Machine-Learning-Based Forecasting of Dengue Fever in Brazilian Cities Using Epidemiologic and Meteorological Variables.

Dengue is a serious public health concern in Brazil and globally. In the absence of a universal vaccine or specific treatments, prevention relies on vector control and disease surveillance. Accurate and early forecasts can help reduce the spread of the disease. In this study, we developed a model for predicting monthly dengue cases in Brazilian cities 1 month ahead, using data from 2007-2019. We compared different machine learning algorithms and feature selection methods using epidemiologic and meteorological variables. We found that different models worked best in different cities, and a random forests model trained on monthly dengue cases performed best overall. It produced lower errors than a seasonal naive baseline model, gradient boosting regression, a feed-forward neural network, or support vector regression. For each city, we computed the mean absolute error between predictions and true monthly numbers of dengue cases on the test data set. The median error across all cities was 12.2 cases. This error was reduced to 11.9 when selecting the optimal combination of algorithm and input features for each city individually. Machine learning and especially decision tree ensemble models may contribute to dengue surveillance in Brazil, as they produce low out-of-sample prediction errors for a geographically diverse set of cities.

Forecasting new diseases in low-data settings using transfer learning.

Recent infectious disease outbreaks, such as the COVID-19 pandemic and the Zika epidemic in Brazil, have demonstrated both the importance and difficulty of accurately forecasting novel infectious diseases. When new diseases first emerge, we have little knowledge of the transmission process, the level and duration of immunity to reinfection, or other parameters required to build realistic epidemiological models. Time series forecasts and machine learning, while less reliant on assumptions about the disease, require large amounts of data that are also not available in early stages of an outbreak. In this study, we examine how knowledge of related diseases can help make predictions of new diseases in data-scarce environments using transfer learning. We implement both an empirical and a synthetic approach. Using data from Brazil, we compare how well different machine learning models transfer knowledge between two different dataset pairs: case counts of (i) dengue and Zika, and (ii) influenza and COVID-19. In the synthetic analysis, we generate data with an SIR model using different transmission and recovery rates, and then compare the effectiveness of different transfer learning methods. We find that transfer learning offers the potential to improve predictions, even beyond a model based on data from the target disease, though the appropriate source disease must be chosen carefully. While imperfect, these models offer an additional input for decision makers for pandemic response.

Higher Energy and Zinc Intakes from Complementary Feeding Are Associated with Decreased Risk of Undernutrition in Children from South America, Africa, and Asia.

BACKGROUND: Few studies have focused on quantitatively analyzing nutrients from infant diets, compromising complementary feeding evaluation and health promotion worldwide. OBJECTIVES: This study aimed to describe dietary intake in infants from 9 to 24 mo of age, determining nutrient intakes associated with the risk of underweight, wasting, and stunting. METHODS: Usual nutrient intakes from complementary feeding were determined by 24-h recalls collected when infants were 9-24 mo of age in communities from 7 low- and middle-income countries: Brazil (n = 169), Peru (n = 199), South Africa (n = 221), Tanzania (n = 210), Bangladesh (n = 208), India (n = 227), and Nepal (n = 229), totaling 1463 children and 22,282 food recalls. Intakes were corrected for within- and between-person variance and energy intake. Multivariable regression models were constructed to determine nutrient intakes associated with the development of underweight, wasting, and stunting at 12, 18, and 24 mo of age. RESULTS: Children with malnutrition presented significantly lower intakes of energy and zinc at 12, 18, and 24 mo of age, ranging from -16.4% to -25.9% for energy and -2.3% to -48.8% for zinc. Higher energy intakes decreased the risk of underweight at 12 [adjusted odds ratio (AOR): 0.90; 95% CI: 0.84, 0.96] and 24 mo (AOR: 0.91; 95% CI: 0.86, 0.96), and wasting at 18 (AOR: 0.91; 95% CI: 0.83, 0.99) and 24 mo (AOR: 0.83; 95% CI: 0.74, 0.92). Higher zinc intakes decreased the risk of underweight (AOR: 0.12; 95% CI: 0.03, 0.55) and wasting (AOR: 0.19; 95% CI: 0.04, 0.92) at 12 mo, and wasting (AOR: 0.05; 95% CI: 0.00, 0.76) at 24 mo. CONCLUSIONS: Higher intakes of energy and zinc in complementary feeding were associated with decreased risk of undernutrition in the studied children. Data suggest these are characteristics to be improved in children's complementary feeding across countries.

