Escola Internacional de Doutoramento
Fa¨ı¸calNda¨ırou
TESE DE DOUTORAMENTO
Modelado matem´atico de certas enfermidades relacionadas coa auga Mathematical modelling of some diseases related to water
Dirixida polos doutores: Iv´anCarlos Area Carracedo (Universidade de Vigo) e Delfim Fernando Marado Torres (Universidade de Aveiro, Portugal)
Ano: 2020
Escola Internacional de Doutoramento
Iv´anCarlos Area Carracedo e Delfim Fernando Marado Torres
FAN CONSTAR que o presente traballo titulado “Modelado matem´atico de certas enfermidades relacionadas coa auga” “Mathematical modelling of some diseases related to water”, que presenta Fa¨ı¸calNda¨ırou para a obtenci´ondo t´ıtulo de Doutor/a, foi elaborado baixo a s´ua direcci´onno pro- grama de doutoramento Programa de Doutoramento en Auga, Sustentabilidade e Desenvolvemento (O03D040P06).
Ourense, 17 de setembro de 2020.
Os Directores da tese de doutoramento
Dr. Iv´anCarlos Area Carracedo Dr. Delfim Fernando Marado Torres
Summary
This thesis dissertation focusses on the study of some infectious diseases dynamics from a double point of view: modelization and control. Our main aim is to formulate new mathematical models and combining them with existing ones in order to analyze the dynamics of diseases related to water. We consider compartmental models described by ordinary di↵erential equations and perform rigorous qualitative and quantitative techniques for acquiring insights into the dynamics of these models. My contribution to the material in this thesis is contained in the following original papers: P1) F. Nda¨ırou, I. Area and D. F. M. Torres. Mathematical Modeling of Japanese Encephalitis Under Aquatic Environmental E↵ects.Submitted; P2) F. Nda¨ırou, I. Area, J. J. Nieto, C. J. Silva, and D. F. M. Torres. Mathematical modeling of Zika disease in pregnant women and newborns with microcephaly in Brazil. Mathematical Methods in the Applied Sciences, 41(18):8929–8941, 2018; P3) F. Nda¨ırou, I. Area, J. J. Nieto, and D. F. M. Torres. Mathematical modeling of COVID-19 transmission dynamics with a case study of Wuhan. Chaos, Solitons & Fractals, 135:109846, 2020; P4) I. Area, F. Nda¨ırou, J. J. Nieto, C. J. Silva, and D. F. M. Torres. Ebola model and opti- mal control with vaccination constraints. Journal of Industrial & Management Optimization, 14(2):427–446, 2018; done in collaboration with my advisors (Professor Iv´anArea and Professor Delfim F. M. Torres) and co-authored with my collaborator and former advisor Professor Juan Jos´eNieto and my collaborator Doctor Cristiana Jo˜aoda Silva. This thesis discusses some recent knowledge and investigation on the transmission dynamics of Ebola disease, Zika disease, Japanese encephalitis disease as well as COVID-19. The following are the main topics: (1) The Ebola virus disease is a severe viral haemorrhagic fever syndrome caused by Ebola virus. This disease is transmitted by direct contact with the body fluids of an infected person and objects contaminated with virus or infected animals, with a death rate close to 90% in humans. Recently, some mathematical models have been presented to analyse the spread of the 2014 Ebola outbreak in West Africa. For this disease, we introduce vaccination of the susceptible population with the aim of controlling the spread of the disease and analyze two optimal control problems related with the transmission of Ebola disease with vaccination. Firstly, we consider the case where the total number of available vaccines in a fixed period of time is limited. Secondly, we analyze the situation where there is a limited supply of vaccines at each instant of time for a fixed interval of time. The optimal control problems have been solved analytically. Finally, we have performed a number of numerical simulations in order to compare the models with vaccination and the model without vaccination, which has recently been shown to fit the real data. Three vaccination scenarios have been considered for our numerical simulations, namely: unlimited supply of vaccines; limited total number of vaccines; and limited supply of vaccines at each instant of time. (2) We propose a compartmental mathematical model for the spread of the COVID-19 disease with special focus on the transmissibility of super-spreaders individuals. We compute the basic reproduction number threshold, we study the local stability of the disease free equilibrium in terms of the basic reproduction number, and we investigate the sensitivity of the model with respect to the variation of each one of its parameters. Numerical simulations show the suitability of the proposed COVID-19 model for the outbreak that occurred in Wuhan, China. (3) We propose a new mathematical model for the spread of Zika virus. Special attention is paid to the transmission of microcephaly. Numerical simulations show the accuracy of the model with respect to the Zika outbreak occurred in Brazil. (4) Also, we propose a mathematical model for the spread of Japanese encephalitis, with emphasis on environmental e↵ects on the aquatic phase of mosquitoes. The model is shown to be biolog- ically well-posed and to have a biologically and ecologically meaningful disease free equilibrium point. Local stability is analyzed in terms of the basic reproduction number and numerical simulations presented and discussed. Resumo
