Co-transmitting STIs¶

STIsim is designed for modeling multiple STIs circulating on the same sexual network. This tutorial shows how to:

1. Run multiple bacterial STIs together
2. Add HIV with coinfection connectors
3. Compare results with and without coinfection interactions

Multiple STIs on a shared network¶

When you pass multiple diseases to a sim, they all transmit on the same StructuredSexual network. Each disease has its own transmission probability, natural history, and clearance dynamics -- but the partnerships are shared.

In [1]:

import stisim as sti

sim = sti.Sim(
    diseases=['ng', 'ct', 'tv'],
    n_agents=2000,
    start=2010,
    stop=2030,
)
sim.run(verbose=0)
sim.plot(key=['ng.prevalence', 'ct.prevalence', 'tv.prevalence'])

import stisim as sti sim = sti.Sim( diseases=['ng', 'ct', 'tv'], n_agents=2000, start=2010, stop=2030, ) sim.run(verbose=0) sim.plot(key=['ng.prevalence', 'ct.prevalence', 'tv.prevalence'])

Initializing sim with 2000 agents

Figure(800x600)

Each disease reaches its own endemic equilibrium determined by its transmission rate, duration of infection, and clearance dynamics. Trichomoniasis typically has the highest prevalence because of its long duration of infection in women.

Customizing disease parameters¶

Use sti_pars to override defaults for each disease:

In [2]:

sim = sti.Sim(
    diseases=['ng', 'ct', 'tv'],
    sti_pars=dict(
        ng=dict(beta_m2f=0.08, init_prev=0.03),
        ct=dict(beta_m2f=0.08, init_prev=0.05),
        tv=dict(beta_m2f=0.12, init_prev=0.08),
    ),
    n_agents=2000,
    start=2010,
    stop=2030,
)
sim.run(verbose=0)
sim.plot(key=['ng.prevalence', 'ct.prevalence', 'tv.prevalence'])

sim = sti.Sim( diseases=['ng', 'ct', 'tv'], sti_pars=dict( ng=dict(beta_m2f=0.08, init_prev=0.03), ct=dict(beta_m2f=0.08, init_prev=0.05), tv=dict(beta_m2f=0.12, init_prev=0.08), ), n_agents=2000, start=2010, stop=2030, ) sim.run(verbose=0) sim.plot(key=['ng.prevalence', 'ct.prevalence', 'tv.prevalence'])

Initializing sim with 2000 agents

Figure(800x600)

Adding HIV with connectors¶

When diseases interact biologically -- for example, having an STI increases susceptibility to HIV -- you model this with connectors. STIsim provides connectors for HIV-STI coinfection interactions.

Connectors modify rel_sus (relative susceptibility) and rel_trans (relative transmissibility) for coinfected agents at each timestep. For example, hiv_ng increases HIV susceptibility for agents who have gonorrhea.

In [3]:

# Create disease modules
hiv = sti.HIV(init_prev=0.10)
ng = sti.Gonorrhea(init_prev=0.03)
ct = sti.Chlamydia(init_prev=0.05)
tv = sti.Trichomoniasis(init_prev=0.08)

# Create connectors for HIV-STI interactions
connectors = [
    sti.hiv_ng(hiv, ng),   # NG increases HIV susceptibility (default: 1.2x)
    sti.hiv_ct(hiv, ct),   # CT increases HIV susceptibility (default: 1.0x)
    sti.hiv_tv(hiv, tv),   # TV increases HIV susceptibility (default: 1.5x)
]

sim = sti.Sim(
    diseases=[hiv, ng, ct, tv],
    connectors=connectors,
    n_agents=2000,
    start=2010,
    stop=2030,
)
sim.run(verbose=0)
sim.plot(key=['hiv.prevalence', 'ng.prevalence', 'ct.prevalence', 'tv.prevalence'])

# Create disease modules hiv = sti.HIV(init_prev=0.10) ng = sti.Gonorrhea(init_prev=0.03) ct = sti.Chlamydia(init_prev=0.05) tv = sti.Trichomoniasis(init_prev=0.08) # Create connectors for HIV-STI interactions connectors = [ sti.hiv_ng(hiv, ng), # NG increases HIV susceptibility (default: 1.2x) sti.hiv_ct(hiv, ct), # CT increases HIV susceptibility (default: 1.0x) sti.hiv_tv(hiv, tv), # TV increases HIV susceptibility (default: 1.5x) ] sim = sti.Sim( diseases=[hiv, ng, ct, tv], connectors=connectors, n_agents=2000, start=2010, stop=2030, ) sim.run(verbose=0) sim.plot(key=['hiv.prevalence', 'ng.prevalence', 'ct.prevalence', 'tv.prevalence'])

Initializing sim with 2000 agents

Figure(800x600)

Note: The HIV dynamics in this tutorial are not realistic -- we have not added ART or HIV testing, which are crucial drivers of the HIV epidemic. Without treatment, HIV prevalence here is governed entirely by natural history and AIDS mortality. For realistic HIV modeling, see the hiv_kenya example, which includes testing, ART, and PrEP interventions.

