Search

We investigate a new dynamical system that describes tumor-host interaction. The equation that describes the untreated tumor growth is based on non-extensive statistical mechanics. Recently, this model has been shown to fit successfully exponential, Gompertz, logistic, and power-law tumor growths. We have been able to include as many hallmarks of cancer as possible. We study also the dynamic response of cancer under therapy. Using our model, we can make predictions about the different outcomes when we change the parameters, and/or the initial conditions. We can determine the importance of different factors to influence tumor growth. We discover synergistic therapeutic effects of different treatments and drugs. Cancer is generally untreatable using conventional monotherapy. We consider conventional therapies, oncogene-targeted therapies, tumor-suppressors gene-targeted therapies, immunotherapies, anti-angiogenesis therapies, virotherapy, among others. We need therapies with the potential to target both tumor cells and the tumors' microenvironment. Drugs that target oncogenes and tumor-suppressor genes can be effective in the treatment of some cancers. However, most tumors do reoccur. We have found that the success of the new therapeutic agents can be seen when used in combination with other cancer-cell-killing therapies. Our results have allowed us to design a combinational therapy that can lead to the complete eradication of cancer., Comment: 35 pages, 6 figures

In this paper, the temporal dynamics of precipitation in northwestern France in 8 relation to the dominant climatic pattern in Europe were investigated. The general trends 9 and the nonstationary behavior of the North Atlantic Oscillation (NAO) were described 10 using fractal analysis and Fourier spectral and continuous wavelet analysis of a NAO 11 index time series over the 1865–2004 period. The 44-year and 8-year multidecadal 12 components showed a clear increasing power during the second half of the last century. 13 The possible link between rainfall variability and the NAO was then investigated. The 14 links between the NAO and rainfall were not as obvious, as NAO-like components were 15 not easily retrieved in the precipitation records: Relationships between the NAO and 16 rainfall were very likely at certain timescales but were not systematically very obvious. 17 For instance, the characteristic quasi-biennal oscillation (QBO) of the NAO was barely 18 detected in precipitation, but a 6-year fluctuation beginning in the 90s was found to be 19 statistically significant at a 95% confidence limit After investigating a possible link 20 with the Southern Oscillation Index, the occurrence of this fluctuation in the beginning of 21 the 90s could be related to the observed shift of the QBO toward slightly higher 22 frequencies in the NAO time series. On the other hand, a modulation of the QBO by a 23 6-year interannual fluctuation would suggest the existence of a 6-year climate pattern 24 that could affect precipitation and, to a lesser extent, the NAO. Cross-wavelet analysis 25 between the NAO and precipitation revealed a loss in correlation across the 1970–2005 26 period, which seemed to be the fact of a QBO-like fluctuation. This loss of correlation 27 was related to the above-mentioned shift of the QBO and 6-year rainfall interannual 28 band since the 90s.

International audience

International audience; In karst systems, rain events often result in a decrease in conductivity (a tracer of dissolved phase transport) and an increase in turbidity (a tracer of suspended solids transport) at wells and springs. We investigated the response to rain events at five karst outlets (three springs and two wells) discharging from the cretaceous chalk aquifer of the Haute-Normandie region, France. As the input signal (rainfall) is relatively homogeneous spatially, the differences in the responses at the sites provide information on the internal organization of the karst systems and their hydrodynamic functioning. We used autocorrelation functions and turbidity-conductivity (T-[Delta]C) hysteresis curves to analyze the responses. At short time scales, autocorrelation functions allow estimation of the inertia of the conduit flow system. The T-[Delta]C curves reflect the processes of deposition, resuspension, and direct transport of suspended solids within the karst network. The results show that: (i) a long memory effect for conductivity indicates storage of water in the karst network and deposition of suspended solids, and the transport of the dissolved phase and suspended solids is not the same; (ii) a short memory effect for conductivity indicates that transport of the dissolved phase and suspended solids is synchronous and direct. On the basis of the results at the five sites, we propose three conceptual models of hydrodynamic functioning.