A presente tese c´entrase no estudo dalgunhas din´amicas de enfermidades infecciosas desde un do- bre punto de vista: modelizaci´one control. O noso principal obxectivo ´eformular novos modelos matem´aticos e combinalos cos existentes para analizar a din´amica das enfermidades relacionadas coa auga. Consideramos modelos compartimentais descritos por ecuaci´ons diferenciais ordinarias e reali- zamos t´ecnicas cualitativas e cuantitativas rigorosas para adquirir informaci´onsobre a din´amica destes modelos. As mi˜nas achegas ao material da tese est´arecollida nos seguintes artigos: P1) F. Nda¨ırou, I. Area and D. F. M. Torres. Mathematical Modeling of Japanese Encephalitis Under Aquatic Environmental E↵ects. Enviado a publicaci´on; P2) F. Nda¨ırou, I. Area, J. J. Nieto, C. J. Silva, and D. F. M. Torres. Mathematical modeling of Zika disease in pregnant women and newborns with microcephaly in Brazil. Mathematical Methods in the Applied Sciences, 41(18):8929–8941, 2018; P3) F. Nda¨ırou, I. Area, J. J. Nieto, and D. F. M. Torres. Mathematical modeling of COVID-19 transmission dynamics with a case study of Wuhan. Chaos, Solitons & Fractals, 135:109846, 2020; P4) I. Area, F. Nda¨ırou, J. J. Nieto, C. J. Silva, and D. F. M. Torres. Ebola model and opti- mal control with vaccination constraints. Journal of Industrial & Management Optimization, 14(2):427–446, 2018; realizados en colaboraci´oncos meus directores de tese (Profesor Iv´anArea e Profesor Delfim F. M. Torres) nos que tam´enparticipou o co-director do meu traballo fin de mestrado, Profesor Juan Jos´e Nieto, e a Doutora Cristiana Jo˜aoda Silva. Esta tese trata de alg´uns co˜necementos e investigaci´ons recentes sobre a din´amica de transmisi´on da enfermidade do ´ebola, a enfermidade de Zika, a enfermidade da encefalite xaponesa e a COVID-19. Os temas principais son os seguintes: (1) En primeiro lugar, a enfermidade do virus do ´ebola ´eunha s´ındrome de febre hemorr´axica viral grave causada polo virus do ´ebola. Esta enfermidade transm´ıtese por contacto directo cos flu´ıdos corporais dunha persoa infectada e obxectos contaminados co virus ou animais in- fectados, cunha taxa de mortalidade pr´oxima ao 90% en humanos. A orixe desta enfermidade non est´aclara. Peter Piot descubriu o ´ebola en 1976 e axudou a conter a primeira crise nese mesmo ano, concretamente en Sud´an(territorio que actualmente est´adividido en dous esta- dos) entre os meses de xu˜no e novembro. Desde ent´onhoubo polo menos 18 crises confirmadas de ´ebola entre 1976 e 2014. Recentemente, present´aronse alg´uns modelos matem´aticos para analizar a propagaci´ondo brote de ´ebola na Africa´ occidental de 2014. Tr´atase dos primeiros pasos para unha seguinte fase do problema, ´edicir, cos modelos existentes na literatura era posible predicir un avance da pandemia no futuro. Posto que non est´aclara a orixe da enfer- midade (se ben boa parte da comunidade cient´ıfica apunta a que poden ser uns determinados morcegos) non se pode predicir cando existir´a,nin onde, un novo surto. Por esta raz´ono problema seguinte ´epensar que facer no caso de ter un surto e como administrar a vacina, no caso de existir. Neste sentido, conv´ensinalar que desde decembro de 2019 existe unha vacina para a enfermidade provocada polo virus do ´ebola, cunha taxa de efectividade suficientemente alta. Pero os problemas non rematan con ter unha vacina, pois ´epreciso ter cantidade sufi- ciente en doses para a s´ua administraci´one, sobre todo, poder contar con persoal m´edico que poida administrar a vacina. Se supo˜nemos un surto do virus do ´ebola non ´edoado contar con persoal sanitario capacitado e disposto a administrar vacina contra esta enfermidade. Por estes motivos e para esta enfermidade, introducimos a vacinaci´onda poboaci´onsusceptible co obxectivo de controlar a propagaci´onda enfermidade e analizamos dous problemas de control ´optimos relacionados coa transmisi´onda enfermidade do ´ebola con vacinaci´on. En primeiro lugar, consideramos o caso onde o n´umero total de vacinas dispo˜nibles nun per´ıodo de tempo fixo ´elimitado. En segundo lugar, analizamos a situaci´onna que hai unha oferta limitada de vacinas en cada instante por un intervalo de tempo fixo. Resolv´eronse os problemas de control ´optimo analiticamente. Finalmente, realizamos unha serie de simulaci´ons num´ericas en para comparar os modelos coa vacinaci´one o modelo sen vacinaci´on, que recentemente demostrou que se axusta aos datos reais. Consider´aronse tres escenarios de vacinaci´ons para as nosas simulaci´ons num´ericas, nomeadamente: subministraci´onilimitada de vacinas; n´umero total limitado de vacinas; e oferta limitada de vacinas en cada instante de tempo. Deste xeito pretendeuse anticipar un problema real con cuesti´ons tan complexas como a vacinaci´on e unha enfermidade altamente letal. Neste ´ultimo aspecto conv´ensinalar que a taxa de mor- talidade