Literatura científica selecionada sobre o tema "Disease Prediction and Monitoring Modelling"

The last two decades have seen several large-scale epidemics of international impact, including human, animal and plant epidemics. Policy makers face health challenges that require epidemic predictions based on limited information. There is therefore a pressing need to construct models that allow us to frame all available information to predict an emerging outbreak and to control it in a timely manner. The aim of this thesis is to develop an early-warning modelling approach that can predict emerging disease outbreaks. Based on Bayesian techniques ideally suited to combine information from different sources into a single modelling and estimation framework, I developed a suite of approaches to epidemiological data that can deal with data from different sources and of varying quality. The SEIR model, particle filter algorithm and a number of influenza-related datasets were utilised to examine various models and methodologies to predict influenza outbreaks. The data included a combination of consultations and diagnosed influenza-like illness (ILI) cases for five influenza seasons. I showed that for the pandemic season, different proxies lead to similar behaviour of the effective reproduction number. For influenza datasets, there exists a strong relationship between consultations and diagnosed datasets, especially when considering time-dependent models. Individual parameters for different influenza seasons provided similar values, thereby offering an opportunity to utilise such information in future outbreaks. Moreover, my findings showed that when the temperature drops below 14°C, this triggers the first substantial rise in the number of ILI cases, highlighting that temperature data is an important signal to trigger the start of the influenza epidemic. Further probing was carried out among Maltese citizens and estimates on the under-reporting rate of the seasonal influenza were established. Based on these findings, a new epidemiological model and framework were developed, providing accurate real-time forecasts with a clear early warning signal to the influenza outbreak. This research utilised a combination of novel data sources to predict influenza outbreaks. Such information is beneficial for health authorities to plan health strategies and control epidemics.

A challenge for the clinical management of Parkinson's disease (PD) is the large within- and between-patient variability in symptom profiles as well as the emergence of motor complications which represent a significant source of disability in patients. This thesis deals with the development and evaluation of methods and systems for supporting the management of PD by using repeated measures, consisting of subjective assessments of symptoms and objective assessments of motor function through fine motor tests (spirography and tapping), collected by means of a telemetry touch screen device. One aim of the thesis was to develop methods for objective quantification and analysis of the severity of motor impairments being represented in spiral drawings and tapping results. This was accomplished by first quantifying the digitized movement data with time series analysis and then using them in data-driven modelling for automating the process of assessment of symptom severity. The objective measures were then analysed with respect to subjective assessments of motor conditions. Another aim was to develop a method for providing comparable information content as clinical rating scales by combining subjective and objective measures into composite scores, using time series analysis and data-driven methods. The scores represent six symptom dimensions and an overall test score for reflecting the global health condition of the patient. In addition, the thesis presents the development of a web-based system for providing a visual representation of symptoms over time allowing clinicians to remotely monitor the symptom profiles of their patients. The quality of the methods was assessed by reporting different metrics of validity, reliability and sensitivity to treatment interventions and natural PD progression over time. Results from two studies demonstrated that the methods developed for the fine motor tests had good metrics indicating that they are appropriate to quantitatively and objectively assess the severity of motor impairments of PD patients. The fine motor tests captured different symptoms; spiral drawing impairment and tapping accuracy related to dyskinesias (involuntary movements) whereas tapping speed related to bradykinesia (slowness of movements). A longitudinal data analysis indicated that the six symptom dimensions and the overall test score contained important elements of information of the clinical scales and can be used to measure effects of PD treatment interventions and disease progression. A usability evaluation of the web-based system showed that the information presented in the system was comparable to qualitative clinical observations and the system was recognized as a tool that will assist in the management of patients.

The advent of recent high throughput sequencing technologies resulted in an unexplored big data of genomics and transcriptomics that might help to answer various research questions in Parkinson’s disease(PD) progression. While the literature has revealed various predictive models that use longitudinal clinical data for disease progression, there is no predictive model based on RNA-Sequence data of PD patients. This study investigates how to predict the PD Progression for a patient’s next medical visit by capturing longitudinal temporal patterns in the RNA-Seq data. Data provided by Parkinson Progression Marker Initiative (PPMI) includes 423 PD patients with a variable number of visits for a period of 4 years. We propose a predictive model based on a Recurrent Neural Network (RNN) with dense connections. The results show that the proposed architecture is able to predict PD progression from high dimensional RNA-seq data with a Root Mean Square Error (RMSE) of 6.0 and rank-order correlation of (r=0.83, p<0.0001) between the predicted and actual disease status of PD. We show empirical evidence that the addition of dense connections and batch normalization into RNN layers boosts its training and generalization capability.

