sulbactam, ticarcillin-clavulanate
c1_2g	Cephalosporins 1–2G	cephalexin, cefazolin, cefuroxime
c3g	Cephalosporins 3G	ceftriaxone, ceftazidime, ceftolozane-tazobactam
c4_5g	Cephalosporins 4–5G	cefepime, ceftaroline, cefiderocol
bl_ni	Novel BL/BLI	ceftazidime-avibactam, meropenem-vaborbactam
carb	Carbapenems	meropenem, imipenem, ertapenem
mono	Monobactams	aztreonam
fq	Fluoroquinolones	ciprofloxacin, levofloxacin, moxifloxacin, ofloxacin
ag	Aminoglycosides	gentamicin, tobramycin, amikacin
mls	Macrolides/Lincosamides	erythromycin, azithromycin, clarithromycin, clindamycin
glyc	Glycopeptides	vancomycin, teicoplanin, dalbavancin
tet	Tetracyclines	tetracycline, doxycycline, minocycline, tigecycline
poly	Polymyxins	colistin
oxa	Oxazolidinones	linezolid, tedizolid
chl	Chloramphenicol	chloramphenicol
sulf	Sulfonamides	sulfanilamide, trimethoprim-sulfamethoxazole
other	Other	daptomycin, quinupristin-dalfopristin, nitrofurantoin, fosfomycin, retapamulin, fusidic acid, metronidazole, fidaxomicin, furazolidone, rifampicin

A.4 Resistance Mechanisms (35)

See Section 7.1 for the full table.

A.5 Enumerations

BacteriaGroup (9 groups)

Group	Description
Enterobacterales	Gram-negative enteric rods
NonFermenter	Non-fermenting Gram-negatives
GramPositive	Gram-positive cocci and rods
Fastidious	Fastidious Gram-negatives and atypicals
EntericPathogen	Specific enteric pathogens (V. cholerae)
Anaerobe	Obligate anaerobes
Spirochete	Spirochetes (T. pallidum)
Helicobacter	Helicobacter/Campylobacter
Mycobacteria	Mycobacteria (M. tuberculosis)

CarriageCompartment (5)

Gut, Respiratory, SkinSoftTissue, Genitourinary, Systemic

ResistanceAcquisitionType (5)

Type	Description
AtInfection	Acquired at community infection
AtInfectionHosp	Acquired at hospital infection
AtInfectionTB	Acquired at carrier-to-infection (treated-by) conversion
DuringInfection	De novo emergence during treatment
HGT	Horizontal gene transfer

InfectionResolutionType (6)

DrugTreatment, NaturalClearance, Death, SepsisDeath, ToxicityDeath, BackgroundDeath

ImmunodeficiencyType (2)

Temporary, Chronic

AgeCategory (7)

Infant, Preschool, SchoolAge, YoungAdult, MiddleAge, Elderly, NotYetBorn

HospitalStatus (2)

Community, Hospital

Region (7)

NorthAmerica, Europe, Asia, Oceania, SouthAmerica, Africa, Home (fallback)

MicrobiomeResistanceLevel (4)

None, Low, Medium, High

---

Appendix B — Parameter Reference

This appendix lists all parameters defined in the model's configuration. Parameters are stored as key–value pairs in a global HashMap and accessed by string key at runtime.

B.1 Global Scalars

These are the ~120 top-level parameters stored in the GlobalScalars struct:

Infection acquisition

Variable	Default	Description
infection_growth_rate_per_day	0.1	Daily bacterial growth increment
infection_initial_level	1.0	Starting bacterial load
infection_clearance_threshold	0.5	Level below which infection resolves
infection_death_threshold	50.0	Level at which death may occur
symptom_onset_threshold	3.0	Level for symptom development
symptom_recheck_interval_days	7.0	Re-evaluation interval
symptom_onset_rate_per_day	0.1	Base symptom development rate
not_under_care_fraction	0.05	Fraction not seeking medical care

Antibiotic treatment initiation

Variable	Default
antibiotic_initiation_base_log_odds	−6.5
antibiotic_initiation_log_odds_symptomatic_infection	6.5
antibiotic_initiation_log_odds_sepsis	6.0
antibiotic_initiation_log_odds_test_identified	0.92
antibiotic_initiation_log_odds_already_on_drug	0.18
antibiotic_initiation_log_odds_immunodeficiency	2.08
antibiotic_initiation_log_odds_no_indication	−1.05

Antibiotic treatment cessation

Variable	Default
treatment_stop_improvement_threshold	2.0
treatment_stop_rate_per_day	0.03
treatment_duration_base_days	7.0

