Skip to main content

An STI on a sexual network

🧠0% 🤖100%

This model restricts transmission to sexual partnerships, which form and dissolve over time, so how fast the infection spreads depends on who partners with whom and for how long, not just on the average number of partners. That needs a model of individual people and their partnerships: a network, and so agents. This page models a curable bacterial STI (an SIS model: people become susceptible again when cured) on Starsim’s sexual network of men and women, with a different chance of transmission per sex act from men to women and from women to men. The parameter values are illustrative, not estimates for a particular infection.

The model

Parameter Value Meaning
Transmission per act, male to female 8% Chance that one sex act with an infected partner infects a woman
Transmission per act, female to male 5% The same, for a man
Duration of infection 5 years Mean time in I (exponential), until cured
Partnerships ss.SexualNet() defaults Partnerships between men and women, with acts per year in each
Agents 2,000
Infected at the start 5%
Time step 1 month Over 10 years 
import starsim as ss

S, I = ss.states('S I')
infection = S >> I @ dict(sexual=ss.peract(m2f=0.08, f2m=0.05)) * I  # Per act, on the sexual network
recovery  = I >> S @ ss.years(5)
sti = ss.Disease(infection, recovery, init_prev=0.05, name='sti')

The chance of transmission is given per act (ss.peract), for each direction of the network, which has men on one side and women on the other. The diagram, from diagram():

sti.diagram()

Result

Since this model has a network, Starsim runs it as agents. With 2,000 people, each run is different, so it is run 20 times, each with its own random seed:

sim = ss.Sim(sti, networks=ss.SexualNet(), n_agents=2_000, start=2000, stop=2010, dt=ss.months(1), verbose=0)
msim = ss.parallel(sim, n_runs=20)
msim.plot('sti')

The plot shows the median of the runs and the range that holds 95% of them. The number infected rises in the first year, as infections pass to the partners of the people infected at the start, and then falls slowly, as people are cured faster than new partnerships bring them new infections. Since in this model an infection can only pass to a current partner, how many partners people have, and how long partnerships last, matter as much as the chance of transmission per act. The number infected at the end, and the infections over the ten years:

msim.summarize().loc[['sti_I', 'sti_infection']]
median low high mean std
result
sti_I 105 87.9 121 104 9.05
sti_infection 252 195 294 247 30.6

Going further