mathematical modeling of dengue viral infection modeling of dengue viral infection ryan nikin-beers...
TRANSCRIPT
Mathematical Modeling of Dengue Viral Infection
Ryan Nikin-Beers
October 1, 2014
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 1 / 45
1 IntroductionBiological Background
2 Within-Host ModelPrimary InfectionSecondary InfectionDiscussion
3 Epidemiological Model
4 Future Work
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 2 / 45
Introduction
1 IntroductionBiological Background
2 Within-Host ModelPrimary InfectionSecondary InfectionDiscussion
3 Epidemiological Model
4 Future Work
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 3 / 45
Introduction Biological Background
1 IntroductionBiological Background
2 Within-Host ModelPrimary InfectionSecondary InfectionDiscussion
3 Epidemiological Model
4 Future Work
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 4 / 45
Introduction Biological Background
Immune System
Innate vs Adaptive
B-cells vs T-cells
Neutralizing Antibodies
Non-neutralizing Antibodies
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 5 / 45
Introduction Biological Background
Immune System
Innate vs Adaptive
B-cells vs T-cells
Neutralizing Antibodies
Non-neutralizing Antibodies
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 5 / 45
Introduction Biological Background
Immune System
Innate vs Adaptive
B-cells vs T-cells
Neutralizing Antibodies
Non-neutralizing Antibodies
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 5 / 45
Introduction Biological Background
Immune System
Innate vs Adaptive
B-cells vs T-cells
Neutralizing Antibodies
Non-neutralizing Antibodies
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 5 / 45
Introduction Biological Background
Dengue Viral Infection
Mosquito-borne disease
50 million infections
500,000 hospitalizations
Tropical regions
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 6 / 45
Introduction Biological Background
Dengue Viral Infection
Mosquito-borne disease
50 million infections
500,000 hospitalizations
Tropical regions
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 6 / 45
Introduction Biological Background
Dengue Viral Infection
Mosquito-borne disease
50 million infections
500,000 hospitalizations
Tropical regions
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 6 / 45
Introduction Biological Background
Dengue Viral Infection
Mosquito-borne disease
50 million infections
500,000 hospitalizations
Tropical regions
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 6 / 45
Introduction Biological Background
Dengue Viral Infection
Four serotypes (strains)
Primary infection leads to dengue fever (DF)
DF characterized by rash, fever, headache
Lifelong immunity from serotype
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 7 / 45
Introduction Biological Background
Dengue Viral Infection
Four serotypes (strains)
Primary infection leads to dengue fever (DF)
DF characterized by rash, fever, headache
Lifelong immunity from serotype
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 7 / 45
Introduction Biological Background
Dengue Viral Infection
Four serotypes (strains)
Primary infection leads to dengue fever (DF)
DF characterized by rash, fever, headache
Lifelong immunity from serotype
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 7 / 45
Introduction Biological Background
Dengue Viral Infection
Four serotypes (strains)
Primary infection leads to dengue fever (DF)
DF characterized by rash, fever, headache
Lifelong immunity from serotype
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 7 / 45
Introduction Biological Background
Dengue Viral Infection
Secondary infection with heterologous serotype
Increased risk of dengue hemorrhagic fever (DHF) or dengue shocksyndrome (DSS)
DHF characterized by high blood platelet count, bleeding, and liverdamage
DSS characterized by high blood pressure, internal bleeding, andpossible shock
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 8 / 45
Introduction Biological Background
Dengue Viral Infection
Secondary infection with heterologous serotype
Increased risk of dengue hemorrhagic fever (DHF) or dengue shocksyndrome (DSS)
DHF characterized by high blood platelet count, bleeding, and liverdamage
DSS characterized by high blood pressure, internal bleeding, andpossible shock
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 8 / 45
Introduction Biological Background
Dengue Viral Infection
Secondary infection with heterologous serotype
Increased risk of dengue hemorrhagic fever (DHF) or dengue shocksyndrome (DSS)
DHF characterized by high blood platelet count, bleeding, and liverdamage
DSS characterized by high blood pressure, internal bleeding, andpossible shock
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 8 / 45
Introduction Biological Background
Dengue Viral Infection
Secondary infection with heterologous serotype
Increased risk of dengue hemorrhagic fever (DHF) or dengue shocksyndrome (DSS)
DHF characterized by high blood platelet count, bleeding, and liverdamage
DSS characterized by high blood pressure, internal bleeding, andpossible shock
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 8 / 45
Introduction Biological Background
Antibody Dependent Enhancement
Neutralizing antibodies bind to virus and attach to Fc-receptors onhost cell
Cells with Fc-receptors not normally infected since antibodyneutralizes virus beforehand
Fc-receptor cells subsequently ingested in a process known asphagocytosis