do ´ebola var´ıa entre o 50 e o 90%, e estudos recentes demostran que determinadas diferenzas xen´eticas xogan un papel fundamental na taxa de mortalidade. Esta cuesti´onest´a baixo an´alise noutra das enfermidades analizadas nesta tese de doutoramento, a COVID-19. De feito, durante a pandemia do ´ebola do ano 2014 que afectou fundamentalmente a Guinea, Liberia e Serra Leoa p´odense observar diferenzas important´ısimas nas taxas de mortalidade, que foron a orixe do estudo xen´etico. En calquera caso, son cuesti´ons moi complexas porque as recomendaci´ons sanitarias durante a pandemia do Ebola´ de 2014 conti˜nan a incineraci´on das persoas falecidas. Esta recomendaci´onest´arecollida no modelo compartimental que anal- iza a evoluci´ondesta enfermidade ao contemplar persoas susceptibles, expostas, infectadas, recuperadas, hospitalizadas, falecidas, incineradas e completamente recuperadas. (2) Propo˜nemos un modelo matem´atico compartimental para a propagaci´onda enfermidade da COVID-19 con especial ´enfase na transmisibilidade dos individuos super-propagadores. Calcu- lamos o n´umero b´asico de reproduci´on, estudamos a estabilidade local do punto de equilibrio libre da enfermidade en termos do n´umero de reproduci´onb´asica, e investigamos a sensibi- lidade do modelo con respecto ´avariaci´onde cada un dos seus par´ametros. As simulaci´ons num´ericas mostran a idoneidade do modelo proposto para o brote de COVID-19 ocorrido en Wuhan, China. A enfermidade da COVID-19 tivo a s´ua orixe posiblemente en China e tam´en con bastante exactitude a finais do ano 2.019 e comezos do ano 2.020. Desde que a OMS come- zou a publicar datos sobre a existencia desta enfermidade comezamos a analizar os datos para tentar dispor dun modelo matem´atico que permitise explicar o avance da enfermidade. Nesta li˜na houbo distintas reuni´ons de traballo e nos primeiros d´ıas de febreiro de 2020 xa ti˜namos o modelo presentado nesta tese de doutoramento. A principal novidade foi a de incorporar unha nova clase de persoas, as denominadas superprogagadoras, que permit´ıa explicar con bastante exactitude os datos que chegaban da China. En comparaci´oncon outras enfermidades m´ais co˜necidas, foi preciso estimar alg´uns dos par´ametros que determinan a taxa de variaci´onde cada un dos compartimentos nos que se dividiu a poboaci´on. Con este modelo, e cando o d´ıa 4 de marzo houbo o primeiro caso en Galicia, elaborouse un informe que foi remitido a todos os partidos pol´ıticos con representaci´onno Parlamento de Galicia advertindo do avance da pandemia. Segundo as simulaci´ons realizadas, o maior n´umero de persoas novas infectadas ter´ıa lugar o d´ıa 5 de abril, coincidindo coa data inicialmente prevista para as elecci´ons galegas. Esta conxectura foi comprobada como certa posteriormente e as elecci´ons foron adiadas para o domingo d´ıa 12 de xullo. Os traballos realizados foron fundamentais para a predici´ondos n´umeros de persoas infectadas, hospitalizadas e nas unidades de coidados intensivos a nivel galego. Tr´atase de problemas de alt´ısima complexidade pois a experiencia indica que unha predici´oncomo a feita no seu d´ıa ´emoi complexa de realizar con predici´ons aca´ıdas. M´ais a´ında, a predici´onde n´umero de persoas hospitalizadas e nas unidades de coidados intensivos nos hospitais require unha predici´onde casos como a antes descrita para, en base a esa onda, extraer as posibles persoas infectadas para predicir as hospitalizaci´ons. Ter predici´ons cun nivel aceptable de fiabilidade pode permitir pensar en decisi´ons tan complexas como: ´enecesario un novo hospital? ´enecesario cancelar as operaci´ons non vitais en toda Galicia? ´epreciso contar con estudantes de ´ultimos anos para tarefas que deberan ser realizadas por profesion- ais nos hospitais? A soluci´onadoptada para frear a pandemia, en base a confinamentos, ten producido numerosas consecuencias a todos os niveis, inclu´ındo os sanitarios, econ´omicos e sociais, sen deixar de mencionar o educativo. Posiblemente esta pandemia puxo enriba da mesa a necesidade de contar con persoas traballando en modelos epidemiol´oxicos, campo de traballo que conta con numeroso persoal investigador e recursos noutros continentes, con es- pecialistas de primeiro nivel mundial en Africa.´ Tr´atase de traballos interdisciplinares, que deben contar con matem´aticos e, por suposto, persoal sanitario. Durante o tempo de dep´osito da tese de doutoramento estaban baixo ensaio cl´ınico, en distintas fases, distintas vacinas con- tra a enfermidade. Novamente, a existencia de vacina d´alugar a un n´umero importante de problemas matem´aticos que c´ompre resolver desde a teor´ıa de control. Basicamente, tr´atase de maximizar os efectos da vacinaci´onminimizando o n´umero de vacinas aplicadas, pois non haber´avacina para todas as persoas que habitamos actualmente a terra. Se ben se trata de datos que se actualizan diariamente, o n´umero de casos de persoas infectadas pola COVID-19 supera xa os 50 mill´ons de persoas en todo o mundo, con arredor do 2,6% de persoas falecidas. Estas cifras var´ıan dun estado a outro e tam´envar´ıan