Drug efficacy

Variable	Default
drug_effect_on_bacteria_per_day	0.5
drug_minimum_effective_level	0.1

Drug selection

Variable	Default
empiric_therapy_broad_spectrum_bonus	0.85
empiric_therapy_ineffective_drug_penalty	0.001
targeted_therapy_narrow_spectrum_bonus	5.0
targeted_therapy_broad_spectrum_penalty	0.1
targeted_therapy_ineffective_drug_penalty	0.001
regional_resistance_penalty_very_high	0.3
regional_resistance_penalty_high	0.5
regional_resistance_penalty_moderate	0.8
regional_resistance_threshold_very_high	0.6
regional_resistance_threshold_high	0.45
regional_resistance_threshold_moderate	0.1

Testing

Variable	Default
bacterial_testing_available_from_day	5,478
resistance_testing_available_from_day	9,131
test_delay_days	3.0
test_rate_per_day	0.2
prob_test_r_done	0.95
test_r_error_probability	0.02
bacterial_testing_base_rate_per_day	0.15
resistance_testing_base_rate_per_day	0.95
bacterial_testing_hospital_multiplier	8.0
resistance_testing_hospital_multiplier	5.0
testing_immunosuppressed_multiplier	2.5
testing_sepsis_multiplier	4.0
bacterial_testing_initial_adoption_rate	0.1
bacterial_testing_max_temporal_multiplier	1.0
resistance_testing_initial_adoption_rate	0.05
resistance_testing_max_temporal_multiplier	1.0

Sepsis onset

Variable	Default
sepsis_baseline_log_odds	−14.0
log_odds_sepsis_infection_level	0.8
log_odds_sepsis_infection_duration	0.005
log_odds_sepsis_onset_immunosuppressed	0.7
log_odds_sepsis_onset_hospitalized	0.5
log_odds_sepsis_onset_not_under_care	1.0
sepsis_age_log_odds_neonatal	1.10
sepsis_age_log_odds_pediatric	0.18
sepsis_age_log_odds_young_adult	0.0
sepsis_age_log_odds_elderly	0.69

Sepsis death

Variable	Default
sepsis_death_base_log_odds	−5.0
sepsis_death_log_odds_age_infant	1.1
sepsis_death_log_odds_age_child	−0.7
sepsis_death_log_odds_age_adult	0.0
sepsis_death_log_odds_age_elderly	0.9
sepsis_death_log_odds_immunosuppressed	1.5
sepsis_death_log_odds_bacteria_level	0.35
sepsis_death_log_odds_duration	0.04
sepsis_death_log_odds_early_phase	0.8
sepsis_death_early_phase_days	3.0
sepsis_death_log_odds_not_under_care	1.4

Sepsis recovery

Variable	Default
sepsis_base_log_odds_of_recovery_per_day	0.0
sepsis_log_odds_bacteria_level	−0.3
sepsis_log_odds_in_hospital	0.8
sepsis_log_odds_age_infant	−0.5
sepsis_log_odds_age_child	0.4
sepsis_log_odds_age_adult	0.0
sepsis_log_odds_age_elderly	−0.7
sepsis_log_odds_immunosuppressed	−1.0
sepsis_minimum_duration_days	1.0

Background mortality

Variable	Default
background_mortality_baseline_log_odds	−14.0
log_odds_mortality_per_year_of_age	0.04
log_odds_mortality_per_year_of_age_squared	0.05
mortality_baseline_1930_multiplier	3.0
mortality_baseline_2035_multiplier	1.0
mortality_improvement_half_life_years	35.0
log_odds_mortality_immunosuppressed	0.916
log_odds_mortality_hospitalized	0.262

Hospitalisation

Variable	Default
hospitalization_base_log_odds	−10.4
hospitalization_log_odds_per_age_year	0.02
hospitalization_log_odds_sepsis	4.4
hospitalization_log_odds_symptomatic_infection	2.5
hospitalization_symptomatic_infection_level_threshold	3.0
hospitalization_recovery_rate_per_day	0.28
hospitalization_max_days	30.0
hospitalization_prevent_discharge_with_sepsis	1.0

Immunodeficiency

Variable	Default
temporary_immunosuppression_onset_rate_per_day	0.00005
temporary_immunosuppression_recovery_rate_per_day	0.01
chronic_immunosuppression_onset_rate_per_day	0.00006
chronic_immunosuppression_recovery_rate_per_day	0.0012
chronic_immunodeficiency_probability_age_0_1	0.3
chronic_immunodeficiency_probability_age_1_18	0.2
chronic_immunodeficiency_probability_age_18_65	0.4
chronic_immunodeficiency_probability_age_65_plus	0.6
antibiotic_infection_prevention_efficacy	0.7

