Search

Due to climate change, we expect an exacerbation of fire in Europe and around the world, with major wildfire events extending to northern latitudes and boreal regions [1]. In this context, it is important to improve our capabilities to anticipate fire danger and understand its driving mechanisms at a global scale. As the earth is an interconnected system, large-scale processes can have an effect on the global climate and fire seasons. For example, extreme fires in Siberia have been linked to previous-year surface moisture conditions and anomalies in the Arctic Oscillation [2]. As part of the ESA-funded project SeasFire (https://seasfire.hua.gr), we gather and harmonize data related to seasonal fire drivers and develop deep learning models that are able to capture spatiotemporal associations with the goal to forecast burned area sizes on a seasonal scale, globally. We publish a global analysis-ready datacube for seasonal fire forecasting for the years 2001-2021 at a spatiotemporal resolution of 0.25 deg x 0.25 deg x 8 days [3]. The datacube includes a combination of variables describing the seasonal fire drivers , namely climate, vegetation, oceanic indices, human factors, land cover and the burned areas. We leverage the availability of big EO data and the advances in Deep Learning modeling [4, 5] to forecast global burned areas, capture the spatio-temporal interactions of the Earth System variables and identify potential teleconnections that determine wildfire regimes under the light of climate change. We present deep learning models that handle the Earth as a system, such as graph neural networks and transformer-based architectures. Applied on the prediction of wildfires at different temporal horizons we reveal that our deep learning models skillfully predict burned area patterns. Exploring the explanation of the models, we reveal important spatio-temporal links.Our approach, using AI to model the earth as a system and capture long spatio-temporal interactions, showcases the potential of an application-specific digital twin. The SeasFire datacube can be exploited as a baseline digital twin for modeling different natural hazards, including floods, heatwaves, and droughts. Thus, we will discuss insights and future directions for digital twins in anticipating climate extremes, inspired by our global wildfire prediction paradigm. [1] Wu, Chao, et al. "Historical and future global burned area with changing climate and human demography." One Earth 4.4 (2021): 517-530.[2] Kim, Jin-Soo, et al. "Extensive fires in southeastern Siberian permafrost linked to preceding Arctic Oscillation." Science advances 6.2 (2020): eaax3308.[3] Alonso, Lazaro, et al. Seasfire Cube: A Global Dataset for Seasonal Fire Modeling in the Earth System. Zenodo, 15 July 2022, p., doi:10.5281/zenodo.6834584.[4] Kondylatos, Spyros et al. “Wildfire Danger Prediction and Understanding with Deep Learning.” Geophysical Research Letters, 2022. doi: 10.1029/2022GL099368[5] Prapas, Ioannis et al. “Deep Learning for Global Wildfire Forecasting.” NeurIPS 2022 workshop on Tackling Climate Change with Machine Learning, doi: 10.48550/arXiv.2211.00534

Deep Learning (DL) has consistently surpassed other Machine Learning methods and achieved state-of-the-art performance in multiple cases. Several modern applications like financial and recommender systems require models that are constantly updated with fresh data. The prominent approach for keeping a DL model fresh is to trigger full retraining from scratch when enough new data are available. However, retraining large and complex DL models is time-consuming and compute-intensive. This makes full retraining costly, wasteful, and slow. In this paper, we present an approach to continuously train and deploy DL models. First, we enable continuous training through proactive training that combines samples of historical data with new streaming data. Second, we enable continuous deployment through gradient sparsification that allows us to send a small percentage of the model updates per training iteration. Our experimental results with LeNet5 on MNIST and modern DL models on CIFAR-10 show that proactive training keeps models fresh with comparable—if not superior—performance to full retraining at a fraction of the time. Combined with gradient sparsification, sparse proactive training enables very fast updates of a deployed model with arbitrarily large sparsity, reducing communication per iteration up to four orders of magnitude, with minimal—if any—losses in model quality. Sparse training, however, comes at a price; it incurs overhead on the training that depends on the size of the model and increases the training time by factors ranging from 1.25 to 3 in our experiments. Arguably, a small price to pay for successfully enabling the continuous training and deployment of large DL models.

Endometriosis is a benign gynecologic disorder, affecting up to 10% of women, characterized by the presence of functional endometrial tissue at ectopic positions generally within the peritoneum. It is a heritable condition influenced by multiple genetic and environmental factors, with an overall heritability estimated at approximately 50%. In this study, we investigated whether single nucleotide polymorphisms (SNPs) rs7521902, rs10859871 and rs11031006, mapping to WNT4, VEZT and FSHB genetic loci, respectively, are associated with risk for endometriosis in a Greek population. This study included 166 women with histologically confirmed endometriosis diagnosed through surgery and 150 normal controls. Genotyping of the rs7521902, rs10859871 and rs11031006 SNPs was performed with Taqman primer/probe sets. A significant association was detected with the AC genotype of rs7521902 (WNT4) in patients with stage III and IV disease only. Evidence for association with endometriosis was also found for the AC genotype of the rs10859871 of VEZT. Notably, a significant difference in the distribution of the AG genotype and the minor allele A of FSHB rs11031006 SNP was found between the endometriosis patients and controls. We found a genetic association between rs11031006 (FSHB) SNP and endometriosis. WNT4 and VEZT genes constitute the most consistently associated genes with endometriosis. In the present study, an association of rs7521902 (WNT4) and rs10859871 (VEZT) was confirmed in women with endometriosis at the genotypic but not the allelic level.