segundo territorios, o que apunta ´a necesidade de an´alise desde o punto de vista xen´etico para poder saber m´ais sobre por que hai persoas que falecen por esta enfermidade e outras que a superan simplemente cun pouco de febre. As taxas de infecci´ona nivel galego de arredor do 1 por cento da poboaci´onest´an moi por debaixo de p.ex. Chile ou Per´uonde as porcentaxes son superiores ao 3 por cento de persoas infectadas. M´ais a´ında, en Per´uxa faleceu m´ais do un por mil da poboaci´ontotal pola COVID-19, dato que dista moito de Galicia onde as cifras oficiais falan de arredor do tres por dez mil da poboaci´ongalega falecida pola COVID-19. Por suposto, ´ecomplexo comparar cifras pois non hai o mesmo n´umero de probas PCRs realizadas nen tan sequera a nivel do estado espa˜nol, menos a´ında en Europa e en ning´un xeito comparable a nivel mundial, pero as cifras que publica a Organizaci´onMundial da Sa´ude son as que se analizan para as comparativas anteriores. Por outra banda, desde o comezo da pandemia estivo moi presente a denominada “inmunidade de reba˜no”. Esta cuesti´onsemella que non ´eposible sen unha vacina pois xa apareceron os primerios casos de persoas reinfectadas pola COVID-19, demostrados cientifi- camente. M´ais a´ında as distintas mutaci´ons do virus da COVID-19, a ´ultima delas a trav´es dos vis´ons, fai pensar que non ´eunha doenza que permita ese tipo de inmunidade. Os estudos de seroprevalencia tampouco indican que sexa posible atinxir a inmunidade de reba˜no, pois as cifras que se publican son moi baixas. (3) No seguinte cap´ıtulo da tese propo˜nemos un novo modelo matem´atico para a propagaci´ondo virus Zika, prestando especial atenci´on´atransmisi´onda microcefalia. Neste estudo realizamos distintas simulaci´ons num´ericas para amosar a precisi´ondo modelo con respecto ao brote de Zika ocorrido en Brasil. O nome do virus do Zika prov´endo bosque Zika en Uganda, onde o virus foi aillado por primeira vez no ano 1.947. Cinco anos m´ais tarde detect´aronse os primeiros casos en humanos. Desde ent´onesta enfermidade tivo distintos surtos, concentrados en Africa´ e en Asia. O surto do ano 2.015-2.016 tivo unha alt´ısima repercusi´onao afectar aos xogos ol´ımpicos de Rio de Janeiro. Esta enfermidade en humanos transm´ıtese a trav´esda picadura dun mosquito infectado da clase Aedes, tanto o Aedes Aegypti como o Aedes Albopictus. Inicialmente a infecci´onnon ´emoi chamativa ao producir erupci´oncut´anea, febre leve, conx- untivite e dor muscular. Debemos ter presente que estes problemas ocultan un dos grandes problemas, pois o virus do Zika est´arelacionado con determinadas desordes neurol´oxicas e malformaci´ons fetais. En marzo do ano 2.016 un equipo de investigadores apuntou ao virus do Zika como causante de microcefalia en mulleres embarazadas, pois o virus infecta un tipo de c´elulas nai neuronais responsables da cortiza cerebral. Como consecuencia desta infecci´on, as c´elulas nai morren ou perden a capacidade de dividirse, polo que a cortiza cerebral non se forma correctamente ou, chegado o caso, non se rexenera. Esta ´eunha das causas posibles para a mencionada microcefalia e outras enfermidades asociadas a este virus. As consecuencias non rematan aqui e outros especialistas indican que este virus tam´enpode ser causante da s´ındrome de Guillain-Barr´e.S´ofai tres anos que un grupo de investigadores determinou, nun artigo pub- licado na prestixiosa revista Science, a estrutura do virus e o que ´ea´ında m´ais importante: as s´uas diferenzas entre o virus do Zika e outros da clase dos flavivirus, como poden ser o dengue, o virus do Nilo Occidental, a febre amarela, a encefalite xaponesa (obxecto de estudo noutro cap´ıtulo da tese) ou o virus da encefalite transmitida por garrapatas. Desde o punto de vista matem´atico a modelizaci´onde enfermidades transmitidas por mosquitos incrementa a s´ua dificultade. Por unha banda, xa que ´epreciso termos en conta d´uas poboaci´ons (humanos e mosquitos) e, consecuentemente e en segundo lugar, ao incrementar o n´umero de inc´ognitas, o sistema de ecuaci´ons diferenciais ten maiores dificultades para a s´ua an´alise e maior custe computacional para as simulaci´ons num´ericas. C´ompre sinalar que o tipo de mosquito que transmite esta(s) enfermidade(s) est´aa propagarse por territorios onde non existiu nos ´ultimos tempos, de xeito que enfermidades ´asque non se lles prestaba atenci´ondesde determinados grupos humanos comezan agora a preocupar. (4) No ´ultimo cap´ıtulo, propo˜nemos un modelo matem´atico para a propagaci´onda encefalite xaponesa, con ´enfase sobre os efectos ambientais na fase acu´atica dos mosquitos. Probamos que o modelo biol´oxico est´aben plantexado e ten un punto de equilibrio libre de enfermidades biol´oxicas, ecoloxicamente significativo. Analizamos a estabilidade local en termos do n´umero de reproduci´onb´asica. Para conclu´ır, ach´eganse simulaci´ons num´ericas do modelo. Tr´atase dunha enfermidade transmitida por mosquitos, que normalmente habitan en zonas con abun- dante auga. Precisan da auga para a s´ua reproduci´one precisan neste caso de humanos