Parkinson’s disease is a chronic neurodegenerative disorder that affects 1-2% of the world’s population over the age of 65. Current treatments that reduce the severity of symptoms cause numerous side-effects and lose efficacy over the course of disease progression. Leucine-rich repeat kinase 2 (LRRK2) is a novel drug target for the development of disease modifying therapeutics for Parkinson’s disease. LRRK2 mutants have elevated kinase activity and, as such, chemical inhibitors have therapeutic potential. The physiological benefits that arise from chemically inhibiting LRRK2 have been proven through the use of generic kinase inhibitors and more recently the selective benzodiazepinone compound LRRK2IN1. LRRK2IN1 is a highly potent inhibitor, exhibiting a half-maximal inhibitory concentration (IC50) of 9 nM in cellular assays. However, LRRK2IN1 is not biologically available in the brain because it has poor physicochemical and pharmacokinetic properties. In previous research we rationally designed a LRRK2IN1 analogue (IN1_G) that was predicted to have improved metabolic stability and blood-brain barrier permeability. Preliminary biological analysis indicated that both LRRK2IN1 and IN1_G potently inhibited LRRK2-associated neuro-inflammation in vitro. However, the high molecular weight, topological polar surface area and lipophilicity of LRRK2IN1 and IN1_G were predicted to be incompatible with functional activity in vivo. Structural modifications were thus required to optimise compounds as neuro-protective treatments for Parkinson’s disease. Biological evaluation of the structural components of LRRK2IN1 and IN1_G indicated that the aniline-bipiperidine 1 motif was a moderately potent inhibitor of neuro-inflammation, whilst the tricyclic diazepinone motif IN1_H had no anti-inflammatory efficacy. In the current research a series of truncated LRRK2IN1/IN1_G analogues were rationally designed to determine if the diazepinone motif could be replaced with low molecular weight bioisosteres without affecting functional potency. In silico property predictions and scoring functions were used to guide the design of truncated analogues. The Schrödinger suite programs LigPrep, QikProp and Marvin were used to predict the physicochemical and pharmacokinetic properties of analogues. The recently described central nervous system multi-parameter optimisation score was used to select analogues that were likely to possess favourable pharmacokinetic and safety profiles. Analogues were docked in a homology model of the LRRK2 kinase domain that was developed in our previous research. Analogues that conformed to the binding mode of known kinase inhibitors and were predicted by GLIDE to bind to the LRRK2 homology model with high affinity were prioritised for synthesis. Twenty analogues were synthesised using methods known in the literature. The substrate scope of Buchwald-Hartwig chemistry was explored. Novel “all-water” chemistry was employed to synthesise N-benzyl aniline analogues. Methodology recently developed in our group was used to synthesise diazepine and oxazepine analogues of IN1_H. Analogues were assessed for anti-inflammatory efficacy in two cell-based assays. Four truncated analogues — 25, 30, 31 and 39 — had equivalent functional efficacy to LRRK2IN1/IN1_G, inhibiting the secretion of pro-inflammatory cytokines from stimulated primary human microglia by more than 43% at concentrations of 1 µM. These analogues were all predicted to have improved pharmacokinetic properties compared to LRRK2IN1/IN1_G and are excellent candidates for further development. The synthetic intermediate 63 was found to be highly potent (57% inhibition of cytokine secretion at 1 µM), which has suggested options for the development of future analogues. The potency of analogues 25, 30, 31 and 39 indicated that the tricyclic diazepinone motif was not essential for anti-inflammatory efficacy. Analogues from this research have been used to identify a role for LRRK2 in the pathology of severe brain cancer glioblastoma. Although their mechanisms of action have not yet been determined, it is clear that analogues developed in this research have potential applications in the treatment of numerous disorders driven by an inflammatory microenvironment. Further optimisation of the analogues developed in this research will provide the first disease-modifying therapeutics for Parkinson’s disease.

