Antiviral Stockpile Model

Antivirals are modeled as a stockpile-constrained intervention. They move people from eligible compartments into T by allocation, not by a disease-rate equation. This is intentionally similar to vaccination: a stockpile is released, distributed to nodes, and then applied within each node according to eligibility and priority. Releases and allocation are deterministic fixed-count transfers; the stockpile model does not draw a random number of doses or randomly sample who receives a dose.

For SEITRS, eligible compartments are usually E and I. For SEITHRD, the antiviral-capable SEIHRD variant, eligible compartments can be E, IA, IP, and IS. For SEAITRD, the standard antiviral route is I -> T; the T compartment is required in this model family even when no doses are released.

Complete Configuration

The only supported antiviral model identity is stockpile-age-risk. A complete five-age-group configuration looks like:

"antiviral_model": {
  "identity": "stockpile-age-risk",
  "parameters": {
    "age_risk_priority_groups": ["0.0", "0.5", "1.0", "1.0", "0.5"],
    "compartment_priority": ["IS"],
    "antiviral_effectiveness_hosp": "0.25",
    "antiviral_capacity_proportion": "0.10",
    "antiviral_half_life_days": "30",
    "antiviral_stockpile": [
      {"day": "0", "amount": "1000"},
      {"day": "14", "amount": "500"}
    ]
  }
}

JSON numbers or numeric strings are accepted for the numeric values shown above.

age_risk_priority_groups

This list uses the same targeting semantics as vaccination. It must contain one value for each configured age group, in the same order as the model’s age groups.

Value	Eligible people in that age group
0.0	No one.
0.5	High-risk people only.
1.0	Everyone, including both low- and high-risk people.

No other values are allowed. For example:

"age_risk_priority_groups": ["0.0", "0.5", "1.0", "1.0", "0.5"]

means:

• Nobody in age group 0 is eligible

• Only high-risk people in age groups 1 and 4 are eligible

• Everyone in age groups 2 and 3 is eligible

This parameter controls age/risk eligibility, not ordering between age groups. Available doses are allocated proportionally among eligible groups. Both vaccinated and unvaccinated people may receive antivirals if they satisfy the age/risk and compartment criteria.

If supplies, daily capacity, and eligible compartment counts are sufficient, everyone who meets the configured age/risk and compartment criteria can receive an antiviral. People outside that eligibility are excluded from the scenario, even if unused doses remain. To test a narrower or broader target population, run a separate scenario with a different eligibility vector.

If omitted, every age group defaults to 1.0.

compartment_priority

This list defines the disease states eligible for antiviral treatment and the order in which they are treated within an age/risk/vaccination group. It does not prioritize one age group over another.

"compartment_priority": ["IS"]

With this SEITHRD configuration, available doses move people from symptomatic infectious IS to T. Every label must exist in the active model’s compartment list. If omitted, the default is ["I"]. For SEITHRD symptomatic treatment scenarios, set compartment_priority explicitly to ["IS"].

Including E, IA, or IP in SEITHRD changes the scenario from symptomatic treatment toward prophylaxis or early treatment. Those people may spend more calendar time in T than someone treated only after entering IS, so the meaning of T_to_R_days and treated infectious time changes. The current stockpile model treats eligible compartment members directly; it does not yet represent a separate proportion or household-contact multiplier, such as treating 1 to 5 family members per infectious person. Use earlier compartments only for explicit prophylaxis scenarios.

antiviral_effectiveness_hosp

This optional SEITHRD parameter is the treated relative reduction in hospitalization risk. It accepts a scalar or one value per age group, with values from 0.0 to 1.0.

For oseltamivir-like treatment (Tamiflu), "0.25" represents a 25% hospitalization risk reduction, or 75% relative risk, for treated people. Treated people can still be hospitalized: SEITHRD moves T -> H or T -> R, and this parameter reduces only the realized T -> H risk. If omitted, the default is 1.0, meaning complete protection from hospitalization for treated people. Set an explicit smaller value, such as 0.25, when modeling partial protection.

The treated hospitalization branch is anchored to the original symptomatic hospitalization clock, IS_to_H_days. Internally, the model uses target_rate = 1 / IS_to_H_days for both untreated IS -> H and treated T -> H; treatment changes the desired realized hospitalization proportion, not the baseline hospitalization clock itself.

For example, suppose a low-risk age group has the following parameters and the default complete protection has been overridden with partial protection:

"IS_to_H_days": "3.0",
"IS_to_R_days": "7.0",
"T_to_R_days": "5.0",
"prop_IS_to_H_lowrisk": ["0.10"],
"antiviral_effectiveness_hosp": "0.25"

Untreated symptomatic people have an eventual hospitalization proportion of 0.10, so about 100 of 1,000 people in IS would eventually move to H. For treated people, the model computes:

treated hospitalization proportion = 0.10 * (1 - 0.25) = 0.075

So about 75 of 1,000 people in T would eventually move to H, while the remaining treated exits go to R. The model still uses the original IS_to_H_days rate for the T -> H target clock, then adjusts the competing T -> H versus T -> R branch multipliers so the realized treated hospitalization proportion is 0.075.

