No description available.

Background The detection of early warning signs is essential in the long-term treatment of bipolar disorders. However, in bipolar patients’ daily life and outpatient treatment the assessment of upcoming state changes faces several difficulties. In this trial, we examine the effectiveness of a smartphone based automated feedback about ambulatory assessed early warning signs in prolonging states of euthymia and therefore preventing hospitalization. This study aims to assess, whether patients experience longer episodes of euthymia, when their treating psychiatrists receive automated feedback about changes in communication and activity. With this additional information an intervention at an earlier stage in the development of mania or depression could be facilitated. We expect that the amount of time will be longer between affective episodes in the intervention group. Methods/design The current study is designed as a randomized, multi-center, observer-blind, active-control, parallel group trial within a nationwide research project on the topic of innovative methods for diagnostics, prevention and interventions of bipolar disorders. One hundred and twenty patients with bipolar disorder will be randomly assigned to (1) the experimental group with included automated feedback or (2) the control group without feedback. During the intervention phase, the psychopathologic state of all participants is assessed every four weeks over 18 months. Kaplan-Meier estimators will be used for estimating the survival functions, a Log-Rank test will be used to formally compare time to a new episode across treatment groups. An intention-to-treat analysis will include data from all randomized patients. Discussion This article describes the design of a clinical trial investigating the effectiveness of a smartphone-based feedback loop. This feedback loop is meant to elicit early interventions at the detection of warning signs for the prevention of affective episodes in bipolar patients. This approach will hopefully improve the chances of a timely intervention helping patients to keep a balanced mood for longer periods of time. In detail, if our hypothesis can be confirmed, clinical practice treating psychiatrists will be enabled to react quickly when changes are automatically detected. Therefore, outpatients would receive an even more individually tailored treatment concerning time and frequency of doctor’s appointments. Trial registration: ClinicalTrials.gov: NCT02782910: Title: “Smartphone-based Ambulatory Assessment of Early Warning Signs (BipoLife_A3)”. Registered May 25 2016. Protocol Amendment Number: 03. Issue Date: 26 March 2018. Author(s): ES.

PhDH / Department of Hydrology and Water Resources / Demand for water resources has been on the increase and is compounded by population growth and related development demands. Thus, numerous sectors are affected by water scarcity and therefore effective management of drought-induced water deficit is of importance. Luvuvhu River Catchment (LRC), a tributary of the Limpopo River Basin in South Africa has witnessed an increasing frequency of drought events over the recent decades. Drought impacts negatively on communities’ livelihoods, development, economy, water resources, and agricultural yields. Drought assessment in terms of frequency and severity using Drought Indices (DI) in different parts of the world has been reported. However, the forecasting and prediction component which is significant in drought preparedness and setting up early warning systems is still inadequate in several regions of the world. This study aimed at characterising, assessing, and predicting drought conditions using DI as a drought quantifying parameter in the LRC. This was achieved through the application of hybrid statistical and machine learning models including predictions via a combination of hybrid models. Rainfall and temperature data were obtained from South African Weather Service, evapotranspiration, streamflow, and reservoir storage data were obtained from the Department of Water and Sanitation while root zone soil moisture data was derived from the NASA earth data Giovanni repository. The Standardised Precipitation Index (SPI), Standardised Precipitation Evapotranspiration Index (SPEI), Standardised Streamflow Index (SSI), and Nonlinear Aggregated Drought Index (NADI) were selected to assess and characterise drought conditions in the LRC. SPI is precipitation based, SPEI is precipitation and evapotranspiration based, SSI is based on streamflow while NADI is a multivariate index based on rainfall, potential evapotranspiration, streamflow, and storage reservoir volume. All indices detected major historical drought events that have occurred and reported over the study area, although the precipitation based indices were the only indices that categorised the 1991/1992 drought as extreme at 6- and 12- month timescales while the streamflow index and multivariate NADI underestimated the event. The most recent 2014/16 drought was also categorised to be extreme by the standardised indices. The study found that the multivariate index underestimates most historical drought events in the catchment. The indices further showed