Comparing with and without connectors¶

To see the effect of coinfection interactions, we compare HIV and syphilis co-transmission with and without an hiv_syph connector.

In reality, syphilis epidemics are sustained by cycles of treatment and reinfection -- people with symptomatic (active) syphilis seek treatment, clearing the infection but leaving them susceptible to reinfection. Without treatment, most cases quickly progress to the latent stage, where they are far less infectious and the connector has little effect.

To make the connector's impact visible without adding treatment interventions, we artificially extend the primary stage duration. This keeps more agents in the active (primary + secondary) stages where the connector amplifies HIV susceptibility.

In [4]:

import matplotlib.pyplot as plt
import starsim as ss

def make_diseases():
    """ Helper to create fresh disease modules """
    hiv = sti.HIV(init_prev=0.02, beta_m2f=0.03)
    syph = sti.Syphilis(
        init_prev=0.10, beta_m2f=0.25,
        rel_trans_primary=5, rel_trans_latent=0.1,
        eff_condom=0.5,
        dur_primary=ss.constant(v=ss.years(2)),  # Extended for illustration
    )
    return hiv, syph

# Without connectors
hiv, syph = make_diseases()
s0 = sti.Sim(diseases=[hiv, syph], n_agents=5000, start=2000, stop=2020)
s0.run(verbose=0)

# With connectors -- syphilis increases HIV susceptibility 2.67x (default)
hiv, syph = make_diseases()
s1 = sti.Sim(
    diseases=[hiv, syph],
    connectors=sti.hiv_syph(hiv, syph),
    n_agents=5000, start=2000, stop=2020,
)
s1.run(verbose=0)

# Compare cumulative HIV infections
fig, ax = plt.subplots()
ax.plot(s0.timevec, s0.results.hiv.cum_infections, label='Without connector')
ax.plot(s1.timevec, s1.results.hiv.cum_infections, label='With hiv_syph connector')
ax.set_xlabel('Year')
ax.set_ylabel('Cumulative HIV infections')
ax.set_title('Effect of syphilis coinfection on HIV acquisition')
ax.legend()
fig

import matplotlib.pyplot as plt import starsim as ss def make_diseases(): """ Helper to create fresh disease modules """ hiv = sti.HIV(init_prev=0.02, beta_m2f=0.03) syph = sti.Syphilis( init_prev=0.10, beta_m2f=0.25, rel_trans_primary=5, rel_trans_latent=0.1, eff_condom=0.5, dur_primary=ss.constant(v=ss.years(2)), # Extended for illustration ) return hiv, syph # Without connectors hiv, syph = make_diseases() s0 = sti.Sim(diseases=[hiv, syph], n_agents=5000, start=2000, stop=2020) s0.run(verbose=0) # With connectors -- syphilis increases HIV susceptibility 2.67x (default) hiv, syph = make_diseases() s1 = sti.Sim( diseases=[hiv, syph], connectors=sti.hiv_syph(hiv, syph), n_agents=5000, start=2000, stop=2020, ) s1.run(verbose=0) # Compare cumulative HIV infections fig, ax = plt.subplots() ax.plot(s0.timevec, s0.results.hiv.cum_infections, label='Without connector') ax.plot(s1.timevec, s1.results.hiv.cum_infections, label='With hiv_syph connector') ax.set_xlabel('Year') ax.set_ylabel('Cumulative HIV infections') ax.set_title('Effect of syphilis coinfection on HIV acquisition') ax.legend() fig

Initializing sim with 5000 agents

Initializing sim with 5000 agents

Out[4]:

Available connectors¶

Connector	Interaction	Default effect
sti.hiv_syph	HIV-Syphilis	Syphilis increases HIV susceptibility 2.67x and transmissibility 1.2x
sti.hiv_ng	HIV-Gonorrhea	NG increases HIV susceptibility 1.2x and transmissibility 1.2x
sti.hiv_ct	HIV-Chlamydia	CT increases HIV susceptibility 1.0x (placeholder)
sti.hiv_tv	HIV-Trichomonas	TV increases HIV susceptibility 1.5x
sti.hiv_bv	HIV-BV	BV (CST4) increases HIV susceptibility and transmissibility 2x
sti.gud_syph	GUD-Syphilis	Bidirectional: each increases susceptibility and transmissibility 2x

All connector parameters can be overridden at creation time, e.g.:

sti.hiv_ng(hiv, ng, rel_sus_hiv_ng=2.0, rel_trans_hiv_ng=2.0)

Exercises¶

1. Add syphilis to the model with an hiv_syph connector. How does it change HIV prevalence compared to the bacterial STIs?
2. Try increasing the rel_sus_hiv_tv parameter to 3.0. What happens to HIV prevalence?
3. Add demographics (demographics='zimbabwe') and observe how population turnover affects the co-transmission dynamics.