para transmitir a enfermidade. A enfermidade ´etransmitida polo mosquito Culex tritaeniorhynchu. Este tipo de mosquito cr´ıa onde hai auga abundante en zonas agr´ıcolas rurais, como os ar- rozais, e inf´ectanse aliment´andose de hospedadores vertebrados (principalmente porcos e aves lim´ıcolas) infectados polo virus da encefalite xaponesa. Os s´ıntomas da enfermidade duran entre 5 e 15 d´ıas e incl´uen febre, dor de cabez, confusi´one dificultades para moverse. Tr´atase dunha enfermidade que pode causar a morte de persoas. Como xa foi sinalado anteriormente, ´eunha enfermidade transmitida por un virus da clase dos flavivirus, polo que o estudo da s´ua evoluci´onpermite co˜necer m´ais sobre outros virus e enfermidades letais para a especie humana. Na actualidade det´ectanse arredor de 68.000 casos cada ano de persoas infectadas por esta enfermidade, pero o n´umero ten que ser maior e s´ounha parte das persoas infec- tadas son detectadas. Segundo os informes da Organizaci´onMundial da Sa´ude, arredor de 17.000 persoas falecen cada ano por esta enfermidade. Os primeiros casos detectados datan do ano 1.871 e prod´ucense grandes surtos cada 2-15 anos, posiblemente debidos a alg´un tipo de comportamento dos mosquitos que a´ında non se co˜nece. A diferenza das outras enfermidades analizadas, non existen datos de seguimento de surtos con casos diarios, sen´onque os n´umeros publicados son mensuais ou anuais, con grandes dificultades para ter estimaci´ons cl´ınicas dos par´ametros do sistema de ecuaci´ons diferenciais proposto. A diferenza de modelos anteriores publicados na literatura, o modelo matem´atico para explicar a expansi´onda enfermidade incor- pora efectos ambientais na fase acu´atica dos mosquitos, como fonte principal de reproduci´on. O ciclo biol´oxico do mosquito presenta as fases de ovo, larva, pupa e adulto. As fase inmaduras (larva e pupa) son acu´aticas, mentre que a de adulto ´eunha fase de vida a´erea. No mod- elo matem´atico proposto, consideramos os factores ambientais dentro de tres poboaci´ons de hospedaxe diferentes: humanos, mosquitos e animais vertebrados (tale como se sinalou, funda- mentalmente porcos e aves lim´ıcolas). De feito, as condici´ons ambientais non hixi´enicas poden mellorar a presenta e o crecemento de poboaci´ons de vectores (mosquitos) que levan a unha r´apida propagaci´onda enfermidade. Isto d´ebese a varios tipos de residuos dom´esticos e p.ex. vertidos ao medio ambiente en zonas residencias, que proporcionan un ambiente moi propicio para o crecemento dos vectores. A cuesti´onfundamental ´eque este tipo de situaci´onnon se puido modelar con compartimentos epidemiol´oxicos, polo que se recurriu a modelos existentes para tratar ese efecto sobre a enfermidade da encefalite xaponesa. No que respecta ao mod- elado de mosquitos e humanos si que se empregaron compartimentos, baixo certas hip´oteses nomeadamente: non se considerou emigraci´onde persoas humanas infectadas; a poboaci´on humana non ´econstante (taxa de mortalidade inducida por enfermidade); e tam´ense supuxo que o coeficiente de transmisi´ondo virus ´econstante e non var´ıa segundo as estaci´ons; no finalmente, no modelo marcouse que todos os mosquitos nacen como susceptibles de contraer a enfermidade. (5) Finalmente, conv´enter presente o impulso ´aciencia aberta desde distintos organismos e in- stituci´ons, como pode ser a Fundaci´onEspa˜nola para la Ciencia y la Tecnolog´ıa, que aspira a po˜ner en valor o labor das universidades, investigadores/as e emprendadores/as. Tr´atase dun debate moi interesante e intenso onde conv´endistinguir ciencia aberta de publicar en aberto e ter presente a cada vez maior presenza de determinadas revistas e editoriais que non cumpren os est´andares b´asicos de calidade. Consid´erase un novo enfoque do proceso cient´ıfico, xa que a ciencia aberta est´abaseada no traballo colaborativo e na dispo˜nibilidade e intercambio de co˜necemento pola comunidade cient´ıfica, a sociedade e as empresas, coa finalidade de aumentar o seu reco˜necemento e o seu (potencial) impacto social e con´omico. Como sinalaba, a ciencia aberta (e en xeral o co˜necemento aberto) ´ealgo m´ais que o acceso aberto a datos e publicaci´ons, e incl´ue a apertura do proceso cient´ıfico a todos, reforzando o concepto de responsabilidade social e cient´ıfica. Neste sentido, como ap´endices da tese est´anpublicados os c´odigos que per- miten facer as simulaci´ons num´ericas dos distintos modelos propostos cun c´odigo libre, tanto para a COVID como para a encefalite xaponesa. Todos os artigos aceptados para publicaci´on foron publicados no repositorio arxiv.org
Agradecementos – Acknowledgements
I express my gratitude to my parents for their love, prayers and support, I say “may Allah care for you as you cared for me when I was young”. I am also grateful to my entire family especially my sisters and brothers. I would like to express my gratitude to my supervisors, Prof. Iv´anArea and Prof. Delfim F. M. Torres, for their endless support, advices, availability and their work to achieve this essay. I say, may Allah give you long life and good health.