that the most prevalent drought events in the LRC were mild drought. Extreme drought events were the least found at 6.42%, 1.08%, 1.56%, and 4.4% for SPI, SPEI, SSI, and NADI, respectively. Standardised indices and NADI showed negative trends and positive upward trends, respectively. The positive trend showed by NADI depicts a decreased drought severity over the study period. Drought events were characterised based on duration, intensity, severity, and frequency of drought events for each decade of the 30 years considered in this study i.e. between 1986 – 1996, 1996 – 2006, 2006 – 2016. This was done to get finer details of how drought characteristics behaved at a 10-year interval over the study period. An increased drought duration was observed between 1986 - 1996 while the shortest duration was observed between 1996 - 2006 followed by 2006 - 2016. NADI showed an overall lowest catchment duration at 1- month timescale compared to the standardised indices. The relationship between drought severity and duration revealed a strong linear relationship across all indices at all timescales (i.e. an R2 of between 0.6353 and 0.9714, 0.6353 and 0.973, 0.2725 and 0.976 at 1-, 6- and 12- month timescales, respectively). In assessing the overall utilisation of an index, the five decision criteria (robustness, tractability, transparency, sophistication, and extendibility) were assigned a raw score of between one and five. The sum of the weighted scores (i.e. raw scores multiplied by the relative importance factor) was the total for each index. SPEI ranked the highest with a total weight score of 129 followed by the SSI with a score of 125 and then the SPI with a score of 106 while NADI scored the lowest with a weight of 84. Since SPEI ranked the highest of all the four indices evaluated, it is regarded as an index that best describes drought conditions in the LRC and was therefore used in drought prediction. Statistical (GAM-Generalised Additive Models) and machine learning (LSTM-Long Short Term Memory) based techniques were used for drought prediction. The dependent variables were decomposed using Ensemble Empirical Mode Decomposition (EEMD). Model inputs were determined using the gradient boosting, and all variables showing some relative off importance were considered to influence the target values. Rain, temperature, non-linear trend, SPEI at lag1, and 2 were found to be important in predicting SPEI and the IMFs (Intrinsic Mode Functions) at 1, 6- and 12- month timescales. Seven models were applied based on the different learning techniques using the SPEI time series at all timescales. Prediction combinations of GAM performed better at 1- and 6- month timescales while at 12- month, an undecomposed GAM outperformed the decomposition and the combination of predictions with a correlation coefficient of 0.9591. The study also found that the correlation between target values, LSTM, and LSTM-fQRA was the same at 0.9997 at 1- month and 0.9996 at 6- and 12- month timescales. Further statistical evaluations showed that LSTM-fQRA was the better model compared to an undecomposed LSTM (i.e. RMSE of 0.0199 for LSTM-fQRA over 0.0241 for LSTM). The best performing GAM and LSTM based models were used to conduct uncertainty analysis, which was based on the prediction interval. The PICP and PINAW results indicated that LSTM-fQRA was the best model to predict SPEI timeseries at all timescales. The conclusions drawn from drought predictions conducted in this study are that machine learning neural networks are better suited to predict drought conditions in the LRC, while for improved model accuracy, time series decomposition and prediction combinations are also implementable. The applied hybrid machine learning models can be used for operational drought forecasting and further be incorporated into existing early warning systems for drought risk assessment and management in the LRC for better water resources management. Keywords: Decomposition, drought, drought indices, early warning system, frequency, machine learning, prediction intervals, severity, water resources. / NRF

This master thesis focuses on enhancing the user interface experience for Airborne Early Warning and Control (AEW&C) system operators in a military context. Collaborating with Saab, a defense sector company, the study employs the research through design and a user-centered design approach to investigate user needs and how they interact with the interface. The research reveals that despite the unique nature of the defense industry, operators respond poorly to a difficult-to-use interface, leading to frustration and underutilization of functionalities. The study identifies contextual factors such as information overload, user interface design and personalization, task difficulty and lack of help systems, stress, and cognitive load. Through the development and testing of a new interface prototype, incorporating design feedback sessions and iterations, the thesis addresses these contextual demands. The findings highlight the significance of introducing user experience in military environments, where complex system engineering and functionality have been traditionally preferred over simplicity and usability.

