The research aims presented in this study form part of a development of decision support systems for Land Engineering, particularly within the field of space knowledge management in risk conditions. The aim of decision-making in this context is characterised by its being part of a complex system composed of different, equally complex sub-systems. Indeed, this complexity is a feature of all systems which include living elements and in any land planning scenario, a human or natural agent is ubiquitous. In order to develop the research study, a problematic sector was chosen; the spread of infection in hospital wards. In particular, it was decided to concentrate specifically on transmission by contact. This topic has proved to be of particular global importance, not only from an economic point of view but more importantly due to its direct risk to human health. The first part of the thesis deals with an investigation of the topic’s principal characteristics as well as the control and prevention measures which are available to experts in the field and which are used to contrast this phenomenon. As statistics demonstrate, current strategies are insufficient in preventing the occurrence of the phenomenon, much less eliminating it altogether, even if there are counter-measures which are relatively efficient in dealing with the problem once it has been officially declared, with all its ensuing ominous consequences. It is thus beneficial to present a research study which provides a more detailed description of the problem, proposing improved intervention strategies and supporting the decision-making required to prevent or control outbreaks of the phenomenon. Although the subject has long been widely researched, above all in the field of epidemiology, the development of new software based on multi-agent paradigms can, in our opinion, play an important contributing role. Thus, it was decided to use a modelling and Event-Based simulation approach, a development of the agent-based system. It proved capable of realistically representing complex human activity and use of space scenarios by integrating the classic bottom up approach with a high-level architecture to plan agent behaviour. The structured knowledge generated during the system problem analysis phase have been attributed to elements of the completed logical model. Moreover, the characteristics that the agents were able to display were fitted with the pre-selected topic and extended. During the development of the model, our aim was to stress the potential of the approach as a means to yielding interesting and detailed considerations. The framework was extended to include other features, for example agent perception of the situation or the influence of environmental conditions on his behaviour. The underlying aim was to increase the descriptive capacity of the framework (and thus the expressive level of the simulation) to achieve a higher degree of complexity, so rendering the model more geared toward the reality of the phenomenon. Compared with other examples in literature, our model was built to overcome a number of conceptual and instrumental limits, above all regarding the spatial spread of contamination. Even if current available data in literature is not detailed enough to allow for an accurate quantification of the weight of interrelations between the variables under study. A further innovative feature is that of interpreting the phenomenon with regard to its relationship with built spaces, both in a physical sense as well as that perceived by agents. It was chosen to analyse this aspect in collaboration with Prof. Yehuda E. Kalay’s research group at the Faculty of Architecture of Technion (Israel). The behaviour of system agents with regard to their conditions and contamination capacity was formulated by the use of a discrete equation. The model and equation were then implemented within a virtual simulation environment, Unity 3D, and the logical functions coded in c#. The simulation of a hypothetical case study and scenarios with initial varied settings helped to verify the efficacy of the modelling and formalisation thanks to the possibility of dynamic visualisation provided by the tool software. It is true that the model requires further calibration, which could also be through data collection to retrieve all the information necessary to feed the system. However, the modelled scenarios and an initial sensitivity analysis provide us with a certain degree of confidence about the validity of the output results. We can thus state that the developed system may be used as decision-making support for this particular modelled phenomenon, thus justifying our choice of agent methodology in the modelling of complex phenomena in the field of urban planning. As research continues, so improvements will be made in the definition of the discrete formula adopted, continuously extending it to reflect real life. A further aim is the application of the system to a real-life case study, providing us with a database in line with the proposed model.