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 9 / 45
Introduction Biological Background
Antibody Dependent Enhancement
Neutralizing antibodies bind to virus and attach to Fc-receptors onhost cell
Cells with Fc-receptors not normally infected since antibodyneutralizes virus beforehand
Fc-receptor cells subsequently ingested in a process known asphagocytosis
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 9 / 45
Introduction Biological Background
Antibody Dependent Enhancement
Neutralizing antibodies bind to virus and attach to Fc-receptors onhost cell
Cells with Fc-receptors not normally infected since antibodyneutralizes virus beforehand
Fc-receptor cells subsequently ingested in a process known asphagocytosis
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 9 / 45
Introduction Biological Background
Antibody Dependent Enhancement
Long-lived antibodies remain from primary infection
These antibodies unequipped to neutralize virus
Fc-receptor cells infected
Leads to more cells being infected, thus more virus produced
More virus correlates with more severe infection
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 10 / 45
Introduction Biological Background
Antibody Dependent Enhancement
Long-lived antibodies remain from primary infection
These antibodies unequipped to neutralize virus
Fc-receptor cells infected
Leads to more cells being infected, thus more virus produced
More virus correlates with more severe infection
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 10 / 45
Introduction Biological Background
Antibody Dependent Enhancement
Long-lived antibodies remain from primary infection
These antibodies unequipped to neutralize virus
Fc-receptor cells infected
Leads to more cells being infected, thus more virus produced
More virus correlates with more severe infection
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 10 / 45
Introduction Biological Background
Antibody Dependent Enhancement
Long-lived antibodies remain from primary infection
These antibodies unequipped to neutralize virus
Fc-receptor cells infected
Leads to more cells being infected, thus more virus produced
More virus correlates with more severe infection
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 10 / 45
Introduction Biological Background
Antibody Dependent Enhancement
Long-lived antibodies remain from primary infection
These antibodies unequipped to neutralize virus
Fc-receptor cells infected
Leads to more cells being infected, thus more virus produced
More virus correlates with more severe infection
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 10 / 45
Introduction Biological Background
Antibody Dependent Enhancement1
1Whitehead, et al. ”Prospects for a dengue virus vaccine.” Nature, 2007.Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 11 / 45
Introduction Biological Background
Vaccination
Currently no vaccine for dengue
Must avoid making population susceptible to more severe disease
Must protect against all serotypes at once
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 12 / 45
Within-Host Model
1 IntroductionBiological Background
2 Within-Host ModelPrimary InfectionSecondary InfectionDiscussion
3 Epidemiological Model
4 Future Work
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 13 / 45
Within-Host Model
Goals
Develop a model for primary infection based on known dynamics
Develop a model for secondary infection with heterologous serotype
Fit secondary infection model to known data
Infer from these results the underlying causes of more severe disease
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 14 / 45
Within-Host Model
Goals
Develop a model for primary infection based on known dynamics
Develop a model for secondary infection with heterologous serotype
Fit secondary infection model to known data
Infer from these results the underlying causes of more severe disease
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 14 / 45
Within-Host Model
Goals
Develop a model for primary infection based on known dynamics
Develop a model for secondary infection with heterologous serotype
Fit secondary infection model to known data
Infer from these results the underlying causes of more severe disease
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 14 / 45
Within-Host Model
Goals
Develop a model for primary infection based on known dynamics
Develop a model for secondary infection with heterologous serotype
Fit secondary infection model to known data
Infer from these results the underlying causes of more severe disease
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 14 / 45
Within-Host Model Primary Infection
1 IntroductionBiological Background
2 Within-Host ModelPrimary InfectionSecondary InfectionDiscussion
3 Epidemiological Model
4 Future Work
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 15 / 45
Within-Host Model Primary Infection
Primary Infection Model3
dT
dt= s − dTT − βTV
1 + ηA,
dI
dt=
βTV
1 + ηA− δI ,
dV
dt= pI − cV − γVA,
dB
dt= −αBV − dBB,
dBa
dt= αBV − kBaV − dBaBa,
dP
dt= rP
(1 − P
KP
)+ kBaV ,
dA
dt= NP − dAA.
(1)
3Nikin-Beers, R and Ciupe, S. “The role of antibody in dengue viral infection.” MathBiosciences, submitted 2014.
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 16 / 45
Within-Host Model Primary Infection
Steady States
System (1) has four steady states. The disease-free steady state is given by
S1 =
(s
dT, 0, 0, 0, 0, 0, 0
).