International audience

Flood Forecasting is an old hydrological problem, which has not yet received a satisfying answer. In the framework of the RIO2 research program, we aimed to compare and improve flood-forecasting tools, while focusing on a very interesting but challenging area: the upper catchment of the Loire River. We relied exclusively on methods involving rainfall-runoff modelling, which is the best-adapted in this area in terms of forecast lead time, as well as in terms of applicability on upland catchments. More generally, the research presented here is a contribution to the development of flood-forecasting tools targeting operational services. Around 15 catchments were studied, with an area between 30 and 5000 km². Several modelling approaches were tested. We obtained promising results, both at the hourly and the daily time step. We started with a methodological review, which was needed as a basis for our comparative approach: we defined criteria that could be used to assess model efficiency in a flood-forecasting context. We also present an approach allowing estimating the maximum possible forecast lead on a given basin. Last, we reviewed the possibilities for updating hydrological models for flood forecasting applications. Then our work focused on intercomparing and improving rainfall-runoff models. We were inspired by the direct assimilation of Artificial Neural Networks to propose a similar solution for a conceptual rainfall-runoff model (GR), which appears quite promising for its efficiency and its robustness. We also had considered improving forecasting efficiency through a multi-model approach. But the test we made (on a single basin) did not give the expected success, and resulted only in an uncertainty processing. Last, we studied the sensitivity of flood forecasting models to the quality of rainfall information, and assessed the minimum number of raingages needed for reliable flood forecasting on the Loire River.The main perspectives for continuing research encompasses pursuing the quest for an efficient multi-model approach, validating the GR-Prévi algorithm on a larger number of watersheds, comparing with physically-based approaches, and optimising raingage networks., La prévision des crues est un problème ancien, qui reste encore aujourd`hui difficile à résoudre. Dans le cadre du programme "Risque d'Inondation" (RIO2), nous avons voulu comparer et améliorer des outils de prévision des crues, en nous concentrant sur une zone particulièrement intéressante (mais difficile) : le haut bassin de la Loire. Nous nous sommes exclusivement intéressés à la modélisation pluie-débit, qui est la mieux adaptée à la prévision dans cette zone, notamment pour ce qui est des délais de prévision possibles, et de son utilisation sur des bassins amont. Ce travail s'inscrivait de manière plus générale dans le cadre du développement d'outils pluie-débit de prévision des crues pour les services opérationnels. Une quinzaine de bassins versants, d'une taille comprise entre 30 et 5000 km² au total ont été étudiés, et plusieurs approches de modélisation ont été testées. Des résultats intéressants ont été obtenus, à la fois au pas de temps horaire et au pas de temps journalier. Nous avons en premier lieu effectué un travail méthodologique indispensable pour pouvoir progresser dans notre démarche comparative : la définition des critères utilisables pour évaluer la performance de modèles de prévision des crues. Nous avons également proposé une démarche permettant d`évaluer le délai maximum de prévision sur un bassin. Enfin, nous avons abordé la question de la mise à jour des modèles hydrologiques pour une application en prévision. Ensuite, notre travail a porté sur un aller-retour entre intercomparaisons et améliorations méthodologiques sur les modèles pluie-débit utilisés. Nous avons proposé d'imiter l`assimilation directe des dernières données de débit connues dans les RNA pour faire progresser les modèles conceptuels (GR) et proposer une version à assimilation directe, qui apparaît très prometteuse par sa performance et par sa robustesse. Nous avons aussi mis en oeuvre une approche multi-modèle, proposée à l`origine dans un double objectif d`amélioration des performances et de la robustesse des prévisions de crue et de calcul des incertitudes. Pour le premier objectif, nous n`avons pas remporté le succès escompté, et l`approche multi-modèle n`aura finalement permis qu`un calcul d`incertitude. Enfin, nous avons étudié la sensibilité des modèles de prévision de crue à la qualité de l`information pluviométrique et proposé une méthode permettant d'évaluer le nombre de pluviomètres nécessaires à une prévision fiable des crues. Les principales perspectives de poursuite du travail engagé concernent l`approche multi-modèle, la validation de l`algorithme GR-Prévi, la comparaison avec des approches à base physique, et l`optimisation des réseaux pluviométriques.