We approach the problem of human action recognition in videos by distinguishing between simple and complex actions. To recognize simple actions, we take advantage of the latest advances with 3D convolutional networks, which are able to offer a generic video snippet descriptor. For the complex ones, which involve interaction between more than one individual, we use the recognized simple human actions of the previous step to generate Event Calculus theories. This way, we aim to achieve a high-level human action understanding, combining the opaque effectiveness of deep learning and the transparent reasoning of computational logic. Our experimental results on a benchmark activity recognition dataset encourage further research towards this direction.

The usefulness of various pathways inhibitors, Everolimus, an inhibitor of mammalian target of rapamycin (mTOR), Infliximab, a monoclonal antibody which blocks the tumor necrosis factor-a (TNF-a), Erlotinib, a tyrosine protein kinase inhibitor of the epidermal growth factor receptor (EGFR), Metformin, an activator of AMP-activated protein kinase enzyme (AMPK) and vascular permeability reducers were explored in an ovarian hyperstimulation syndrome (OHSS) rat model. Sixty-three female Wistar rats were randomly divided in seven groups. The control group received saline, while the OHSS group received recombinant -- follicle-stimulating hormone (rec-FSH) for four consecutive days. The other five groups received rec-FSH for 4 d and Everolimus daily, Infliximab once, Erlotinib daily, Metformin daily and Vitamin C daily, respectively. All groups received human chorionic gonadotropin (hCG) at the fifth day. The efficacy of Everolimus administration for various intervals was also explored. Significantly reduced ovarian weight was observed in the Everolimus group (rec-FSH + hCG + mTOR inhibitor) compared to the OHSS group (p < 0.001). The Everolimus group also showed the lowest progesterone (PRG) concentration (p = 0.007). The Erlotinib group (rec-FSH + hCG + EGFR inhibitor) presented with the lowest graafian follicle number, while the Everolimus group was characterized by the lowest corpus luteum number. The vascular permeability and the estradiol levels did not differ between groups. Finally, the Everolimus intra-comparison showed no difference in all measured outcomes. Studying the different pathways linked to vascular endothelial growth factor (VEGF) pathway, we conclude that targeting mTOR pathways is beneficial for reducing ovarian weight and PRG levels in an OHSS animal model.

The efficacy of pathways inhibition and the combined effect of Everolimus (mTOR inhibitor) and Verapamil (CYP3A inhibitor) in ovarian hyperstimulation syndrome (OHSS) need to be tested. Therefore, the impact of a leucotriene receptor antagonist, an anticoagulant, a GnRH antagonist as well as Everolimus plus Verapamil (at various doses and days of administration) on an OHSS rat model was tested. Sixty three female Wistar rats were randomly divided into seven groups. The control group received saline, while the OHSS group received rec-FSH for four consecutive days. The other five groups received rec-FSH for four days and Montelukast daily, Heparin daily, GnRH antagonist daily, Everolimus plus Verapamil in the last two days (half days group) and Everolimus plus Verapamil (half dose group) daily, respectively. All groups received also hCG at the fifth day. Significantly reduced ovarian weight was observed in the Everolimus plus Verapamil groups (half days and half-dose groups) and the Montelukast group compared to the OHSS group (p = 0.001 and p = 0.001, respectively). The vascular permeability was significantly reduced in the Everolimus plus Verapamil group (half dose group) and the GnRH antagonist group compared to the OHSS group (p

OBJECTIVE: The efficacy of vascular endothelial growth factor (VEGF), COX-2, calcium and aromatase inhibitors in an ovarian hyperstimulation syndrome (OHSS) rat model was tested. METHODS: One hundred and eight female Wistar rats were randomly divided in nine groups. The control group received saline, while the OHSS group received rec-FSH for 4 consecutive days. The other seven groups received rec-FSH (4d) and Bevacizumab twice, Parecoxib daily, Verapamil daily, Parecoxib daily and Bevacizumab twice, Verapamil daily and Bevacizumab twice, Parecoxib and Verapamil daily, Letrozole and Meloxicam daily, respectively. All groups received also hCG at the 5th day. RESULTS: All intervention groups were characterized by reduced vascular permeability compared to the OHSS group, which in the groups of Verapamil (Calcium inhibition) and Parecoxib + Verapamil (COX-2 + Calcium inhibition) presented significant statistical difference. The Verapamil group showed the lowest corpus luteum formation, while theParecoxib (COX-2 inhibition), the Parecoxib + Verapamil (COX-2 + Calcium inhibition), the Bevacizumab + Parecoxib (VEGF + COX-2 inhibition) and the Bevacizumab + Verapamil (VEGF + Calcium inhibition) groups were also characterized by lower corpus luteum numbers compared to the OHSS group. Furthermore, lower graafian follicle formation was observed in the above groups, while the ovarian weight and the hormonal profile were not significantly affected. CONCLUSIONS: Studying the different check points of the VEGF pathway, we conclude that targeting calcium pathways could be beneficial for the vascular permeability control in an OHSS animal model.

Climate change exacerbates the occurrence of extreme droughts and heatwaves, increasing the frequency and intensity of large wildfires across the globe. Forecasting wildfire danger and uncovering the drivers behind fire events become central for understanding relevant climate-land surface feedback and aiding wildfire management. In this work, we leverage Deep Learning (DL) to predict the next day's wildfire danger in a fire-prone part of the Eastern Mediterranean and explainable Artificial Intelligence (xAI) to diagnose model attributions. We implement DL models that capture the temporal and spatio-temporal context, generalize well for extreme wildfires, and demonstrate improved performance over the traditional Fire Weather Index. Leveraging xAI, we identify the substantial contribution of wetness-related variables and unveil the temporal focus of the models. The variability of the contribution of the input variables across wildfire events hints into different wildfire mechanisms. The presented methodology paves the way to more robust, accurate, and trustworthy data-driven anticipation of wildfires., The 28th IUGG General Assembly (IUGG2023) (Berlin 2023)