Cardiovascular disease is a large problem in the intensive care unit (ICU) due to its high prevalence in modern society. In the ICU, intensive monitoring is required to help diagnose cardiac and circulatory dysfunction. However, complex interactions between the patient, disease, and treatment can hide the underlying disorder. As a result, clinical staff must often rely on their skill, intuition, and experience to choose therapy, increasing variability in care and patient outcome. To simplify this clinical scenario, model-based methods have been created to track subject-specific disease and treatment dependent changes in patient condition, using only clinically available measurements. The approach has been tested in two pig studies on acute pulmonary embolism and septic shock and in a human study on surgical recovery from mitral valve replacement. The model-based method was able to track known pathophysiological changes in the subjects and identified key determinants of cardiovascular health such as cardiac preload, afterload, and contractility. These metrics, which can be otherwise difficult to determine clinically, can be used to help provide targets for goal-directed therapies to help provide deliver the optimal level of therapy to the patient. Hence, this model-based approach provides a feasible and potentially practical means of improving patient care in the ICU.

Thesis (MSc (Computer Science))-- University of Stellenbosch, 2001.<br>ENGLISH ABSTRACT: This thesis explores the use of neural networks for predicting difficult, real-world time series. We first establish and demonstrate methods for characterising, modelling and predicting well-known systems. The real-world system we explore is seismic event data obtained from a South African gold mine. We show that this data is chaotic. After preprocessing the raw data, we show that neural networks are able to predict seismic activity reasonably well.<br>AFRIKAANSE OPSOMMING: Hierdie tesis ondersoek die gebruik van neurale netwerke om komplekse, werklik bestaande tydreekse te voorspel. Ter aanvang noem en demonstreer ons metodes vir die karakterisering, modelering en voorspelling van bekende stelsels. Ons gaan dan voort en ondersoek seismiese gebeurlikheidsdata afkomstig van ’n Suid-Afrikaanse goudmyn. Ons wys dat die data chaoties van aard is. Nadat ons die rou data verwerk, wys ons dat neurale netwerke die tydreekse redelik goed kan voorspel.<br>Integrated Seismic Systems International

The supply of electricity is important in modern society, so the outages of the electric grid should be few and short, especially for the transmission grid. A summary of the history of the Swedish electrical system is presented. The objective is to be able to plan the maintenance better by following the condition of the equipment. The risk matrix can be used to choose which component to be maintained. The risk matrix is improved by adding a dimension, the uncertainty of the probability. The risk can be reduced along any dimension: better measurements, preventive maintenance or more redundancy. The number of dimensions can be reduced to two by following iso-risk lines calculated for the beta distribution. This thesis lists twenty surveys about circuit breakers and disconnectors, with statistics about the failures and the lifetime. It also presents about forty condition-measuring methods for circuit breakers and disconnectors, mostly applicable to the electric contacts and the mechanical parts. A method for scheduling thermography based on analysis of variance of the current is tried. Its aim is to reduce the uncertainty of thermography and it is able to explain two thirds of the variation using the time of the day, the day of the week and the week number as explanatory variables. However, the main problem remains as the current is in general too low. A system with IR sensors has been installed at the nine contacts of six disconnectors with the purpose of avoiding outages for maintenance if the contacts are in a good condition. The measured temperatures are sent by radio and regressed against the square of the current, the best exponent found. The coefficient of determination $R^2$ is high, greater than 0.9. The higher the regression coefficient is, the more heat is produced at the contact. So this ranks the different contacts. Finally a framework for lifetime modelling and condition measuring is presented. Lifetime modelling consists in associating a distribution of time to failure with each subpopulation. Condition measuring means measuring a parameter and estimating its value in the future. If it exceeds a threshold, maintenance should be carried out. The effect of