Il percorso di ricerca presentato in questo lavoro si inserisce nell’ambito dello sviluppo dei sistemi di supporto alla decisione nel campo dell’Ingegneria del Territorio, nel caso particolare della gestione della conoscenza spaziale in condizioni di rischio. L’oggetto di decisione in questo ambito è caratterizzato per essere un sistema complesso, composto di svariati sotto sistemi anch’essi complessi. La complessità è infatti una caratteristica propria di tutti i sistemi che comprendono elementi viventi e nella pianificazione territoriale l’agente umano o naturale è ubiquo. Per poter sviluppare la ricerca si è scelto un settore problema che è quello della diffusione delle infezioni nei reparti ospedalieri. In particolare si è scelto di concentrarsi sulla trasmissione tramite contatto. Questo tema è stato dimostrato essere una questione di rilevanza planetaria, da un punto di vista economico e ancor più per il rischio diretto alla salute umana. Nella prima parte della tesi il tema è indagato, sia rispetto alle sue precipue caratteristiche, sia rispetto alle misure di controllo e prevenzione che esperti del settore hanno a disposizione ed attuano per contrastare il fenomeno. Come le statistiche dimostrano le attuali strategie non sono sufficienti a evitare il verificarsi del fenomeno, né tantomeno potrebbero eliminarlo, quanto piuttosto sono delle contromisure più o meno efficaci ed intraprese quando il problema è conclamato, con tutte le nefaste conseguenze. La ricerca di un più raffinata descrizione del problema, così da suggerire migliori strategie di intervento e supportare la presa di decisioni atte a prevenire l’insorgenza del fenomeno o a controllarlo, è auspicabile. Per quanto la questione sia stata ampiamente indagata nei secoli soprattutto nel settore della epidemiologia, lo sviluppo di nuovi strumenti software basati sul paradigma del sistema multi agente può a nostro parere dare un contributo rilevante. Si è quindi scelto di utilizzare un approccio di modellazione e simulazione Event-Based, che è un’evoluzione del sistema ad agenti puro. Esso si è dimostrato capace di rappresentare realisticamente complessi scenari di attività umane e d’uso di spazi tramite l’integrazione del classico approccio bottom-up con una architettura di alto livello per la pianificazione dei comportamenti degli agenti. La conoscenza strutturata che si è generata nella fase di analisi del sistema problema è stata attribuita agli elementi del modello logico realizzato. Le caratteristiche che gli agenti erano in grado di esprimere sono state accordate al tema prescelto ed ampliate. Inoltre nello sviluppo del modello si è voluto stressare le potenzialità dell’approccio scelto al fine di poter introitare interessanti considerazione di dettaglio. Il framework è stato esteso per rappresentare aspetti quali ad esempio la percezione del contesto da parte degli agenti e l’influenza delle condizioni ambientali sui loro comportamenti. Il fine sotteso è stato quello di aumentare le capacità descrittive del l’approccio e di conseguenza l’espressività della simulazione verso una complessificazione che rendesse il modello più calzante alla realtà del fenomeno. Rispetto ai modelli di letteratura quello qui presentato è stato costruito per superarne alcune limitazioni concettuali e strumentali soprattutto riguardanti l’inclusione dell’aspetto di diffusione spaziale della contaminazione, ancorché i dati attualmente disponibili in letteratura non sono sufficientemente fini per poter correttamente quantificare il peso delle interrelazioni fra le variabili considerate. L’interpretazione del fenomeno rispetto alla sua relazione con gli spazi costruiti, sia fisici sia considerati nella comprensione di essi da parte degli agenti, è un aspetto innovativo che si è scelto di analizzare in collaborazione con il gruppo di ricerca del Prof. Yehuda E. Kalay presso la facoltà di architettura del Technion, (IL). Dunque il comportamento degli agenti del sistema rispetto alla loro condizione e capacità di contaminazione è stato formulato attraverso una equazione discreta. Il modello e la equazione sono stati poi implementati in un ambiente di simulazione virtuale, Unity 3D e le funzioni logiche codificate in C#. La simulazione di un caso di studio ipotetico e di scenari iniziali variamente settati è servita a verificare la bontà della modellazione e della formalizzazione, grazie alla possibilità di visualizzazione dinamica che offre lo strumento software. Per quanto il modello necessiti di ulteriori passaggi di calibrazione, anche attraverso una campagna di data-collection mirata a raccogliere in maniera contestuale tutti i dati necessari come input per il sistema, gli scenari modellati e una prima analisi di sensitività ci offrono un certo grado di confidenza sulla bontà degli output. Possiamo quindi affermare che il modello sviluppato può fungere da supporto alle decisioni per lo specifico fenomeno modellato ed ancor più conforta la nostra scelta della metodologia ad agenti per la modellazione di fenomeni complessi nel campo della pianificazione territoriale. Come proseguo della ricerca ci si propone di migliorare la definizione della formulazione discreta adottata, estendendola al continuo così da rispecchiare la realtà ed è inoltre auspicabile l’applicazione del sistema ad un caso di studio reale che ci offra una base dati coerente con il modello proposto.