Centrality in earthquake multiplex networks.

Seismic time series has been mapped as a complex network, where a geographical region is divided into square cells that represent the nodes and connections are defined according to the sequence of earthquakes. In this paper, we map a seismic time series to a temporal network, described by a multiplex network, and characterize the evolution of the network structure in terms of the eigenvector centrality measure. We generalize previous works that considered the single layer representation of earthquake networks. Our results suggest that the multiplex representation captures better earthquake activity than methods based on single layer networks. We also verify that the regions with highest seismological activities in Iran and California can be identified from the network centrality analysis. The temporal modeling of seismic data provided here may open new possibilities for a better comprehension of the physics of earthquakes.

The role of community structure on the nature of explosive synchronization.

In this paper, we analyze explosive synchronization in networks with a community structure. The results of our study indicate that the mesoscopic structure of the networks could affect the synchronization of coupled oscillators. With the variation of three parameters, the degree probability distribution exponent, the community size probability distribution exponent, and the mixing parameter, we could have a fast or slow phase transition. Besides, in some cases, we could have communities which are synchronized inside but not with other communities and vice versa. We also show that there is a limit in these mesoscopic structures which suppresses the transition from the second-order phase transition and results in explosive synchronization. This could be considered as a tuning parameter changing the transition of the system from the second order to the first order.

Unfolding the Complexity of the Global Value Chain: Strength and Entropy in the Single-Layer, Multiplex, and Multi-Layer International Trade Networks.

The worldwide trade network has been widely studied through different data sets and network representations with a view to better understanding interactions among countries and products. Here we investigate international trade through the lenses of the single-layer, multiplex, and multi-layer networks. We discuss differences among the three network frameworks in terms of their relative advantages in capturing salient topological features of trade. We draw on the World Input-Output Database to build the three networks. We then uncover sources of heterogeneity in the way strength is allocated among countries and transactions by computing the strength distribution and entropy in each network. Additionally, we trace how entropy evolved, and show how the observed peaks can be associated with the onset of the global economic downturn. Findings suggest how more complex representations of trade, such as the multi-layer network, enable us to disambiguate the distinct roles of intra- and cross-industry transactions in driving the evolution of entropy at a more aggregate level. We discuss our results and the implications of our comparative analysis of networks for research on international trade and other empirical domains across the natural and social sciences.

Gingerol fraction from Zingiber officinale protects against gentamicin-induced nephrotoxicity.

Nephrotoxicity is the main complication of gentamicin (GM) treatment. GM induces renal damage by overproduction of reactive oxygen species and inflammation in proximal tubular cells. Phenolic compounds from ginger, called gingerols, have been demonstrated to have antioxidant and anti-inflammatory effects. We investigated if oral treatment with an enriched solution of gingerols (GF) would promote a nephroprotective effect in an animal nephropathy model. The following six groups of male Wistar rats were studied: (i) control group (CT group); (ii) gingerol solution control group (GF group); (iii) gentamicin treatment group (GM group), receiving 100 mg/kg of body weight intraperitoneally (i.p.); and (iv to vi) gentamicin groups also receiving GF, at doses of 6.25, 12.5, and 25 mg/kg, respectively (GM+GF groups). Animals from the GM group had a significant decrease in creatinine clearance and higher levels of urinary protein excretion. This was associated with markers of oxidative stress and nitric oxide production. Also, there were increases of the mRNA levels for proinflammatory cytokines (tumor necrosis factor alpha [TNF-α], interleukin-1ß [IL-1ß], IL-2, and gamma interferon [IFN-γ]). Histopathological findings of tubular degeneration and inflammatory cell infiltration reinforced GM-induced nephrotoxicity. All these alterations were attenuated by previous oral treatment with GF. Animals from the GM+GF groups showed amelioration in renal function parameters and reduced lipid peroxidation and nitrosative stress, in addition to an increment in the levels of glutathione (GSH) and superoxide dismutase (SOD) activity. Gingerols also promoted significant reductions in mRNA transcription for TNF-α, IL-2, and IFN-γ. These effects were dose dependent. These results demonstrate that GF promotes a nephroprotective effect on GM-mediated nephropathy by oxidative stress, inflammatory processes, and renal dysfunction.