The virus persistence in the absence of antibody responses is given by
S2 = (T2, I2,V2, 0, 0, 0, 0).
The virus persistence in the presence of antibody responses is given by
S3 = (T3, I3,V3, 0, 0,KP ,KA).
The antibody-induced virus clearance is given by
S4 =
(s
dT, 0, 0, 0, 0,KP ,KA
).
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 17 / 45
Within-Host Model Primary Infection
Stability
Steady state S2 exists when R0 = βpsdT cδ
> 1Steady states S1 and S2 are unstable.Steady state S3 is locally asymptotically stable if Rp
0 = R0
(1+ηKA)(1+γcKA)
> 1
Steady state S4 is locally asymptotically stable if Rp0 < 1.
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 18 / 45
Within-Host Model Primary Infection
Primary Infection Dynamics
Primary infection viremia peak 3-4 days after infection
Primary infection viremia clear at 7-10 days
Viremia always clears (S4 is locally asymptotically stable)
Antibodies above limit of detection between one to two weeks afterinfection
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 19 / 45
Within-Host Model Primary Infection
Primary Infection Dynamics
Primary infection viremia peak 3-4 days after infection
Primary infection viremia clear at 7-10 days
Viremia always clears (S4 is locally asymptotically stable)
Antibodies above limit of detection between one to two weeks afterinfection
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 19 / 45
Within-Host Model Primary Infection
Primary Infection Dynamics
Primary infection viremia peak 3-4 days after infection
Primary infection viremia clear at 7-10 days
Viremia always clears (S4 is locally asymptotically stable)
Antibodies above limit of detection between one to two weeks afterinfection
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 19 / 45
Within-Host Model Primary Infection
Primary Infection Dynamics
Primary infection viremia peak 3-4 days after infection
Primary infection viremia clear at 7-10 days
Viremia always clears (S4 is locally asymptotically stable)
Antibodies above limit of detection between one to two weeks afterinfection
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 19 / 45
Within-Host Model Primary Infection
Primary Infection Dynamics
0 20 40 60 80 100104
10 5
106
Day After Infection
Tar
get
Cel
ls
Target Cells
0 2 4 6 8 10
10 2
104
106
Day After Infection
Infe
cted
Cel
ls
Infected Cells
0 2 4 6 8 10
10 3
106
109
Day After Infection
Vir
alL
oad
Virus
0 2 4 6 8 1010 -8
10 -1
106
Day After Infection
BC
ells
B Cells
0 10 20 30 40 50
10 1
10 3
10 5
Day After Infection
Pla
sma
Cel
ls
Plasma Cells
0 10 20 30 40 50
10 10
10 12
10 14
Day After Infection
Ant
ibo
die
s
Antibodies
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 20 / 45
Within-Host Model Secondary Infection
1 IntroductionBiological Background
2 Within-Host ModelPrimary InfectionSecondary InfectionDiscussion
3 Epidemiological Model
4 Future Work
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 21 / 45
Within-Host Model Secondary Infection
Secondary Infection Model
The new model is
dT
dt= s − dTT − β1TV1
1 + ηA1− β2TV2
1 + ηA2,
dIidt
=βiTVi
1 + ηAi− δIi ,
dV1
dt= p1I1 − (c + γ1A1)V1,
dV2
dt= p2I2 − (c + γ2A2 − γEA1)V2,
dB
dt= −αBV1 − αBV2 − dBB,
dBai
dt= αBVi − kBaiVi − dBaBai ,
dPi
dt= rPi (1 − Pi
KP) + kBaiVi ,
dAi
dt= NPi − dAAi ,
(2)
where i = 1, 2.
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 22 / 45
Within-Host Model Secondary Infection
Steady States
Assume S4 is stable in original model. Then we only consider four steadystates of model 2.
S5 = (s
dT, 0, 0, 0, 0, 0, 0, 0, 0,KP , 0,KA, 0),
S6 = (T6, 0, I6, 0,V6, 0, 0, 0,KP , 0,KA, 0),
S7 = (T7, 0, I7, 0,V7, 0, 0, 0,KP ,KP ,KA,KA),
S8 = (s
dT, 0, 0, 0, 0, 0, 0, 0,KP ,KP ,KA,KA).
(3)
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 23 / 45
Within-Host Model Secondary Infection
Stability
Steady state S6 only exists when R02 = β2p2sdT cδ
> 1.Steady states S5 and S6 are unstable.Steady state S7 is locally asymptotically stable whenRs0 = R02
(1+ηKA)(1+γ2−γE
cKA)
> 1
Steady state S8 is locally asymptotically stable when Rs0 < 1.