[pt] A Internet das Coisas (IoT) se propõe a interligar o mundo físico e a Internet, o que abre a possibilidade de desenvolvimento de diversas aplicações, principalmente na área da saúde. Essas aplicações requerem um grande número de sensores para coletar informações continuamente, gerando grandes fluxos de dados, muitas vezes excessivos, redundantes ou sem significado para as operações do sistema. Essa geração massiva de dados de sensores desperdiça recursos computacionais para adquirir, transmitir, armazenar e processar informações, levando à perda de eficiência desses sistemas ao longo do tempo. Além disso, os dispositivos IoT são projetados para serem pequenos e portáteis, alimentados por baterias, para maior mobilidade e interferência minimizada no ambiente monitorado. No entanto, esse design também resulta em restrições de consumo de energia, tornando a vida útil da bateria um desafio significativo que precisa ser enfrentado. Além disso, esses sistemas geralmente operam em ambientes imprevisíveis, o que pode gerar alarmes redundantes e insignificantes, tornando-os ineficazes. No entanto, um sistema auto-adaptativo que identifica e prevê riscos iminentes através de um sistema de pontuação de alertas antecipados (EWS) pode lidar com esses problemas. Devido ao seu baixo custo de processamento, a referência EWS pode ser incorporada em dispositivos vestíveis e sensores, permitindo um melhor gerenciamento das taxas de amostragem, transmissões, produção de alarmes e consumo de energia. Seguindo a ideia acima, esta tese apresenta uma solução que combina um sistema EWS com um algoritmo auto-adaptativo em aplicações IoT de monitoramento de pacientes. Desta forma, promovendo uma redução na aquisição e transmissão de dados , diminuindo alarmes não acionáveis e proporcionando economia de energia para esses dispositivos. Além disso, projetamos e desenvolvemos um protótipo de hardware capaz de embarcar nossa proposta, evidenciando a sua viabilidade técnica. Além disso, usando nosso protótipo, coletamos dados reais de consumo de energia dos componentes de hardware que foram usados durante nossas simulações com dados reais de pacientes provenientes de banco de dados públicos. Nossos experimentos demonstraram grandes benefícios com essa abordagem, reduzindo em 87 por cento os dados amostrados, em 99 por cento a carga total das mensagens transmitidas do dispositivo de monitoramento, 78 por cento dos alarmes e uma economia de energia de quase 82 por cento. No entanto, a fidelidade do monitoramento do estado clínico dos pacientes apresentou um erro absoluto total médio de 6,8 por cento (mais ou menos 5,5 por cento), mas minimizado para 3,8 por cento (mais ou menos 2,8 por cento) em uma configuração com menores ganhos na redução de dados. A perda de detecção total dos alarmes dependendo da configuração de frequências e janelas de tempo analisadas ficou entre 0,5 por cento e 9,5 por cento, com exatidão do tipo de alarme entre 89 por cento e 94 por cento. Concluindo, este trabalho apresenta uma abordagem para o uso mais eficiente de recursos computacionais, de comunicação e de energia para implementar aplicativos de monitoramento de pacientes baseados em IoT. / [en] The Internet of Things (IoT) proposes to connect the physical world to the Internet, which opens up the possibility of developing various applications, especially in healthcare. These applications require a huge number of sensors to collect information continuously, generating large data flows, often excessive, redundant, or without meaning for the system s operations. This massive generation of sensor data wastes computational resources to acquire, transmit, store, and process information, leading to the loss of efficiency of these systems over time. In addition, IoT devices are designed to be small and portable, powered by batteries, for increased mobility and minimized interference with the monitored environment. However, this design also results in energy consumption restrictions, making battery lifetime a significant challenge that needs to be addressed. Furthermore, these systems often operate in unpredictable environments, which can generate redundant and negligible alarms, rendering them ineffective. However, a self-adaptive system that identifies and predicts imminent risks using early-warning scores (EWS) can cope with these issues. Due to its low processing cost, EWS guidelines can be embedded in wearable and sensor devices, allowing better management of sampling rates, transmissions, alarm production, and energy consumption. Following the aforementioned idea, this thesis presents a solution combining EWS with a self-adaptive algorithm for IoT patient monitoring applications. Thus, promoting a reduction in data acquisition and transmission, decreasing non-actionable alarms, and providing energy savings for these devices. In addition, we designed and developed a hardware prototype capable of embedding our proposal, which attested to its technical feasibility. Moreover, using our wearable prototype, we collected the energy consumption data of hardware components and used them during our simulations with real patient data from public datasets. Our experiments demonstrated great benefits of our approach, reducing by 87 percent the sampled data, 99 percent the total payload of the transmitted messages from the monitoring device, 78 percent of the alarms, and an energy saving of almost 82 percent. However, the fidelity of monitoring the clinical status of patients showed a mean total absolute error of 6.8 percent (plus-minus 5.5 percent) but minimized to 3.8 percent (plus-minus 2.8 percent) in a configuration with lower data reduction gains. The total loss of alarm detection depends on the configuration of frequencies and time windows, remaining between 0.5 percent and 9.5 percent, with an accuracy of the type of alarm between 89 percent and 94 percent. In conclusion, this work presents an approach for more efficient use of computational, communication, and energy resources to implement IoT-based patient monitoring applications.