Cluster explosive synchronization in complex networks.

The emergence of explosive synchronization has been reported as an abrupt transition in complex networks of first-order Kuramoto oscillators. In this Letter we demonstrate that the nodes in a second-order Kuramoto model perform a cascade of transitions toward a synchronous macroscopic state, which is a novel phenomenon that we call cluster explosive synchronization. We provide a rigorous analytical treatment using a mean-field analysis in uncorrelated networks. Our findings are in good agreement with numerical simulations and fundamentally deepen the understanding of microscopic mechanisms toward synchronization.

Analysis of functional connectivity using machine learning and deep learning in different data modalities from individuals with schizophrenia.

Objective. Schizophrenia(SCZ) is a severe mental disorder associated with persistent or recurrent psychosis, hallucinations, delusions, and thought disorders that affect approximately 26 million people worldwide, according to the World Health Organization. Several studies encompass machine learning (ML) and deep learning algorithms to automate the diagnosis of this mental disorder. Others study SCZ brain networks to get new insights into the dynamics of information processing in individuals suffering from the condition. In this paper, we offer a rigorous approach with ML and deep learning techniques for evaluating connectivity matrices and measures of complex networks to establish an automated diagnosis and comprehend the topology and dynamics of brain networks in SCZ individuals.Approach.For this purpose, we employed an functional magnetic resonance imaging (fMRI) and electroencephalogram (EEG) dataset. In addition, we combined EEG measures, i.e. Hjorth mobility and complexity, with complex network measurements to be analyzed in our model for the first time in the literature.Main results.When comparing the SCZ group to the control group, we found a high positive correlation between the left superior parietal lobe and the left motor cortex and a positive correlation between the left dorsal posterior cingulate cortex and the left primary motor. Regarding complex network measures, the diameter, which corresponds to the longest shortest path length in a network, may be regarded as a biomarker because it is the most crucial measure in different data modalities. Furthermore, the SCZ brain networks exhibit less segregation and a lower distribution of information. As a result, EEG measures outperformed complex networks in capturing the brain alterations associated with SCZ.Significance. Our model achieved an area under receiver operating characteristic curve (AUC) of 100% and an accuracy of 98.5% for the fMRI, an AUC of 95%, and an accuracy of 95.4% for the EEG data set. These are excellent classification results. Furthermore, we investigated the impact of specific brain connections and network measures on these results, which helped us better describe changes in the diseased brain.

Diagnosis of autism spectrum disorder based on functional brain networks and machine learning.

Autism is a multifaceted neurodevelopmental condition whose accurate diagnosis may be challenging because the associated symptoms and severity vary considerably. The wrong diagnosis can affect families and the educational system, raising the risk of depression, eating disorders, and self-harm. Recently, many works have proposed new methods for the diagnosis of autism based on machine learning and brain data. However, these works focus on only one pairwise statistical metric, ignoring the brain network organization. In this paper, we propose a method for the automatic diagnosis of autism based on functional brain imaging data recorded from 500 subjects, where 242 present autism spectrum disorder considering the regions of interest throughout Bootstrap Analysis of Stable Cluster map. Our method can distinguish the control group from autism spectrum disorder patients with high accuracy. Indeed the best performance provides an AUC near 1.0, which is higher than that found in the literature. We verify that the left ventral posterior cingulate cortex region is less connected to an area in the cerebellum of patients with this neurodevelopment disorder, which agrees with previous studies. The functional brain networks of autism spectrum disorder patients show more segregation, less distribution of information across the network, and less connectivity compared to the control cases. Our workflow provides medical interpretability and can be used on other fMRI and EEG data, including small data sets.