With the numbers above, the T -> H target rate is still 1 / IS_to_H_days = 1 / 3 = 0.333 per day before branch weighting. The competing-clock adjustment combines that target clock with T -> R = 1 / T_to_R_days = 1 / 5 = 0.200 per day and chooses branch weights that realize 7.5% hospitalized among treated people. Treatment does not replace IS_to_H_days; it lowers the realized hospitalization proportion relative to the untreated IS pathway and shortens the recovery clock from IS_to_R_days = 7.0 to T_to_R_days = 5.0.

antiviral_effectiveness_death

This SEAITRD antiviral parameter is the treated relative reduction in mortality risk. It accepts a scalar or one value per age group, with values from 0.0 to 1.0. Provide it when an antiviral model is configured for SEAITRD.

For example, "0.25" represents a 25% mortality risk reduction, or 75% relative risk, for treated people. In SEAITRD, this parameter reduces the mortality intensity used for treated T -> D events:

treated mortality rate = risk-specific I_to_D rate * (1 - antiviral_effectiveness_death)

For high-risk SEAITRD groups, the risk-specific untreated rate is I_to_D_invdays * highrisk_death_multiplier. Set 0.0 to preserve the untreated SEAITRD mortality rate for treated people. Set 1.0 only when treated people should have no T -> D event.

antiviral_capacity_proportion

This value limits how many people can be treated in one node on one day. The daily limit is:

\[\left\lfloor \text{antiviral_capacity_proportion} \times \text{node population} \right\rfloor.\]

For example, "0.10" allows at most 10% of a node’s total population to begin treatment that day. Unused doses roll forward to the next day. The default is 1.0, allowing treatment up to the node’s full population per day.

antiviral_half_life_days

This optional positive value models stockpile loss over time. For a half-life \(h\), stock remaining after day 0 is multiplied each day by:

\[0.5^{1/h}.\]

For example, "30" gives a 30-day stockpile half-life. Use null, or omit the parameter, to disable stockpile decay. This affects unused doses, not people already in T.

antiviral_stockpile

This list schedules doses entering the network stockpile:

"antiviral_stockpile": [
  {"day": "0", "amount": "1000"},
  {"day": "14", "amount": "500"}
]

Each day is an integer simulation day and each amount is a dose count. Multiple entries on the same day are combined. A negative release day is reassigned to day 0 with a warning. If no eligible people are available, unused doses roll forward to the next day. If omitted, the stockpile is empty. The configured amount is the number of doses entering the stockpile on that day; it is not noised or sampled.

On each release day, doses are distributed among nodes in proportion to each node’s currently eligible population. Within a node, doses are allocated proportionally among eligible age/risk/vaccination groups and then applied according to compartment_priority. Fractional proportional shares are turned into integer transfer counts with deterministic rounding and availability limits. The only reason fewer people move into T than the release amount is that doses are limited by eligibility, capacity, stockpile decay, or available people in the prioritized compartments.

Disease Parameters

Disease parameters then describe treated people:

Parameter	Meaning
T_to_R_days	Average duration from treated infectious to recovered.
rel_inf_T_to_I	Relative infectiousness of treated people compared with untreated I.
rel_inf_T_to_IS	Relative infectiousness of treated people compared with symptomatic IS in SEITHRD.
antiviral_effectiveness_hosp	Antiviral-model parameter that reduces T -> H risk in SEITHRD; scalar or one value per age group. Defaults to 1.0, complete protection from hospitalization.
antiviral_effectiveness_death	Antiviral-model parameter that reduces T -> D risk in SEAITRD; scalar or one value per age group. Required when SEAITRD uses an antiviral model.

In SEITRS and SEITHRD, if the compartment list includes T but no antiviral model is configured, the simulator warns. Those models can still run, but no people enter T.

There are no disease-model rates such as E_to_T_days or I_to_T_days. Movement into T is fully determined by released doses, daily capacity, the population in compartment_priority. In SEITRS and SEITHRD, movement out of T is controlled by T_to_R_days; SEITHRD also allows treated people to enter H, with risk reduced by antiviral_effectiveness_hosp. SEAITRD also uses T_to_R_days, separately from I_to_R_days, and derives treated mortality from I_to_D_invdays and antiviral_effectiveness_death.

SEAITRD

SEAITRD requires T, and T means Treated, consistently with the other antiviral-capable models. Anyone in T is assumed to be receiving an antiviral, but no one enters T without stockpile allocation unless the input explicitly initializes that compartment.

In the Gillespie stochastic SEAITRD model, untreated infection trajectories do not enter T. Natural progression uses E -> A -> I; A behaves like a pre-symptomatic compartment and does not recover or die directly.

When the optional stockpile model allocates an antiviral dose to someone in I, the simulator:

1. moves one person into T;

2. marks the person’s old source-compartment event trajectory as stale;

3. queues a new trajectory beginning in T.

Stockpile-created T entries follow the existing SEAITRD competing events: T -> R or T -> D. Treated recovery uses T_to_R_days, while untreated recovery uses I_to_R_days. Treated mortality is derived from the risk-specific untreated mortality rate and antiviral_effectiveness_death.

SEAITRD contact events are stored at the demographic-group level, not linked to person identifiers. Antiviral reconciliation replaces progression events but keeps the existing contact queue. This is consistent with the current SEAITRD assumption that A, T, and I use the same infectious contact process.