Only 70% of the Swedish students graduate from upper secondary school within the given time frame. Earlier research has shown that unfinished degrees disadvantage the individual student, policy makers and society. A first step for preventing dropouts is to indicate students about to fail courses. Thus the purpose is to identify tendencies whether a student will pass or not pass a course. In addition, the thesis accounts for the development of an Early Warning System to be applied to signal which students need additional support from a professional teacher. The used algorithm Random Forest functioned as a binary classification model of a failed grade against a passing grade. Data in the study are in samples of approximately 700 students from an upper secondary school within the Stockholm municipality. The chosen method originates from a Design Science Research Methodology that allows the stakeholders to be involved in the process. The results showed that the most dominant indicators for classifying correct were Absence, Previous grades and Mathematics diagnosis. Furthermore, were variables from the Learning Management System predominant indicators when the system also was utilised by teachers. The prediction accuracy of the algorithm indicates a positive tendency for classifying correctly. On the other hand, the small number of data points imply doubt if an Early Warning System can be applied in its current state. Thus, one conclusion is in further studies, it is necessary to increase the number of data points. Suggestions to address the problem are mentioned in the Discussion. Moreover, the results are analysed together with a review of the potential Early Warning Systemfrom a didactic perspective. Furthermore, the ethical aspects of the thesis are discussed thoroughly. / Endast 70% av svenska gymnasieelever tar examen inom den givna tidsramen. Tidigare forskning har visat att en oavslutad gymnasieutbildning missgynnar både eleven och samhället i stort. Ett första steg mot att förebygga att elever avviker från gymnasiet är att indikera vilka studenter som är på väg mot ett underkänt betyg i kurser. Därmed är syftet med rapporten att identifiera vilka trender som bäst indikerar att en elev kommer klara en kurs eller inte. Dessutom redogör rapporten för utvecklandet av ett förebyggande varningssystem som kan appliceras för att signalera vilka studenter som behöver ytterligare stöd från läraren och skolan. Algoritmen som användes var Random Forest och fungerar som en binär klassificeringsmodell av ett underkänt betyg mot ett godkänt. Den data som använts i studien är datapunkter för ungefär 700 elever från en gymnasieskola i Stockholmsområdet. Den valda metoden utgår från en Design Science Researchmetodik vilket möjliggör för intressenter att vara involverade i processen. Resultaten visade att de viktigaste variablerna var frånvaro, tidigare betyg och resultat från Stockholmsprovet (kommunal matematikdiagnos). Vidare var variabler från lärplattformen en viktig indikator ifall lärplattformen användes av läraren. Algoritmens noggrannhet indikerade en positiv trend för att klassificeringen gjordes korrekt. Å andra sidan är det tveksamt ifall det förebyggande systemet kan användas i sitt nuvarande tillstånd då mängden data som användes för att träna algoritmen var liten. Därav är en slutsats att det är nödvändigt för vidare studier att öka mängden datapunkter som används. I Diskussionen nämns förslag på hur problemet ska åtgärdas. Dessutom analyseras resultaten tillsammans med en utvärdering av systemet från ett didaktiskt perspektiv. Vidare diskuteras rapportens etiska aspekter genomgående.