[8] and [10]-Gingerol reduces urothelial damage in ifosfamide-induced hemorrhagic cystitis via JAK/STAT/FOXO signaling pathway via IL-10.

Acrolein is the main toxic metabolite of ifosfamide (IFO) that causes urothelial damage by oxidative stress and inflammation. Here, we investigate the molecular mechanism of action of gingerols, Zingiber officinale bioactive molecules, as an alternative treatment for ifosfamide-induced hemorrhagic cystitis. Female Swiss mice were randomly divided into 5 groups: control; IFO; IFO + Mesna; and IFO + [8]- or [10]-gingerol. Mesna (80 mg/kg, i.p.) was given 5 min before, 4 and 8 h after IFO (400mg/kg, i.p.). Gingerols (25 mg/kg, p.o.) were given 1 h before and 4 and 8 h after IFO. Animals were euthanized 12 h after IFO injection. Bladders were submitted to macroscopic and histological evaluation. Oxidative stress and inflammation were assessed by malondialdehyde (MDA) or myeloperoxidase assays, respectively. mRNA gene expression was performed to evaluate mesna and gingerols mechanisms of action. Mesna was able to protect bladder tissue by activating NF-κB and NrF2 pathways. However, we demonstrated that gingerols acted as an antioxidant and anti-inflammatory agent stimulating the expression of IL-10, which intracellularly activates JAK/STAT/FOXO signaling pathway.

The structure and resilience of financial market networks.

Financial markets can be viewed as a highly complex evolving system that is very sensitive to economic instabilities. The complex organization of the market can be represented in a suitable fashion in terms of complex networks, which can be constructed from stock prices such that each pair of stocks is connected by a weighted edge that encodes the distance between them. In this work, we propose an approach to analyze the topological and dynamic evolution of financial networks based on the stock correlation matrices. An entropy-related measurement is adopted to quantify the robustness of the evolving financial market organization. It is verified that the network topological organization suffers strong variation during financial instabilities and the networks in such periods become less robust. A statistical robust regression model is proposed to quantity the relationship between the network structure and resilience. The obtained coefficients of such model indicate that the average shortest path length is the measurement most related to network resilience coefficient. This result indicates that a collective behavior is observed between stocks during financial crisis. More specifically, stocks tend to synchronize their price evolution, leading to a high correlation between pair of stock prices, which contributes to the increase in distance between them and, consequently, decrease the network resilience.

From subcritical behavior to a correlation-induced transition in rumor models.

Rumors and information spreading emerge naturally from human-to-human interactions and have a growing impact on our everyday life due to increasing and faster access to information, whether trustworthy or not. A popular mathematical model for spreading rumors, data, or news is the Maki-Thompson model. Mean-field approximations suggested that this model does not have a phase transition, with rumors always reaching a fraction of the population. Conversely, here, we show that a continuous phase transition is present in this model. Moreover, we explore a modified version of the Maki-Thompson model that includes a forgetting mechanism, changing the Markov chain's nature and allowing us to use a plethora of analytic and numeric methods. Particularly, we characterize the subcritical behavior, where the lifespan of a rumor increases as the spreading rate drops, following a power-law relationship. Our findings show that the dynamic behavior of rumor models is much richer than shown in previous investigations.

The rise and fall of countries in the global value chains.