Resistance

Variable	Default
mechanism_assignment_probability	0.8

Microbiome and carriage

Variable	Default
default_microbiome_clearance_probability_per_day	0.01
microbiome_clearance_probability_on_drug_treatment	0.8
default_microbiome_disruption_log_odds	0.3
microbiome_resistance_multiplier_on_acquisition	0.50
infection_from_microbiome_dampening	0.70
antibiotic_disruption_decay_half_life_days	30.0
carriage_duration_log_odds_coefficient	−0.01
carriage_duration_max_log_odds_effect	−2.0
antibiotic_clearance_log_odds_per_unit_activity	0.5
carrier_resistance_inheritance_probability	0.50
community_resistance_dilution_factor	0.50
majority_r_window_days	100
majority_r_min_total_samples	10
majority_r_freeze_at_last_positive	0.0
microbiome_majority_promotion_rate_per_day	0.02

Toxicity

Variable	Default
default_drug_toxicity_death_hazard_per_unit_level	0.0
default_toxicity_reservoir_half_life_days	1.5
toxicity_death_base_log_odds	−8.0
toxicity_death_log_odds_per_reservoir_unit	2.0
toxicity_death_log_odds_age_infant	0.6
toxicity_death_log_odds_age_child	0.2
toxicity_death_log_odds_age_adult	0.0
toxicity_death_log_odds_age_elderly	0.8
toxicity_death_log_odds_immunosuppressed	0.9
toxicity_death_log_odds_hospitalized	0.25
toxicity_discontinuation_base_log_odds	−3.0
toxicity_discontinuation_log_odds_per_reservoir_unit	1.5
toxicity_discontinuation_log_odds_sepsis	−1.5
toxicity_avoidance_penalty_multiplier	0.05
toxicity_avoidance_window_days	14.0

HGT

Variable	Default
hgt_base_probability	1×10⁻⁵
hgt_co_infection_multiplier	10.0
hgt_hospital_multiplier	5.0
hgt_microbiome_multiplier	2.0
hgt_gut_compartment_multiplier	2.0

Travel

Variable	Default
travel_probability_per_day	0.00005
north_america_travel_multiplier	3.0
europe_travel_multiplier	3.5
oceania_travel_multiplier	2.5
asia_travel_multiplier	1.5
south_america_travel_multiplier	0.8
africa_travel_multiplier	0.3

B.2 Per-Bacteria Parameters (42 bacteria × N parameters)

Generated with key pattern bacteria_{name}_{param}:

Parameter suffix	Description
acquisition_log_odds	Baseline acquisition rate
hospital_acquisition_log_odds	Hospital acquisition rate
microbiome_vs_infection_log_odds	Likelihood of carriage vs infection
clearance_probability_per_day_no_treatment	Natural clearance rate
growth_rate_multiplier	Bacteria-specific growth modifier
symptom_onset_override	Override for symptom onset threshold
hgt_donor_probability	Probability of being an HGT donor
resistance_floor_enabled	Whether resistance floors apply
resistance_floor_ramp_years	Ramp-up period for floors
{drug_class}_resistance_floor	Per-class floor target

B.3 Per-Drug Parameters (58 drugs × N parameters)

Generated with key pattern drug_{name}_{param}:

Parameter suffix	Default	Description
half_life_days	Drug-specific	PK half-life
initial_level	10.0	Administration level
double_dose_multiplier	2.0	Double-dose level
spectrum_breadth	3.0	Microbiome disruption breadth
toxicity_death_hazard_per_unit_level	0.0	Per-drug toxicity hazard
toxicity_reservoir_half_life_days	1.5	Per-drug toxicity decay

B.4 Per-Region Parameters (6 regions × N parameters)

Generated with key pattern {region}_{param}:

Parameter suffix	Description
hospitalization_log_odds	Regional hospitalisation modifier
sepsis_onset_log_odds	Regional sepsis onset modifier
sepsis_mortality_multiplier	Regional sepsis death modifier
sepsis_recovery_log_odds	Regional sepsis recovery modifier
mortality_log_odds	Regional background mortality modifier
testing_multiplier	Regional testing rate modifier
antibiotic_initiation_log_odds	Regional prescribing modifier
drug_{drug}_availability	Per-drug availability (0.0–1.0)

B.5 Per-Mechanism Parameters (35 mechanisms × N parameters)

Generated with key pattern for mechanism-specific parameters:

Parameter pattern	Description
mech_{mechanism}_emergence_rate	Base emergence rate
mech_{mechanism}_reversion_rate	Fitness-cost reversion rate
mech_{mechanism}_enhancement_{drug_class}	Enhancement multiplier (how much resistance the mechanism confers against each drug class)
mech_{mechanism}_global_enhancement_multiplier	Legacy global multiplier

B.6 Cross-Indexed Parameters

Potency matrix (42 × 52)

Key: drug_{drug}_for_bacteria_{bacteria}_potency_when_no_r

Values: 0.0 (no activity) to 1.0 (maximum activity). See Section 6.5.