Over the past decades, public concern about global environmental change has grown, following the progressive increase in both frequency and intensity of extreme events. Even though the problem is global, it has proved to have very different societal and environmental impacts at local level, further widening the gap between disadvantaged and advantaged communities, according to the degree of vulnerability of their social, economic and environmental systems. Among the various anthropogenic activities, the agricultural sector is particularly linked to global environmental change by a two-way relationship: on the one hand, intensive mono-cultures, together with intensive livestock production, compromise the environment and produce huge CO$_2$ emissions (one of the most important factors behind global warming); on the other hand, smallholder farming is one of the most endangered sectors by global environmental change, precisely because it depends heavily on the natural resources of the territory, including favourable weather and climate. Scientific research, supported by international institutions, has been working on this subject for several decades, analysing phenomena at global and local scale and providing medium and long-term forecasts capable of directing economic and political strategies. Such complex investigations become even more complex in contexts lacking reliable environmental data, where their low-quality and low representativeness weaken their reliability, compromising the reliability of the outcomes as well. This thesis seeks to respond to the increasing need of realistically addressing environmental phenomena that threaten rural communities and the environment on which they depend in low-income countries, by investigating two of the main environmental factors affecting tropical farming practices: river-floodplain dynamics and climate change. Despite data-related constraints, the environment of tropical rural areas still provides a unique opportunity to study several near-natural processes, such as the morphodynamics of mostly free-flowing rivers. Especially in foothill regions, unconfined or partially confined conditions of tropical rivers allow evaluating the natural dynamics of erodible river corridors, with erosion and accretion shaping their interactions with the adjacent floodplain and related human activities. At the same time, the complex terrain characterizing the river valleys at the foothills of high mountain chains also offers the opportunity to study interesting local meteorological processes, especially considering the interaction between synoptic-scale dynamics and local convective phenomena. In this context, local bottom-up initiatives and new and tailored-to-context strategies for adaptation to the ongoing environmental change are deepened following a multidisciplinary approach. This PhD research has been framed within an international cooperation project entitled “Sustainable Development and Fight against Climate Change in the Upper Huallaga basin (Peru)”, promoted by Mandacarù ONLUS, and funded by the Autonomous Province of Trento. The project aimed to enhance the resilience of the local farmers of the Upper Huallaga valley (Peru), facing the consequences of climate change and implementing new agricultural initiatives with a special attention to plantain and banana fields. Thanks to the support of the involved partners (Redesign by PROMER s.a.c., the Universidad Agraria Nacional de la Selva de Tingo Maria, in Peru, and the Edmund Mach Foundation of San Michele all’Adige, in Italy), the project provided the opportunity to carry out a consistent set of fieldwork activities over an 8-months period collecting hydro-morphological data, interviewing the local population, and installing two weather stations. The PhD thesis has been structured along two main parts, related to to the assessment of climate change effects on local agricultural practices, and the interplay between river-floodplain dynamics and floodplain agriculture. The part on the assessment of climate change includes two main research elements. First, a novel approach is used to evaluate climate change in data-scarce contexts: non-conventional data sources (population survey) are compared with conventional data sources (few local historical weather stations and global reanalysis data series – ERA5), to better account for the sub-daily time scale (local conventional sources only provide daily data), correlating weather changes perceived by farmers (more thunderstorms and longer drought periods) with climate variations deduced from quantitative data. Second, after having determined the most impacting meteorological variables on crops through the survey, a weather early-warning system has been developed to provide agro-meteorological forecasts to the \textit{bananeros} (banana farmers) of the Upper Huallaga valley. The system, based on the Weather Research and Forecasting (WRF) model, and enhanced with the assimilation of real-time observations from local meteorological stations installed during the project fieldwork, issues an alert when the predicted wind speed exceeds thresholds related to potential damage to the harvest, and spreads the warning via text messages. Such alerting system contains several novel features in relation to the socio-environmental context, allowing to discuss its potential for replication in analogous, vulnerable situations. The part on river-floodplain dynamics also includes two main research elements. First, a remote-sensing analysis is conducted at reach scale in two different reaches of the Huallaga River, quantifying geomorphological river trajectories and land use changes in the adjacent floodplain. The outcomes show that river morphology reacts differently depending on the agricultural systems (extensive or intensive) in the nearby floodplain, revealing a high geomorphological sensitivity of such a near-natural, highly dynamic river reach. Second, riverine agriculture within the erodible river corridor is analysed in association with riverine islands dynamics, at the geomorphic unit scale, evaluating the morphological evolution and agricultural suitability of two cultivated fluvial islands. The three main drivers of agricultural suitability within river erodible corridors, i.e. river disturbance, cultivation windows of opportunity, and soil suitability are quantified, allowing to generalize a process-based conceptual model of riverine islands as complex-adaptive-systems.

No / To build an early warning score (EWS) based exclusively on routinely undertaken laboratory tests that might provide early discrimination of in-hospital death and could be easily implemented on paper. Using a database of combined haematology and biochemistry results for 86,472 discharged adult patients for whom the admission specialty was Medicine, we used decision tree (DT) analysis to generate a laboratory decision tree early warning score (LDT-EWS) for each gender. LDT-EWS was developed for a single set (n=3496) (Q1) and validated in 22 other discrete sets each of three months long (Q2, Q3...Q23) (total n=82,976; range of n=3428 to 4093) by testing its ability to discriminate in-hospital death using the area under the receiver-operating characteristic (AUROC) curve. The data generated slightly different models for male and female patients. The ranges of AUROC values (95% CI) for LDT-EWS with in-hospital death as the outcome for the validation sets Q2-Q23 were: 0.755 (0.727-0.783) (Q16) to 0.801 (0.776-0.826) [all patients combined, n=82,976]; 0.744 (0.704-0.784, Q16) to 0.824 (0.792-0.856, Q2) [39,591 males]; and 0.742 (0.707-0.777, Q10) to 0.826 (0.796-0.856, Q12) [43,385 females]. CONCLUSIONS: This study provides evidence that the results of commonly measured laboratory tests collected soon after hospital admission can be represented in a simple, paper-based EWS (LDT-EWS) to discriminate in-hospital mortality. We hypothesise that, with appropriate modification, it might be possible to extend the use of LDT-EWS throughout the patient's hospital stay.