Countries become global leaders by controlling international and domestic transactions connecting geographically dispersed production stages. We model global trade as a multi-layer network and study its power structure by investigating the tendency of eigenvector centrality to concentrate on a small fraction of countries, a phenomenon called localization transition. We show that the market underwent a significant drop in power concentration precisely in 2007 just before the global financial crisis. That year marked an inflection point at which new winners and losers emerged and a remarkable reversal of leading role took place between the two major economies, the US and China. We uncover the hierarchical structure of global trade and the contribution of individual industries to variations in countries' economic dominance. We also examine the crucial role that domestic trade played in leading China to overtake the US as the world's dominant trading nation. There is an important lesson that countries can draw on how to turn early signals of upcoming downturns into opportunities for growth. Our study shows that, despite the hardships they inflict, shocks to the economy can also be seen as strategic windows countries can seize to become leading nations and leapfrog other economies in a changing geopolitical landscape.

Nonuniform random graphs on the plane: A scaling study.

We consider random geometric graphs on the plane characterized by a nonuniform density of vertices. In particular, we introduce a graph model where n vertices are independently distributed in the unit disk with positions, in polar coordinates (l,θ), obeying the probability density functions ρ(l) and ρ(θ). Here we choose ρ(l) as a normal distribution with zero mean and variance σ∈(0,∞) and ρ(θ) as a uniform distribution in the interval θ∈[0,2π). Then, two vertices are connected by an edge if their Euclidean distance is less than or equal to the connection radius ℓ. We characterize the topological properties of this random graph model, which depends on the parameter set (n,σ,ℓ), by the use of the average degree 〈k〉 and the number of nonisolated vertices V_{×}, while we approach their spectral properties with two measures on the graph adjacency matrix: the ratio of consecutive eigenvalue spacings r and the Shannon entropy S of eigenvectors. First we propose a heuristic expression for 〈k(n,σ,ℓ)〉. Then, we look for the scaling properties of the normalized average measure 〈X[over ¯]〉 (where X stands for V_{×}, r, and S) over graph ensembles. We demonstrate that the scaling parameter of 〈V_{×}[over ¯]〉=〈V_{×}〉/n is indeed 〈k〉, with 〈V_{×}[over ¯]〉≈1-exp(-〈k〉). Meanwhile, the scaling parameter of both 〈r[over ¯]〉 and 〈S[over ¯]〉 is proportional to n^{-γ}〈k〉 with γ≈0.16.

Comparison of theoretical approaches for epidemic processes with waning immunity in complex networks.

The role of waning immunity in basic epidemic models on networks has been undervalued while being noticeably fundamental for real epidemic outbreaks. One central question is which mean-field approach is more accurate in describing the epidemic dynamics. We tackled this problem considering the susceptible-infected-recovered-susceptible (SIRS) epidemic model on networks. Two pairwise mean-field theories, one based on recurrent dynamical message-passing (rDMP) theory and the other on the pair quenched mean-field (PQMF) theory, are compared with extensive stochastic simulations on large networks of different levels of heterogeneity. For waning immunity times longer than or comparable with the recovering time, rDMP outperforms PQMF theory on power-law networks with degree distribution P(k)∼k^{-γ}. In particular, for γ>3, the epidemic threshold observed in simulations is finite, in qualitative agreement with rDMP, while PQMF leads to an asymptotically null threshold. The critical epidemic prevalence for γ>3 is localized in a finite set of vertices in the case of the PQMF theory. In contrast, the localization happens in a subextensive fraction of the network in rDMP theory. Simulations, however, indicate that localization patterns of the actual epidemic lay between the two mean-field theories, and improved theoretical approaches are necessary to understanding the SIRS dynamics.

Application of machine learning and complex network measures to an EEG dataset from ayahuasca experiments.