HGT probability matrix (42 × 42)

Key: hgt_prob_{source}_to_{target}

Values: see Section 9.2 for the probability rules.

Resistance emergence rates (42 × 35)

Key: bacteria_{bacteria}_mechanism_{mechanism}_emergence_rate

Values: base rate × incidence band multiplier. See Section 7.3.

Demographic distribution (6 × 18)

Key: demo_{region}_age_{start}_{end}

Values: probability weight for each region-age combination.

B.7 Syndrome Scoring Templates (10 syndromes)

See Section 6.2 for the full empiric scoring tables for all 10 syndromes.

B.8 Drug Penetration Table (10 syndromes × 18 drug classes)

See Section 6.4 for the penetration values by syndrome and drug class.

B.9 Cross-Resistance Groups

See Section 7.6 for per-bacteria cross-resistance groupings.

B.10 Drug Introduction Dates

See Section 6.6 for the full table of 58 drug introduction time steps.

B.11 Age-Specific Risk Templates

See Section 3.1 for the 7 risk templates and their multiplier vectors.

B.12 Regional Drug Availability

See Section 6.6 for the per-region availability tables.

---

Appendix C — Output Specification

C.1 Output File

Each simulation run produces a single CSV file:

amr_simulation_output_analysis_outputs/simulation_summary_NNNNNN.csv

where NNNNNN is a zero-padded run identifier.

C.2 Row Structure

Each row represents one simulated day. The number of rows equals the total number of time steps (default 38,325).

C.3 Column Categories

Scalar columns (per-timestep)

Column	Type	Description
day	int	Simulation day (0-indexed)
year	float	Calendar year (1930.0 + day/365.25)
total_alive	int	Living individuals
total_infected	int	Individuals with ≥1 active infection
total_on_treatment	int	Individuals receiving ≥1 antibiotic
total_in_hospital	int	Hospitalised individuals
total_sepsis	int	Individuals with sepsis
total_died_infection	int	Cumulative infection deaths
total_died_sepsis	int	Cumulative sepsis deaths
total_died_background	int	Cumulative background deaths
total_died_toxicity	int	Cumulative drug toxicity deaths
total_new_infections	int	New infections this day
drug_stops_due_to_toxicity	int	Drug courses stopped for toxicity this day
total_carriers	int	Individuals carrying ≥1 bacteria
total_immunosuppressed	int	Immunosuppressed individuals
policy_name	string	Active policy branch name

Per-bacteria columns (~42 each)

Pattern	Description
{bacteria}_infected	Currently infected count
{bacteria}_carriers	Current carrier count
{bacteria}_deaths	Cumulative deaths from this species
{bacteria}_new_infections	New infections this day
{bacteria}_new_infections_community	From community acquisition
{bacteria}_new_infections_hospital	From hospital acquisition
{bacteria}_new_infections_carrier	From carrier-to-infection
{bacteria}_sepsis	Current sepsis count

Per-drug columns (~58 each)

Pattern	Description
{drug}_prescribed	Courses initiated this day
{drug}_active_treatments	Currently on this drug

Per-bacteria × per-drug columns (~2,436 each)

Pattern	Description
{bacteria}_{drug}_activity_r	Mean resistance (activity_r)
{bacteria}_{drug}_majority_r	Population-level resistance prevalence

Per-region columns (~6 each)

Pattern	Description
{region}_infected	Regional infection count
{region}_hospitalized	Regional hospital count
{region}_deaths	Regional death count

C.4 Total Column Count

With 42 bacteria, 58 drugs, and 6 regions, the CSV contains approximately:

• ~16 scalar columns
• ~336 per-bacteria columns (42 × 8)
• ~116 per-drug columns (58 × 2)
• ~4,872 per-bacteria-per-drug columns (42 × 58 × 2)
• ~18 per-region columns (6 × 3)
• Total: ~5,358 columns

C.5 Infection Journey Logs

When enabled, individual infection journeys are logged to the infection_journeys/ directory as CSV files, capturing:

• Infection acquisition details
• Resistance profile at acquisition and over time
• Treatment episodes
• Clinical outcome (clearance, death, ongoing)
• Mechanism gains and losses

---

This document describes the model as implemented in the Rust codebase. All variable names correspond to parameter keys used in src/config.rs.