Contents
Chapter 1. Introduction 3 1. A brief review on mathematical models 4 2. A brief review on optimal control measures 5
Chapter 2. Mathematical background 7 1. Mathematical modelling 7 2. Mathematical analysis 9 3. Optimal control 12
Chapter 3. Ebola model and optimal control with vaccination constraints 19 1. Introduction 19 2. Initial mathematical model for Ebola 21 3. Mathematical model for Ebola with vaccination 23 4. Optimal control with an end-point state constraint 24 5. Optimal control with a mixed state control constraint 26 6. Numerical simulations 28 7. Discussion 34
Chapter 4. Mathematical modeling of COVID-19 dynamics with a case study of Wuhan 37 1. Introduction 37 2. The Proposed COVID-19 Compartment Model 37 3. Qualitative Analysis of the Model 38 4. Sensitivity Analysis 41 5. Numerical Simulations: The Case Study of Wuhan 42 6. Discussion 43
Chapter 5. Some models related to mosquitoes, water and environments 45 1. Mathematical modelling of Zika disease 46 2. Modeling of Japanese Encephalitis under Aquatic Environmental E↵ects 58
Appendix A. SageMath code to compute R0 and its sensitivity indexes of Chapter 4 67
Appendix B. Matlab code for the Wuhan case study of Chapter 4 71
Appendix C. Python code for the Japanese encephalitis simulations 75
Appendix. Bibliography 79
1
CHAPTER 1
Introduction
Mathematical or epidemiological modeling can be seen as fundamental techniques that support the understanding of the spread of infectious diseases in populations. It plays an important role in analyzing, predicting, controlling potential outbreaks [9, 10, 32, 105].
In this PhD thesis, we study dynamics of some infectious diseases from a double mathematical point of view: modelization and control. In modelization, we are concerned by formulating new mathematical models and combining them with existing ones in order to analyze the dynamics of diseases, some of them transmitted by mosquitoes, water or environment.
In control theories, we are interested in the study of measures that will help to combat the progression of the diseases. Here, we will emphasize either on the notion of basic reproduction number which quantifies the transmission potential of a disease; or on the notion of optimal control that deals with the problem of finding the best control strategy for a given dynamics of a disease.
It might be interesting to notice that, the su↵ering of people from infectious diseases surpasses the individual problems but goes to collectives, countries and even the whole world. It impacts the social living and thus the economic by keeping children away from school and adults away from work. In fact, nowadays with the novel coronavirus (COVID-19), many countries are in lockdown putting their citizens away from their activities [68, 76]. The dissertation is organized as follows: We begin by chapter 2, mathematical background related to our work. We focus on three points: mathematical modeling, dynamical systems and optimal control theory. Mainly, in order to highlight the novelty of our work and the related mathematical models proposed, we begin by recalling the basic and simple models that are assumed to be the foundations of this field of mathematical epidemiology. Then, we emphasize on the mathematical analysis tools with close view on dynamical systems. We conclude the chapter, by optimal control tools and precisely by stating the Maximum Principle of problems related to our study. In Chapter 3, we firstly consider a mathematical model for the 2014 Ebola outbreak and introduce vaccination of susceptible individuals with the aim of controlling the spread of the disease. Then, analyze, two optimal control problems related to limitation on the total number of available vaccines. This is mathematically known as state and control constraints optimal control problem and represents a public health problem of limited total number of vaccines. Lastly, we present some numerical experiments for the scenarios: unlimited supply of vaccines; limited total number of available vaccines; and limited supply of vaccines at each instant of time. Chapter 4 is devoted to the novel pandemic COVID-19. We propose a compartmental mathematical model for the spread of the COVID-19 disease with special focus on the transmissibility of superspreaders individuals. We compute the basic reproduction number threshold, we study the local stability of the disease free equilibrium in terms of the basic reproduction number, and we investigate the sensitivity of the model with respect to the variation of each one of its parameters. Numerical simulations show the suitability of the proposed COVID-19 model for the outbreak that occurred in Wuhan, China. At the end, we extend the study by considering the reality of Galiza, Spain and Vigo.
3 4 1. INTRODUCTION
We end by chapter 5, in which we emphasize on diseases related to water. Thus, we propose and study two di↵erent models: the first one is about the Zika outbreak occurred in Brazil with a special attention to the transmission of microcephaly to newborn babies. And the second model is on a study of a Japanese encephalitis diseases from multiples source of transmission with a focus on an environmental discharge. For the Zika disease, we propose a model that includes human and mosquito compartments with an aquatic phase of mosquitoes. The novelty of our work is that the human populations is restricted to women individuals with four epidemiological states: susceptible pregnant women; infected pregnant women; women who gave birth to babies without neurological disorder; and women who gave birth to babies with neurological disorder due microcephaly. This restriction on populations appears to be useful to estimate the number of newborn babies with microcephaly as it has been shown in our published paper [92]. Regarding the Japanese encephalitis disease, the proposed model consider environmental factors within three di↵erent host populations: human, mosquito and vertebrate animals (pigs, wading birds) interconnected with the aquatic phase of mosquitoes. Our main contribution is related to considering environmental discharge e↵ect on that aquatic phase of mosquitoes. Our results show that infected populations decrease highlighting the importance of considring environmental e↵ect on the aquatic phase as primarily source of reproduction of mosquitoes. Now we present a brief state of the art on mathematical models and optimal control theories.