Ayahuasca is a blend of Amazonian plants that has been used for traditional medicine by the inhabitants of this region for hundreds of years. Furthermore, this plant has been demonstrated to be a viable therapy for a variety of neurological and mental diseases. EEG experiments have found specific brain regions that changed significantly due to ayahuasca. Here, we used an EEG dataset to investigate the ability to automatically detect changes in brain activity using machine learning and complex networks. Machine learning was applied at three different levels of data abstraction: (A) the raw EEG time series, (B) the correlation of the EEG time series, and (C) the complex network measures calculated from (B). Further, at the abstraction level of (C), we developed new measures of complex networks relating to community detection. As a result, the machine learning method was able to automatically detect changes in brain activity, with case (B) showing the highest accuracy (92%), followed by (A) (88%) and (C) (83%), indicating that connectivity changes between brain regions are more important for the detection of ayahuasca. The most activated areas were the frontal and temporal lobe, which is consistent with the literature. F3 and PO4 were the most important brain connections, a significant new discovery for psychedelic literature. This connection may point to a cognitive process akin to face recognition in individuals during ayahuasca-mediated visual hallucinations. Furthermore, closeness centrality and assortativity were the most important complex network measures. These two measures are also associated with diseases such as Alzheimer's disease, indicating a possible therapeutic mechanism. Moreover, the new measures were crucial to the predictive model and suggested larger brain communities associated with the use of ayahuasca. This suggests that the dissemination of information in functional brain networks is slower when this drug is present. Overall, our methodology was able to automatically detect changes in brain activity during ayahuasca consumption and interpret how these psychedelics alter brain networks, as well as provide insights into their mechanisms of action.

NeoBat Interactions: A data set of bat-plant interactions in the Neotropics.

Data papers and open databases have revolutionized contemporary science, as they provide the long-needed incentive to collaborate in large international teams and make natural history information widely available. Nevertheless, most data papers have focused on species occurrence or abundance, whereas interactions have received much less attention. To help fill this gap, we have compiled a georeferenced data set of interactions between 93 bat species of the family Phyllostomidae (Chiroptera) and 501 plant species of 68 families. Data came from 169 studies published between 1957 and 2007 covering the entire Neotropical Region, with most records from Brazil (34.5% of all study sites), Costa Rica (16%), and Mexico (14%). Our data set includes 2571 records of frugivory (75.1% of all records) and nectarivory (24.9%). The best represented bat genera are Artibeus (28% of all records), Carollia (24%), Sturnira (10.1%), and Glossophaga (8.8%). Carollia perspicillata (187), Artibeus lituratus (125), Artibeus jamaicensis (94), Glossophaga soricina (86), and Artibeus planirostris (74) were the bat species with the broadest diets recorded based on the number of plant species. Among the plants, the best represented families were Moraceae (17%), Piperaceae (15.4%), Urticaceae (9.2%), and Solanaceae (9%). Plants of the genera Cecropia (46), Ficus (42), Piper (40), Solanum (31), and Vismia (27) exhibited the largest number of interactions. These data are stored as arrays (records, sites, and studies) organized by logical keys and rich metadata, which helped to compile the information on different ecological and geographic scales, according to how they should be used. Our data set on bat-plant interactions is by far the most extensive, both in geographic and taxonomic terms, and includes abiotic information of study sites, as well as ecological information of plants and bats. It has already facilitated several studies and we hope it will stimulate novel analyses and syntheses, in addition to pointing out important gaps in knowledge. Data are provided under the Creative Commons Attribution 4.0 International License. Please cite this paper when the data are used in any kind of publication related to research, outreach, and teaching activities.

Publisher's Note: Entropy of weighted recurrence plots [Phys. Rev. E 90, 042919 (2014)].

This corrects the article DOI: 10.1103/PhysRevE.90.042919.

Estimating complex cortical networks via surface recordings- a critical note.

We discuss potential caveats when estimating topologies of 3D brain networks from surface recordings. It is virtually impossible to record activity from all single neurons in the brain and one has to rely on techniques that measure average activity at sparsely located (non-invasive) recording sites. Effects of this spatial sampling in relation to structural network measures like centrality and assortativity were analyzed using multivariate classifiers. A simplified model of 3D brain connectivity incorporating both short- and long-range connections served for testing. To mimic M/EEG recordings we sampled this model via non-overlapping regions and weighted nodes and connections according to their proximity to the recording sites. We used various complex network models for reference and tried to classify sampled versions of the "brain-like" network as one of these archetypes. It was found that sampled networks may substantially deviate in topology from the respective original networks for small sample sizes. For experimental studies this may imply that surface recordings can yield network structures that might not agree with its generating 3D network.