maintenance of the contacts is shown for four disconnectors. An extension of the risk matrix with uncertainty, a survey of statistics and condition monitoring methods, a system with IR sensors at contacts, a thermography scheduling method and a framework for lifetime modelling and condition measuring are presented. They can improve the planning of outages for maintenance. Finally a framework for lifetime modelling and condition measuring is presented. Lifetime modelling consists in associating a distribution of time to failure with each subpopulation. Condition measuring means measuring a parameter and estimating its value in the future. If it exceeds a threshold, maintenance should be carried out. The effect of maintenance of the contacts is shown for four disconnectors. An extension of the risk matrix with uncertainty, a survey of statistics and condition monitoring methods, a system with IR sensors at contacts, a thermography scheduling method and a framework for lifetime modelling and condition measuring are presented. They can improve the planning of outages for maintenance.<br>Elförsörjningen är viktig i det moderna samhället, så avbrotten bör vara få och korta, särskilt i stamnätet. En kortfattad historik över det svenska elsystemet presenteras. Målet är att kunna planera avbrotten för underhåll bättre genom att veta mera om apparaternas skick. Det är svårt att planera avbrott för underhåll och utbyggnad. Riskmatrisen är verktyg för att välja vad som ska underhållas och den kan förbättras genom att lägga till en dimension, sannolikhetens osäkerhet. Risken kan minskas längs med varje dimension: bättre mätningar, förebyggande underhåll och mer redundans. Antalet dimensioner kan igen bli två genom att följa linjer med samma risk, som är beräknade för betafördelningen. Denna avhandling tar upp tjugo studier av fel i brytare och frånskiljare med data om felorsak och livslängd. Den har också en översikt av ett fyrtiotal olika metoder för tillståndsmätningar för brytare och frånskiljare, som huvudsakligen rör de elektriska kontakterna och de mekaniska delarna. Ett system med IR sensorer har installerats på de nio kontakterna på sex frånskiljare. Målet är att minska antalet avbrott för underhåll genom att skatta skicket när frånskiljarna är i drift. De uppmätta temperaturerna tas emot genom radio och behandlas genom regression mot kvadraten av strömmen, då den bästa exponenten för strömmen visade sig vara 2,0. Förklaringsfaktorn $R^2$ är hög, över 0,9. För varje kontakt ger det en regressionskoefficient. Ju högre koefficienten är, desto mer värme utvecklas det i kontakten, vilket kan leda till skador på materialet. Koefficienterna ger en rangordning av frånskiljarna. Systemet kan också användas för att minska eller öka den tillåtna strömmen baserat på skicket. Slutligen förklaras ett ramverk för livslängdsmodellering och tillståndsmätning. Livslängdsmodellering innebär att koppla en fördelning för tiden till fel med varje delpopulation. Med tillståndsmätning avses att mäta en parameter och skatta dess värde i framtiden. Om den överskrider en tröskel, måste apparaten underhållas. Effekten av underhåll visas för fyra frånskiljare. En utveckling av riskmatrisen med osäkerheten, en sammanställning av statistik och metoder för tillståndsövervakning, ett system med IR-sensor vid kontakerna, en metod för termografiplanering och ett ramverk för livslängdsmodellering och tillståndsmätningar presenteras. De kan förbättra avbrottsplaneringen.<br>El suministro de energía eléctrica es importante en la sociedad moderna. Por eso los cortes eléctricos deben ser poco frecuentes y de poca duración, sobre todo en la red de transmisión. Esta tesis resume la historia del sistema eléctrico sueco. El objetivo es planificar los cortes mejor siguiendo la condición de los aparatos. La matriz de riesgo se utiliza muchas veces para escoger en qué aparatos debería realizarse mantenimiento. Esta matriz se puede mejorar añadiendo una dimensión: la incertidumbre de la probabilidad. El riesgo puede ser disminuido siguiendo cada una de las tres dimensiones: mejores mediciones, mantenimiento preventivo y mayor redundancia. El número de dimensiones puede reducirse siguiendo líneas del mismo riesgo calculadas para la distribución beta. Esta tesis presenta veinte estudios de fallos en interruptores y seccionadores con datos sobre la causa y el tiempo hasta la avería. Contiene también una visión general de cuarenta métodos para medir la condición de seccionadores e interruptores, aplicables en su mayoría a los contactos eléctricos y los componentes mecánicos. Se ha instalado un sistema con sensores infrarrojos en los seis contactos de nueve