1. A brief review on mathematical models Mathematical epidemiology models goes back to the work of Daniel Bernouilli on inoculation against Small pox. In fact, in 1760, he formulated and studied a model for Small pox. Later, at the beginning of twentieth century, mathematical epidemiology based on compartmental models was established, namely by Hamer in 1906. Then this subject developed as follows: on the one hand, by a sequence of three papers by Kermack and McKendrick in late 1920s after a seminal work [59, 60, 61], and on the other hand by a remarkable work of R. Ross in the context of malaria modeling in early 1900s. In the study of compartmental models, the population under study is divided into compartments or classes according to the disease status of each single individual in the population. The simplest and most used model is the so-called SIR (Susceptible class, Infected class, and recovered class) model, an extension is SEIR (Susceptible class, Infected class, Exposed class and recovered class) model. The rates of transfer between compartments are expressed mathematically as derivatives with respect to time of the sizes of the compartments leading to a system of di↵erential equations model. Noticed that modeling infectious diseases dynamics with partial di↵erential equations, specially with age distribution has been anticipated by Lotka in early 1920s. During time many researchers as McKendrick, Norton, Fisher, Leslie, Andrewartha and Birch etc, had considered such techniques of modeling leading to the fundation of this field of studies. Disease transmission process is basically of two types: Direct contact that is between infected individuals and susceptible individuals in the commu- • nity. For example, diseases as Measles, Smallpox, Ebola, etc. Indirect contact that is due to the presence of carriers in the environment. Either through • mosquitoes for vector borne diseases like Zika, Dengue, Chikungunya, Malaria etc, or through water for water-borne diseases like cholera, typhoid, giardiasis, etc, or through air for airborne diseases like Tuberculosis etc. Several mathematical models have been proposed and studied for infectious diseases transmision. For direct contact transmission, we may refer to recent works on 2014 Ebola outbreak [95, 132, 133]. Also, for indirect contact transmission, we may refer to recent works on Zika disease [4, 92, 89]. And very recently a compartmental mathematical model for the spread of the COVID-19 disease with special focus on the transmissibility of super-spreaders individuals has been analyzed [93]. In this thesis, for the 2. A BRIEF REVIEW ON OPTIMAL CONTROL MEASURES 5 diseases related with water we will consider models that involve both human and mosquito populations interconnected by an aquatic phase. Some years ago very few people was considering that mathematical modelling could have so huge interest and relations with economy. Nowadays, it seems that the analysis of infectious diseases, even of not widely spread at this time, could be crucial in future pandemics.
2. A brief review on optimal control measures Optimal control, the theory of mathematics that deals with the problem of finding the best control strategy for a given system, is an extension of the Calculus of Variations, born in 1696 after Johann Bernouilli’s work, proposing solution to the Brachistochrone problem. Furthermore, Newton, Leibniz and L’Hˆopital gave correct solutions to the Brachistochrone problem. However, it is in fact, with the formulation of the celebrated Pontryagin Maximun principle [100] that the theory of Optimal control emerged and gained much interest through use of computer simulations, which are helpful for plotting and analyzing the optimal trajectories. Numerical schemes are indeed valuable tools for testing theories, assessing quantitative conjectures, and we aim through this thesis to investigate numerical method for computing optimal solutions for some given problem and their optimal states. The method of optimal control is to extremize (minimize or maximize) a (cost) functional subject to a state dynamical system and boundaries conditions. In the theory of mathematical models for the spread of infectious disease in populations, when a disease invade a community, it is necessary to look for control measures as well as studying optimal control strategies in order to combat the progression of the disease. there is a threshold quantity denoted universally now by R0, named earlier by Mac Donald [79] as basic reproduction number which quantifies the transmission potential of a disease. If the basic reproduction number falls below one (R0 < 1), i.e. the infective may not pass the infection on during the infectious period, the infection dies out. On the other hand, if R0 > 1 the disease persists in the populations and it is urgent to react by considering control measures. In [11], for the first time on the Ebola literature, they were investigated and analyzed two optimal control problems related with the transmission of the 2014 Ebola outbreak with vaccination constraints. Nowadays, water-borne diseases are of great concern as they are related to water quantity and/or quality issues and could be a↵ected by climate change. Also, everyday there is an estimated of 4500 deaths occurring among children under the age of 14 in the world due to water-borne diseases like Cholera, Typhoid, Giardiasis, Shigella, etc. Recall, the notable work of John Snow since 1855 [119], about the Broad Street Cholera outbreak in London, which suggested that consumption of contaminated water by susceptible individuals is the the main source of Cholera infection, not as was believed at that period through air by fecal-oral transmission. Recent literature about water borne diseases can be found in [81, 88] and references therein. Moreover, optimal control is a beautiful subject of studies, with many applications in epidemic models, see for instance [108, 109, 30, 116].