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 24 / 45
Within-Host Model Secondary Infection
Data4
4Wang, et al. “Slower Rates of Clearance of Viral Load and Virus-ContainingImmune Complexes in Patients with Dengue Hemorrhagic Fever.” Clin Infect Dis, 2006.
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 25 / 45
Within-Host Model Secondary Infection
Secondary DF vs DHF
Higher viral peak in DHF
Slower viral clearance in DHF
Similar viral peak time
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 26 / 45
Within-Host Model Secondary Infection
Secondary DF vs DHF
Higher viral peak in DHF
Slower viral clearance in DHF
Similar viral peak time
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 26 / 45
Within-Host Model Secondary Infection
Secondary DF vs DHF
Higher viral peak in DHF
Slower viral clearance in DHF
Similar viral peak time
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 26 / 45
Within-Host Model Secondary Infection
Data Fitting
Assume only neutralizing antibodies play a role in enhancingsecondary infection
Implies higher β2 in more severe infection since more Fc-receptorbearing cells will be infected
So,γE = 0, p2 = p1, γ1 = γ2, and β2 = (1 + ξKA)β1However, from fitting the data we get that βDHF
2 < βDF2 < β1.
We also get that the peak of DF is higher than DHF, which is inconsistentwith the dynamics of secondary DF and DHF.This implies there is not higher infectivity rate β2 in more severe infection.So, we set β2 = β1.
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 27 / 45
Within-Host Model Secondary Infection
Data Fitting
Assume only neutralizing antibodies play a role in enhancingsecondary infection
Implies higher β2 in more severe infection since more Fc-receptorbearing cells will be infected
So,γE = 0, p2 = p1, γ1 = γ2, and β2 = (1 + ξKA)β1However, from fitting the data we get that βDHF
2 < βDF2 < β1.
We also get that the peak of DF is higher than DHF, which is inconsistentwith the dynamics of secondary DF and DHF.This implies there is not higher infectivity rate β2 in more severe infection.So, we set β2 = β1.
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 27 / 45
Within-Host Model Secondary Infection
Data Fitting
Now we fit p2 > p1, γ2 and γE , which correspond to the effectnon-neutralizing antibodies have on secondary infection.We get results more consistent with known viral dynamics of secondaryinfection.
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 28 / 45
Within-Host Model Secondary Infection
Secondary DF vs DHF
0 2 4 6 8 10
1000
104
105
106
107
108
109
Day After Infection
Vir
alL
oad
Secondary DF
0 2 4 6 8 10
1000
104
105
106
107
108
109
Day After Infection
Vir
alL
oad
Secondary DHF
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 29 / 45
Within-Host Model Secondary Infection
Primary vs Secondary Infections
Higher viral peak in secondary infections
Earlier viral peak time in secondary infections
Faster viral clearance time in secondary infections
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 30 / 45
Within-Host Model Secondary Infection
Primary vs Secondary Infections
Higher viral peak in secondary infections
Earlier viral peak time in secondary infections
Faster viral clearance time in secondary infections
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 30 / 45
Within-Host Model Secondary Infection
Primary vs Secondary Infections
Higher viral peak in secondary infections
Earlier viral peak time in secondary infections
Faster viral clearance time in secondary infections
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 30 / 45
Within-Host Model Secondary Infection
Primary vs Secondary DF
0 5 10 500 505 510
10 3
106
109
Time After Primary Infection
Vir
alL
oad
Primary vs Secondary DF
Table : Comparison of Primary DF and Secondary DF Viral Dynamics
Disease Level Viral Peak (RNA/ml) Viral Peak Time (days) Clearance Time (days)
Primary 7.7 × 107 3.9 8.4DF 5.8 × 108 1.8 6.1
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 31 / 45
Within-Host Model Secondary Infection
Primary vs Secondary DHF
0 5 10 500 505 510
10 3
106
109
Time After Primary Infection
Vir
alL
oad
Primary vs Secondary DHF
Table : Comparison of Primary DF and Secondary DHF Viral Dynamics
Disease Level Viral Peak (RNA/ml) Viral Peak Time (days) Clearance Time (days)
Primary 7.7 × 107 3.9 8.4DHF 9.6 × 108 1.7 8.3
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 32 / 45
Within-Host Model Discussion
1 IntroductionBiological Background
2 Within-Host ModelPrimary InfectionSecondary InfectionDiscussion
3 Epidemiological Model
4 Future Work
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 33 / 45
Within-Host Model Discussion
Model Dynamics
Model of primary infection consistent with known dynamics
Model of secondary infection can distinguish between secondary DFand secondary DHF
Model of secondary infection can distinguish between primary andsecondary infection
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 34 / 45
Within-Host Model Discussion
Model Dynamics
Model of primary infection consistent with known dynamics