seccionadores. El objetivo es disminuir los cortes de servicio para mantenimiento, estimando la condición con el seccionador en servicio. Las temperaturas son transmitidas por radio y se hace una regresión con el cuadrado de la corriente, ya que el mejor exponente de la corriente resultó ser 2,0. $R^2$ alcanza un valor de 0,9 indicando un buen ajuste de los datos por parte del modelo. Existe un coeficiente de regresión para cada contacto y este sirve para ordenar los contactos según la necesidad de mantenimiento, ya que cuanto mayor sea el coeficiente más calor se produce en el contacto. Finalmente se explica que el modelado de tiempo hasta la avería consiste en asignar una distribución estadística a cada equipo. La monitorización del estado consiste en medir y estimar un parámetro y luego predecir su valor en el futuro. Si va a sobrepasar un cierto límite, el equipo necesitará de mantenimiento. Se presenta el efecto de mantenimiento de cuatro seccionadores. Un desarrollo de la matriz de riesgo, un conjunto de estadísticas y métodos de monitoreo de condición, un sistema de sensores IR situados cerca de los contactos, en método de planificación de termografía y un concepto para explicar la modelización de tiempo hasta la avería y de la monitorización de la condición han sido presentados y hace posible una mejor planificación de los cortes de servicio.<br><p>QC 20170928</p>

Diabetes mellitus is a chronic disease characterized by dysfunctions of the normal regulation of glucose concentration in the blood. In Type 1 diabetes the pancreas is unable to produce insulin, while in Type 2 diabetes derangements in insulin secretion and action occur. As a consequence, glucose concentration often exceeds the normal range (70-180 mg/dL), with short- and long-term complications. Hypoglycemia (glycemia below 70 mg/dL) can progress from measurable cognition impairment to aberrant behaviour, seizure and coma. Hyperglycemia (glycemia above 180 mg/dL) predisposes to invalidating pathologies, such as neuropathy, nephropathy, retinopathy and diabetic foot ulcers. Conventional diabetes therapy aims at maintaining glycemia in the normal range by tuning diet, insulin infusion and physical activity on the basis of 4-5 daily self-monitoring of blood glucose (SMBG) measurements, obtained by the patient using portable minimally-invasive lancing sensor devices. New scenarios in diabetes treatment have been opened in the last 15 years, when minimally invasive continuous glucose monitoring (CGM) sensors, able to monitor glucose concentration in the subcutis continuously (i.e. with a reading every 1 to 5 min) over several days (7-10 consecutive days), entered clinical research. CGM allows tracking glucose dynamics much more effectively than SMBG and glycemic time-series can be used both retrospectively, e.g. to optimize metabolic control therapy, and in real-time applications, e.g. to generate alerts when glucose concentration exceeds the normal range thresholds or in the so-called “artificial pancreas”, as inputs of the closed loop control algorithm. For what concerns real time applications, the possibility of preventing critical events is, clearly, even more appealing than just detecting them as they occur. This would be doable if glucose concentration were known in advance, approximately 30-45 min ahead in time. The quasi continuous nature of the CGM signal renders feasible the use of prediction algorithms which could allow the patient to take therapeutic decisions on the basis of future instead of current glycemia, possibly mitigating/ avoiding imminent critical events. Since the introduction of CGM devices, various methods for short-time prediction of glucose concentration have been proposed in the literature. They are mainly based on black box time series models and the majority of them uses only the history of the CGM signal as input. However, glucose dynamics are influenced by many factors, e.g. quantity of ingested carbohydrates, administration of drugs including insulin, physical activity, stress, emotions and inter- and intra-individual variability is high. For these reasons, prediction of glucose time course is a challenging topic and results obtained so far may be improved. The aim of this thesis is to investigate the possibility of predicting future glucose concentration, in the short term, using new models based on neural networks (NN) exploiting, apart from CGM history, other available information. In particular, we first develop an original model which uses, as inputs, the CGM signal and information on timing and carbohydrate content of ingested meals. The prediction algorithm is based on a feedforward NN in parallel with a linear