CHAPTER 2
Mathematical background
In this chapter, we present the mathematical concepts, tools and definitions that are used in this thesis. Mainly, in order to highlight the novelty of our work and the related mathematical models proposed, we begin by recalling the basic and simple models that are assumed to be the foundations of the field of mathematical epidemiology. Then, we emphasize on the mathematical analysis tools with close view on dynamical systems. We conclude the chapter, by optimal control tools and precisely by stating the Maximum Principle of problems related to our study.
1. Mathematical modelling 1.1. The SIR and SEIR models. The classical SIR (Susceptible–Infected–Recovered) and SEIR (Susceptible–Exposed–Infected–Recovered) allow the determination of critical threshold for a disease to occur or persist in a population and during a relatively short period of time. The SIR model is described by the following ODE system:
dS(t) I(t) = S(t), dt N 8 >dI(t) I(t) (1) > = S(t) I(t), > dt N <> dR(t) = I(t), > dt > > where the parameter is the e↵ective: contact rate of infective individuals and susceptible, and stands for recovery rate, and N the constant total population size. The SEIR model extends the basic SIR model in the sense that an extra compartment is introduced, precisely the so-called Exposed compartment E¯. The Exposed compartment E¯ characterize individuals who are infected but not yet capable to transmit the virus to non infected individuals. The ODE system describing the evolution of individuals and transmission of the virus is the following one:
dS¯(t) I¯(t) = S¯(t), dt N 8 ¯ ¯ >dE(t) I(t) ¯ > = S ✏E(t), > dt N (2) > >dI¯(t) <> = ✏E¯(t) I¯(t), dt > >dR¯(t) > = I¯(t). > dt > > Where is the e↵ective contact rate: of infective individuals and susceptible, ✏ > 0 is the rate constant at which exposed individuals become infectious, > 0 the rate constant that infectious individuals become recovered.
7 8 2. MATHEMATICAL BACKGROUND
I(t) In both models the term S(t) represents the force of infection a disease at each instant of N time t. This term known as mass action law had its origin in chemical reaction kinetics and describes the transmission process due to the contacts between susceptible and infected individuals. Moreover, the Law of Mass Action has applicability in many areas of science. In chemistry, it is called Fundamental Law of Chemical Kinetics (the study of rates of chemical reactions), formulated in 1864–1879 by the Norwegian scientists Cato M. Guldberg (1836–1902) and Peter Waage (1833– 1900). The law states that for a homogeneous system, the rate of any simple chemical reaction is proportional to the probability that the reacting molecules will be found together in a small volume (see for instance [128] and references therein). Applied to population processes, if the individuals in population mix homogeneously, the rate of interaction between two di↵erent cohorts of the population is proportional to the product of the numbers in each of the cohorts concerned. If several processes occur simultaneously, then the e↵ects on the numbers in any given cohort from these processes are assumed to be additive. Therefore in case of epidemic modeling, the law is applied to rates of transition of individuals between two interacting compartments of the population (e.g. susceptibles who become infectives after an adequate contact). In brief, the rates of transfer between compartments are expressed mathematically as derivatives with respect to time of the sizes of the compartments leading to a system of di↵erential equations model [21, 52, 53]. However, the complexity in the transmission of a virus due to the existence of several source of infection makes the original SIR model not suitable for complex diseases. Taking this fact into consid- eration, it is clear that for complex diseases, it is necessary to pay more attention on the formulation of the related force of infection. For instance, we have studied in this thesis (see Chapter 3 and Chapter 4), two di↵erent complex models having multiple source of infection: A model for Ebola transmission dynamics consisting of 8 mutually exclusive compartments; • This model consider 4 di↵erent infectives classes as source of infection to formulate the force of infection of ebola. A model for corona disease consisting of 7 mutually exclusive compartments. Its consider 3 • infectives classes as source of infection to formulate the related force of infection. Besides the Law of Mass Action, there exists other form of techniques modeling the transmission dynam- ics of infectious diseases in accordance with the pathogenic agents. Next, we will consider transmission dynamics from two di↵erent species (mosquitoes and human population): mosquito-borne diseases.
1.2. The Ross-MacDonald type SIS model. The Ross-Macdonald model is a basis foundation for a broader theory of mosquito-borne disease transmission and control [65, 79, 83, 111, 117]. This kind of model consider two set of ordinary di↵erential equations: one for human populations and the other for mosquito populations. For example, to highlight both categories of populations, a basic Ross- MacDonald model can be label as ShIhSh-SmIm, where the human populations is referred by using h and the mosquito populations is referred by using m. Thus the model consists of 4 compartments with 2 compartments for humans that is ShIh(susceptible human–infected human), and 2 compartments for mosquitoes SmIm( susceptible mosquito–for infected mosquito). The classical Ross-MacDonald SIS type model is described by the following ordinary di↵erential equations:
dS I h = µ N B m S + r I µ S , dt h h mh N h h h h h 8 m >dIh Im > = B mh Sh (rh + µh)Ih, > dt Nm (3) > >dSm Ih <> = µmNm B hm + µm Sm, dt Nh ⇣ ⌘ >dIm I > = B hm Sm µmIm, > dt N > > ⇣ ⌘ :> 2. MATHEMATICAL ANALYSIS 9 where Nh and Nm denote the total human population and total mosquito population respectively. The rest of the parameters are described in table 1.
Parameter Description
1/µh average lifespan of humans (in days) Proportion of bites that produce infection in human B average number of bites (per day)