Model of secondary infection can distinguish between secondary DFand secondary DHF
Model of secondary infection can distinguish between primary andsecondary infection
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 34 / 45
Within-Host Model Discussion
Model Dynamics
Model of primary infection consistent with known dynamics
Model of secondary infection can distinguish between secondary DFand secondary DHF
Model of secondary infection can distinguish between primary andsecondary infection
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 34 / 45
Within-Host Model Discussion
Biological Explanation
Neutralizing antibodies do not increase infectivity rate as described byADE
Clearance rate of virus by non-neutralizing antibodies is decreased insecondary infection
Non-neutralizing antibodies from primary infection bind to virus fromsecondary infection, thus not allowing the virus to be cleared byantibodies specific to secondary infection
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 35 / 45
Within-Host Model Discussion
Biological Explanation
Neutralizing antibodies do not increase infectivity rate as described byADE
Clearance rate of virus by non-neutralizing antibodies is decreased insecondary infection
Non-neutralizing antibodies from primary infection bind to virus fromsecondary infection, thus not allowing the virus to be cleared byantibodies specific to secondary infection
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 35 / 45
Within-Host Model Discussion
Biological Explanation
Neutralizing antibodies do not increase infectivity rate as described byADE
Clearance rate of virus by non-neutralizing antibodies is decreased insecondary infection
Non-neutralizing antibodies from primary infection bind to virus fromsecondary infection, thus not allowing the virus to be cleared byantibodies specific to secondary infection
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 35 / 45
Epidemiological Model
1 IntroductionBiological Background
2 Within-Host ModelPrimary InfectionSecondary InfectionDiscussion
3 Epidemiological Model
4 Future Work
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 36 / 45
Epidemiological Model
Model with Two Strains
S
I1p R1 I2s
I2p R2 I1s
R12
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 37 / 45
Epidemiological Model
Model Equations
dS
dt= µ− β1I1pS − β1φ1I1sS − β2I2pS − β2φ2I2sS − µS
dI1pdt
= β1I1pS + β1φ1I1sS − σI1p
dI2pdt
= β2I2pS + β2φ2I2sS − σI2p
dI1sdt
= β1I1pR2 + β1φ1I1sR2 − σI1s
dI2sdt
= β2I2pR1 + β2φ2I2sR1 − σI2s
dR1
dt= σI1p − β2I2pR1 − β2φ2I2sR1 − µR1
dR2
dt= σI2p − β1I1pR2 − β1φ1I1sR2 − µR2
dR12
dt= σI1s + σI2s − µR12
(4)
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 38 / 45
Epidemiological Model
Epidemiological Data 5
5Ferguson, et al. ”The effect of antibody-dependent enhancement on thetransmission dynamics and persistence of multiple-strain pathogens.” PNAS USA, 1999.
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 39 / 45
Epidemiological Model
Model Dynamics
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 40 / 45
Epidemiological Model
Model Dynamics
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 41 / 45
Future Work
1 IntroductionBiological Background
2 Within-Host ModelPrimary InfectionSecondary InfectionDiscussion
3 Epidemiological Model
4 Future Work
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 42 / 45
Future Work
Original Antigenic Sin
Another explanation for results from within-host model may beoriginal antigenic sin
Effector cells specific to primary infection unable to clear virus easily
Determine whether OAS model can explain data
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 43 / 45
Future Work
Original Antigenic Sin
Another explanation for results from within-host model may beoriginal antigenic sin
Effector cells specific to primary infection unable to clear virus easily
Determine whether OAS model can explain data
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 43 / 45
Future Work
Original Antigenic Sin
Another explanation for results from within-host model may beoriginal antigenic sin
Effector cells specific to primary infection unable to clear virus easily
Determine whether OAS model can explain data
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 43 / 45
Future Work
Multi-scale Modeling
Transmission of virus is dependent on severity of infection (β vs βφ)
Determine effect of enhancement from within-host model
Use these results in epidemiological model
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 44 / 45
Future Work
Multi-scale Modeling
Transmission of virus is dependent on severity of infection (β vs βφ)
Determine effect of enhancement from within-host model
Use these results in epidemiological model
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 44 / 45
Future Work
Multi-scale Modeling
Transmission of virus is dependent on severity of infection (β vs βφ)
Determine effect of enhancement from within-host model
Use these results in epidemiological model
Ryan Nikin-Beers Mathematical Modeling of DVI October 1, 2014 44 / 45