predictor. Results are promising: the predictor outperforms widely used state of art techniques and forecasts are accurate and allow obtaining a satisfactory time anticipation. Then we propose a second model, which exploits a different NN architecture, a jump NN, which combines benefits of both feedforward NN and linear algorithm obtaining performance similar to the previously developed predictor, although the simpler structure. To conclude the analysis, information on doses of injected bolus of insulin are added as input of the jump NN and the relative importance of every input signal in determining the NN output is investigated by developing an original sensitivity analysis. All the proposed predictors are assessed on real data of Type 1 diabetics, collected during the European FP7 project DIAdvisor. To evaluate the clinical usefulness of prediction in improving diabetes management we also propose a new strategy to quantify, using an in silico environment, the reduction of hypoglycemia when alerts and relative therapy are triggered on the basis of prediction, obtained with our NN algorithm, instead of CGM. Finally, possible inclusion of additional pieces of information such as physical activity is investigated, though at a preliminary level. The thesis is organized as follows. Chapter 1 gives an introduction to the diabetes disease and the current technologies for CGM, presents state of art techniques for short-time prediction of glucose concentration of diabetics and states the aim and the novelty of the thesis. Chapter 2 discusses NN paradigms from a theoretical point of view and specifies technical details common to the design and implementation of all the NN algorithms proposed in the following. Chapter 3 describes the first prediction model we propose, based on a NN in parallel with a linear algorithm. Chapter 4 presents an alternative simpler architecture, based on a jump NN, and demonstrates its equivalence, in terms of performance, with the previously proposed algorithm. Chapter 5 further improves the jump NN, by adding new inputs and investigating their effective utility by a sensitivity analysis. Chapter 6 points out possible future developments, as the possibility of exploiting information on physical activity, reporting also a preliminary analysis. Finally, Chapter 7 describes the application of NN for generation of preventive hypoglycemic alerts and evaluates improvement of diabetes management in a simulated environment. Some concluding remarks end the thesis.<br>Il diabete mellito è una patologia cronica caratterizzata da disfunzioni della regolazione della concentrazione di glucosio nel sangue. Nel diabete di Tipo 1 il pancreas non produce l'ormone insulina, mentre nel diabete di Tipo 2 si verificano squilibri nella secrezione e nell'azione dell'insulina. Di conseguenza, spesso la concentrazione glicemica eccede le soglie di normalità (70-180 mg/dL), con complicazioni a breve e lungo termine. L'ipoglicemia (glicemia inferiore a 70 mg/dL) può risultare in alterazione delle capacità cognitive, cambiamenti d'umore, convulsioni e coma. L'iperglicemia (glicemia superiore a 180 mg/dL) predispone, nel lungo termine, a patologie invalidanti, come neuropatie, nefropatie, retinopatie e piede diabetico. L'obiettivo della terapia convenzionale del diabete è il mantenimento della glicemia nell'intervallo di normalità regolando la dieta, la terapia insulinica e l'esercizio fisico in base a 4-5 monitoraggi giornalieri della glicemia, (Self-Monitoring of Blood Glucose, SMBG), effettuati dal paziente stesso usando un dispositivo pungidito, portabile e minimamente invasivo. Negli ultimi 15 anni si sono aperti nuovi orizzonti nel trattamento del diabete, grazie all'introduzione, nella ricerca clinica, di sensori minimamente invasivi (Continuous Glucose Monitoring, CGM) capaci di misurare la glicemia nel sottocute in modo quasi continuo (ovvero con una misurazione ogni 1-5 min) per parecchi giorni consecutivi (dai 7 ai 10 giorni). I sensori CGM permettono di monitorare le dinamiche glicemiche in modo più fine delle misurazioni SMBG e le serie temporali di concentrazione glicemica possono essere utilizzate sia retrospettivamente, per esempio per ottimizzare la terapia di controllo metabolico, sia prospettivamente in tempo reale, per esempio per generare segnali di allarme quando la concentrazione glicemica oltrepassa le soglie di normalità o nel “pancreas artificiale”. Per quanto concerne le applicazioni in tempo reale, poter prevenire gli eventi critici sarebbe chiaramente più attraente che semplicemente individuarli, contestualmente al loro verificarsi. Ciò sarebbe fattibile se si conoscesse la concentrazione