Serious illness in childhood is a rare occurrence, but accounts for 20% of childhood deaths. Early recognition and treatment of serious illness is vital if the child is to recover without long-term disability. It is known that vital signs such as heart rate, respiratory rate, temperature, and oxygen saturation can be used to identify children who are at high risk of serious illness. This thesis presents research into the development of a vital signs monitor, designed for use in the initial assessment of unwell children at their first point of contact with a medical practitioner. Child-friendly monitoring techniques are used to obtain vital signs, which can then be combined using data fusion techniques to assist clinicians in identifying children with serious illness. Existing normal ranges for heart rate and respiratory rate in childhood vary considerably, and do not appear to be based on clinical evidence. This thesis presents a systematic meta-analysis of heart rate and respiratory rate from birth to 18 years of age, providing evidence-based curves which can be used to assess the degree of abnormality in these important vital signs. Respiratory rate is particularly difficult to measure in children, but is known to be predictive of serious illness. Current methods of automated measurement can be distressing, or are time-consuming to apply. This thesis therefore presents a novel method for estimating the respiratory rate from an optical finger sensor, the pulse oximeter, which is routinely used in clinical practice. Information from multiple vital signs can be used to identify children at risk of serious illness. A number of data fusion techniques were tested on data collected from children attending primary and emergency care, and shown to outperform equivalent existing scoring systems when used to identify those with more serious illness.

La crise alimentaire de 2008 a suscité un regain d’intérêt pour les questions agricoles et de sécurité alimentaire dans les pays en développement. Partant du constat que près de 27% de la population d’Afrique Sub-saharienne souffre de malnutrition, cette thèse a pour objectif de contribuer à une meilleure compréhension des causes complexes de l’insécurité alimentaire. Le premier chapitre étudie les mécanismes de transmission des variations du prix mondial du riz aux prix domestiques dans trois pays ouest-africain: le Sénégal, le Tchad et le Mali. Les résultats indiquent que le prix du riz importé à Dakar et le prix du riz local à Bamakorépondent de façon asymétrique aux variations du prix mondial. Le chapitre 2 teste la présence d’obstacles aux échanges agricoles entre pays d’Afrique de l’Ouest et du Centre. Il ressort de l’analyse que le passage des frontières est coûteux. Toutefois, le coût associé au passage de la frontière est plus faible entre pays membre d’une même union économique et monétaire. Le chapitre 3 a pour objectif le renforcement des systèmes d’alertes précoces des crises alimentaires existants au Sahel. Il montre qu’il est possible d’anticiper les crises de prix avec six mois d’avance en analysant les mouvements passés des prix des céréales. Enfin, le chapitre 4 s’intéresse à la vulnérabilité des ménages face aux chocs pluviométriques. Il révèle que les ménages ruraux au Burkina Faso n’ont pas la capacité d’assurer ou d’absorber ces chocs climatiques. / This doctoral thesis is in line with the renewed interest in research on agriculture and food security, following the 2008 global food crisis. The aim of this thesis is to contribute to a better understanding of the complex issues surrounding food security. The first chapter investigates whether the changes in the international price of rice are transmitted to the domestic prices of rice in Senegal, Mali and Chad. Results indicate that the domestic prices of imported rice in Dakar and of local rice in Bamako react differently to changes in the world price depending on whether the world price is rising or falling. Chapter 2 analyses by how much trade barriers at the border and transport costs impede the integration of agricultural markets in West and Central Africa. Results highlight the role played by borders in explaining price deviations between markets. Additionally, belonging to an economic union and sharingthe same currency appear as major determinants of market integration. The third chapter aims at providing new early warning indicators based on food prices in Mali, Niger and Burkina Faso. Our analysis reveals that price crisis can be predicted about 6 months in advance through the observation of past price movements. Chapter 4 focuses on the analysis of children’s vulnerability to climate shocks in Burkina Faso. By combining health data originating from a 2008 household survey with meteorological data, we show the importance of weather conditions in prenatal period and in the first year of life on the future nutritional status of the children.