Event-Based Modelling and Simulation of Hospital Acquired Infection Propagation Dynamics by Contact Transmission in Hospital Wards / Esposito, Dario. - ELETTRONICO. - (2018). [10.60576/poliba/iris/esposito-dario_phd2018]
Event-Based Modelling and Simulation of Hospital Acquired Infection Propagation Dynamics by Contact Transmission in Hospital Wards
Esposito, Dario
2018-01-01
Abstract
The research aims presented in this study form part of a development of decision support systems for Land Engineering, particularly within the field of space knowledge management in risk conditions. The aim of decision-making in this context is characterised by its being part of a complex system composed of different, equally complex sub-systems. Indeed, this complexity is a feature of all systems which include living elements and in any land planning scenario, a human or natural agent is ubiquitous. In order to develop the research study, a problematic sector was chosen; the spread of infection in hospital wards. In particular, it was decided to concentrate specifically on transmission by contact. This topic has proved to be of particular global importance, not only from an economic point of view but more importantly due to its direct risk to human health. The first part of the thesis deals with an investigation of the topic’s principal characteristics as well as the control and prevention measures which are available to experts in the field and which are used to contrast this phenomenon. As statistics demonstrate, current strategies are insufficient in preventing the occurrence of the phenomenon, much less eliminating it altogether, even if there are counter-measures which are relatively efficient in dealing with the problem once it has been officially declared, with all its ensuing ominous consequences. It is thus beneficial to present a research study which provides a more detailed description of the problem, proposing improved intervention strategies and supporting the decision-making required to prevent or control outbreaks of the phenomenon. Although the subject has long been widely researched, above all in the field of epidemiology, the development of new software based on multi-agent paradigms can, in our opinion, play an important contributing role. Thus, it was decided to use a modelling and Event-Based simulation approach, a development of the agent-based system. It proved capable of realistically representing complex human activity and use of space scenarios by integrating the classic bottom up approach with a high-level architecture to plan agent behaviour. The structured knowledge generated during the system problem analysis phase have been attributed to elements of the completed logical model. Moreover, the characteristics that the agents were able to display were fitted with the pre-selected topic and extended. During the development of the model, our aim was to stress the potential of the approach as a means to yielding interesting and detailed considerations. The framework was extended to include other features, for example agent perception of the situation or the influence of environmental conditions on his behaviour. The underlying aim was to increase the descriptive capacity of the framework (and thus the expressive level of the simulation) to achieve a higher degree of complexity, so rendering the model more geared toward the reality of the phenomenon. Compared with other examples in literature, our model was built to overcome a number of conceptual and instrumental limits, above all regarding the spatial spread of contamination. Even if current available data in literature is not detailed enough to allow for an accurate quantification of the weight of interrelations between the variables under study. A further innovative feature is that of interpreting the phenomenon with regard to its relationship with built spaces, both in a physical sense as well as that perceived by agents. It was chosen to analyse this aspect in collaboration with Prof. Yehuda E. Kalay’s research group at the Faculty of Architecture of Technion (Israel). The behaviour of system agents with regard to their conditions and contamination capacity was formulated by the use of a discrete equation. The model and equation were then implemented within a virtual simulation environment, Unity 3D, and the logical functions coded in c#. The simulation of a hypothetical case study and scenarios with initial varied settings helped to verify the efficacy of the modelling and formalisation thanks to the possibility of dynamic visualisation provided by the tool software. It is true that the model requires further calibration, which could also be through data collection to retrieve all the information necessary to feed the system. However, the modelled scenarios and an initial sensitivity analysis provide us with a certain degree of confidence about the validity of the output results. We can thus state that the developed system may be used as decision-making support for this particular modelled phenomenon, thus justifying our choice of agent methodology in the modelling of complex phenomena in the field of urban planning. As research continues, so improvements will be made in the definition of the discrete formula adopted, continuously extending it to reflect real life. A further aim is the application of the system to a real-life case study, providing us with a database in line with the proposed model.File | Dimensione | Formato | |
---|---|---|---|
30 ciclo - ESPOSITO Dario.pdf
Open Access dal 26/01/2019
Tipologia:
Tesi di dottorato
Licenza:
Tutti i diritti riservati
Dimensione
8.98 MB
Formato
Adobe PDF
|
8.98 MB | Adobe PDF | Visualizza/Apri |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.