glicemia futura con circa 30-45 min di anticipo. La natura quasi continua del segnale CGM rende possibile l'uso di algoritmi predittivi che possono, potenzialmente, permettere ai pazienti diabetici di ottimizzare le decisioni terapeutiche sulla base della glicemia futura, invece che attuale, dando loro l'opportunità di limitare l'impatto di eventi pericolosi per la salute, se non di evitarli. Dopo l'introduzione nella pratica clinica dei dispositivi CGM, in letteratura, sono stati proposti vari metodi per la predizione a breve termine della glicemia. Si tratta principalmente di algoritmi basati su modelli di serie temporali e la maggior parte di essi utilizza solamente la storia del segnale CGM come ingresso. Tuttavia, le dinamiche glicemiche sono determinate da molti fattori, come la quantità di carboidrati ingeriti durante i pasti, la somministrazione di farmaci, compresa l'insulina, l'attività fisica, lo stress, le emozioni. Inoltre, la variabilità inter- e intra- individuale è elevata. Per questi motivi, predire l'andamento glicemico futuro è difficile e stimolante e c'è margine di miglioramento dei risultati pubblicati finora in letteratura. Lo scopo di questa tesi è investigare la possibilità di predire la concentrazione glicemica futura, nel breve termine, utilizzando modelli basati su reti neurali (Neural Network, NN) e sfruttando, oltre alla storia del segnale CGM, altre informazioni disponibili. Nel dettaglio, inizialmente svilupperemo un nuovo modello che utilizza, come ingressi, il segnale CGM e informazioni relative ai pasti ingeriti, (istante temporale e quantità di carboidrati). L'algoritmo predittivo sarà basato su una NN di tipo feedforward, in parallelo ad un modello lineare. I risultati sono promettenti: il modello è superiore ad algoritmi stato dell'arte ampiamente utilizzati, la predizione è accurata e il guadagno temporale è soddisfacente. Successivamente proporremo un nuovo modello basato su una differente architettura di NN, ovvero una “jump NN”, che fonde i benefici di una NN di tipo feedforward e di un algoritmo lineare, ottenendo risultati simili a quelli del modello precedentemente proposto, nonostante la sua struttura notevolmente più semplice. Per completare l'analisi, valuteremo l'inclusione, tra gli ingressi della jump NN, di segnali ottenuti sfruttando informazioni sulla terapia insulinica (istante temporale e dose dei boli iniettati) e valuteremo l'importanza e l'influenza relativa di ogni ingresso nella determinazione del valore glicemico predetto dalla NN, sviluppando un'originale analisi di sensitività. Tutti i modelli proposti saranno valutati su dati reali di pazienti diabetici di Tipo 1, raccolti durante il progetto Europeo FP7 (7th Framework Programme, Settimo Programma Quadro) DIAdvisor. Per valutare l'utilità clinica della predizione e il miglioramento della gestione della terapia diabetica proporremo una nuova strategia per la quantificazione, in simulazione, della riduzione del numero e della gravità degli eventi ipoglicemici nel caso gli allarmi, e la relativa terapia, siano determinati sulla base della concentrazione glicemica predetta, utilizzando il nostro algoritmo basato su NN, invece che su quella misurata dal sensore CGM. Infine, investigheremo, in modo preliminare, la possibilità di includere, tra gli ingressi della NN, ulteriori informazioni, come l'attività fisica. La tesi è organizzata come descritto in seguito. Il Capitolo 1 introduce la patologia diabetica e le attuali tecnologie CGM, presenta le tecniche stato dell'arte utilizzate per la predizione a breve termine della glicemia di pazienti diabetici e specifica gli scopi e le innovazioni della presente tesi. Il Capitolo 2 introduce le basi teoriche delle NN e specifica i dettagli tecnici che abbiamo scelto di adottare per lo sviluppo e l'implementazione di tutte le NN proposte in seguito. Il Capitolo 3 descrive il primo modello proposto, basato su una NN in parallelo a un algoritmo lineare. Il Capitolo 4 presenta una struttura alternativa più semplice, basata su una jump NN, e dimostra la sua equivalenza, in termini di prestazioni, con il modello precedentemente proposto. Il Capitolo 5 apporta ulteriori miglioramenti alla jump NN, aggiungendo nuovi ingressi e investigando la loro utilità effettiva attraverso un'analisi di sensitività. Il Capitolo 6 indica possibili sviluppi futuri, come l'inclusione di informazioni sull'attività fisica, presentando anche un'analisi preliminare. Infine, il Capitolo 7 applica la NN per la generazione di allarmi preventivi per l'ipoglicemia, valutando, in simulazione, il miglioramento della gestione del diabete. Alcuni commenti e osservazioni concludono la tesi.