Overview
β’ Introduction to epidemics on networksβ’ Description of moment-closure
representationβ’ Description of βMessage-passingβ
representationβ’ Comparison of methods
Route 2: Water flow (down stream)Modelling aquatic infectious disease
Jonkers et al. (2010) Epidemics
The SIR compartmental model
SI
R
Infection
Removal
All processes Poisson
SusceptibleInfectiousRemoved
States of individual nodes could be:
Moment closure & BBGKY hierarchy
iS Probability that node i is Susceptible
N
jjiiji ISTS
1
i
N
jjiiji IgISTI
1
Sharkey, K.J. (2008) J. Math Biol., Sharkey, K.J. (2011) Theor. Popul. Biol.
i jπ ππ
N
jjiiji ISTS
1
i
N
jjiiji IgISTI
1
jijk
jiijjikikik
kjijkji ISgISTISITISSTIS
i j i ji j
k
i j
k
Sharkey, K.J. (2008) J. Math Biol., Sharkey, K.J. (2011) Theor. Popul. Biol.
Moment closure & BBGKY hierarchy
N
jjiiji ISTS
1
i
N
jjiiji IgISTI
1
jijk
jiijjikikik
kjijkji ISgISTISITISSTIS
ik
kjijk
jikikji ISSSSITSS
Sharkey, K.J. (2008) J. Math Biol., Sharkey, K.J. (2011) Theor. Popul. Biol.
Hierarchy provably exact at all orders
j
kjji
kjiB
CBBACBA iA
jB
kC
To close at second order can assume:
Moment closure & BBGKY hierarchy
Sharkey, K.J. (2008) J. Math Biol., Sharkey, K.J. (2011) Theor. Popul. Biol.
Random Network of 100 nodes
Sharkey, K.J. (2008) J. Math Biol., Sharkey, K.J. (2011) Theor. Popul. Biol.
Random Network of 100 nodes
Random K-Regular Network
Sharkey, K.J. (2008) J. Math Biol., Sharkey, K.J. (2011) Theor. Popul. Biol.
Locally connected Network
Sharkey, K.J. (2008) J. Math Biol., Sharkey, K.J. (2011) Theor. Popul. Biol.
Example: Tree graph
For any tree, these equations are exact
Sharkey, Kiss, Wilkinson, Simon. B. Math. Biol . (2013)
Extensions to Networks with Clustering
1 2
3
4
1 2
3
5 4
β¨ πΌ 1 πΌ 2π3 πΌ 4 β©=β¨ πΌ 1 πΌ 2π3 β© β¨π3 πΌ 4 β©
β¨π3 β©
β¨ πΌ 1 πΌ 2π3 πΌ 4 πΌ5 β©=β¨ πΌ 1 πΌ 2π3 β© β¨π3 πΌ 4 πΌ 5 β©
β¨π3 β©
Kiss, Morris, Selley, Simon, Wilkinson (2013) arXiv preprint arXiv:1307.7737
Application to SIS dynamics
i
N
jjiiji IgISTS
1
i
N
jjiiji IgISTI
1
jijijk
jiijjikikik
kjijkji IIgISgISTISITISSTIS
jijiik
kjijk
jikikji SIgISgISSSSITSS
j
kjji
kjiB
CBBACBA Closure:
Nagy, Simon Cent. Eur. J. Math. 11(4) (2013)
Exact on tree networks
Can be extended to exact models on clustered networks
Can be extended to other dynamics (e.g. SIS)
Problem: Limited to Poisson processes
Moment-closure model
Karrer and Newman Message-Passing
Cavity state
i
Fundamental quantity: : Probability that i has not received an infectious contact from j when i is in the cavity state.
j
Karrer B and Newman MEJ, Phys Rev E 84, 036106 (2010)
Karrer and Newman Message-Passing
Cavity state
Fundamental quantity: : Probability that i has not received an infectious contact from j when i is in the cavity state.
i
j
is the probability that j has not received an infectious contact by time t from any of its neighbours when i and j are in the cavity state.
Karrer B and Newman MEJ, Phys Rev E 84, 036106 (2010)
Karrer and Newman Message Passing
Fundamental quantity: : Probability that i has not received an infectious contact from j when i is in the cavity state.
is the probability that j has not received an infectious contact by time t from any of its neighbours when i and j are in the cavity state.
Define: of being infected is: (Combination of infection process and removal ).
π ππ (π )ππ=π ππ (π )(1ββ«0
π
π π (π β² )ππ β²)Message passing equation:
β¨π π β©=π§ π βπβπ π
π» πβ π (π‘ ) β¨π π β©=β«0
π‘
π (π ) [1β β¨ππ β©π‘ βπ ]ππ β¨ πΌ π β©=1β β¨ππ β©β β¨π π β©
Karrer B and Newman MEJ, Phys Rev E 84, 036106 (2010)
1 if j initially susceptibleΒΏ π ππ (π )β«
π
β
π π (π β² )ππ β²
1) Applies to arbitrary transmission and removal processes
2) Not obvious to see how to extend it to other scenarios including generating exact models with clustering and dynamics such as SIS
Karrer B and Newman MEJ, Phys Rev E 84, 036106 (2010)
Useful to relate the two formalisms to each other
Karrer and Newman Message-Passing
Relationship to moment-closure equations
β¨π ππ π β©=π§ π π§ πΞ¦ ππ (π‘ )Ξ¦π
π (π‘ )
β¨π π πΌ π β©=π§ πΞ¦ ππ (π‘ )[1βπ§ πΞ¦π
π (π‘ )ββ«0
π‘
( π ππ (π )+πππ (π ) )Γ (1βπ§ π )Ξ¦ππ (π‘βπ )ππ ]
π» πβ π=1ββ«0
π‘
π ππ (π ) [1βπ§ πΞ¦ππ (π‘βπ ) ]ππ
When the contact processes are Poisson, we have:
ππ» πβ π (π‘ ) ππ‘
=βπ ππ [π» πβ π (π‘ )βπ§ πΞ¦ππ (π‘ )ββ«
0
π‘
π π (π )πβπ πππ (1βπ§πΞ¦ππ (π‘βπ ) )ππ ]
so:
Wilkinson RR and Sharkey KJ, Phys Rev E 89, 022808 9 (2014)
Relationship to moment-closure equations
When the removal processes are also Poisson:
jijk
jiijjikikik
kjijkji ISgISTISITISSTIS
jijk
jiijjikikik
kjijkji ISgISTISITISSTIS
β¨ π΄ππ΅ ππΆπ β©=β¨π΄ππ΅ π β© β¨π΅ ππΆπ β©
β¨πΆπ β©Wilkinson RR and Sharkey KJ, Phys Rev E 89, 022808 9 (2014)
Relationship to moment-closure equations
When the removal process is fixed, Let
Wilkinson RR and Sharkey KJ, Phys Rev E 89, 022808 9 (2014)
Summary part 1
Exact correspondence with stochastic simulation for tree networks.
Extensions to: β’ Exact models in networks with clusteringβ’ Non-SIR dynamics (eg SIS).
Pair-based moment closure:
Message passing:Exact on trees for arbitrary transmission and removal processes
Not clear how to extend to models with clustering or other dynamics
Limited to Poisson processes
Summary part 2
Extension of message passing models to include:a)Heterogeneous initial conditionsb)Heterogeneous transmission and removal processes
Extension of the pair-based moment-closure models to include arbitrary removal processes.
Linking the models enabled:
Proof that the pair-based SIR models provide a rigorous lower bound on the expected Susceptible time series.