Inequality in economic shock exposures across the global firm-level supply network.

For centuries, national economies have been engaging in international trade and production. The resulting international supply networks not only increase wealth for countries, but also allow for economic shocks to propagate across borders. Using global, firm-level supply network data, we estimate a country's exposure to direct and indirect economic losses caused by the failure of a company in another country. We show the network of international systemic risk-flows. We find that rich countries expose poor countries stronger to systemic risk than vice-versa. The risk is highly concentrated, however, higher risk levels are not compensated with a risk premium in GDP levels, nor higher GDP growth. Our findings put the often praised benefits for developing countries from globalized production in a new light, by relating them to risks involved in the production processes. Exposure risks present a new dimension of global inequality that most affects the poor in supply shock crises.

Quantifying the impact of homophily and influencer networks on song popularity prediction.

Forecasting the popularity of new songs has become a standard practice in the music industry and provides a comparative advantage for those that do it well. Considerable efforts were put into machine learning prediction models for that purpose. It is known that in these models, relevant predictive parameters include intrinsic lyrical and acoustic characteristics, extrinsic factors (e.g., publisher influence and support), and the previous popularity of the artists. Much less attention was given to the social components of the spreading of song popularity. Recently, evidence for musical homophily-the tendency that people who are socially linked also share musical tastes-was reported. Here we determine how musical homophily can be used to predict song popularity. The study is based on an extensive dataset from the last.fm online music platform from which we can extract social links between listeners and their listening patterns. To quantify the importance of networks in the spreading of songs that eventually determines their popularity, we use musical homophily to design a predictive influence parameter and show that its inclusion in state-of-the-art machine learning models enhances predictions of song popularity. The influence parameter improves the prediction precision (TP/(TP + FP)) by about 50% from 0.14 to 0.21, indicating that the social component in the spreading of music plays at least as significant a role as the artist's popularity or the impact of the genre.

Forecasting the Early Impact of COVID-19 on Physician Supply in EU Countries.

BACKGROUND: Many countries faced health workforce challenges even before the pandemic, such as impending retirements, negative population growth, or sub-optimal allocation of resources across health sectors. Current quantitative models are often of limited use, either because they require extensive individual-level data to be properly calibrated, or (in the absence of such data) because they are too simplistic to capture important demographic changes or disruptive epidemiological shocks such as the SARS-CoV-2 pandemic. Method: We propose a population-dynamic and stock-flow-consistent approach to physician supply forecasting that is complex enough to account for dynamically changing behaviour, while requiring only publicly available time-series data for full calibration. We demonstrate the utility of this model by applying it to 21 European countries to forecast the supply of generalist and specialist physicians to 2040, and the impact of increased health care utilisation due to Covid on this supply. RESULTS: Compared with the workforce needed to maintain physician density at 2019 levels, we find that in many countries there is indeed a significant trend towards decreasing generalist density at the expense of increasing specialist density. The trends for specialists are exacerbated by expectations of negative population growth in many Southern and Eastern European countries. Compared to the expected demographic changes in the population and the health workforce, we expect a limited impact of Covid on these trends, even under conservative modelling assumptions. Finally, we generalise the approach to a multi-professional, multi-regional and multi-sectoral model for Austria, where we find an additional suboptimal distribution in the supply of contracted versus non-contracted (private) physicians. CONCLUSION: It is therefore vital to develop tools for decision-makers to influence the allocation and supply of doctors across specialties and sectors to address these imbalances.

Estimating the loss of economic predictability from aggregating firm-level production networks.

To estimate the reaction of economies to political interventions or external disturbances, input-output (IO) tables-constructed by aggregating data into industrial sectors-are extensively used. However, economic growth, robustness, and resilience crucially depend on the detailed structure of nonaggregated firm-level production networks (FPNs). Due to nonavailability of data, little is known about how much aggregated sector-based and detailed firm-level-based model predictions differ. Using a nearly complete nationwide FPN, containing 243,399 Hungarian firms with 1,104,141 supplier-buyer relations, we self-consistently compare production losses on the aggregated industry-level production network (IPN) and the granular FPN. For this, we model the propagation of shocks of the same size on both, the IPN and FPN, where the latter captures relevant heterogeneities within industries. In a COVID-19 inspired scenario, we model the shock based on detailed firm-level data during the early pandemic. We find that using IPNs instead of FPNs leads to an underestimation of economic losses of up to 37%, demonstrating a natural limitation of industry-level IO models in predicting economic outcomes. We ascribe the large discrepancy to the significant heterogeneity of firms within industries: we find that firms within one sector only sell 23.5% to and buy 19.3% from the same industries on average, emphasizing the strong limitations of industrial sectors for representing the firms they include. Similar error levels are expected when estimating economic growth, CO2 emissions, and the impact of policy interventions with industry-level IO models. Granular data are key for reasonable predictions of dynamical economic systems.

Unraveling cradle-to-grave disease trajectories from multilayer comorbidity networks.

We aim to comprehensively identify typical life-spanning trajectories and critical events that impact patients' hospital utilization and mortality. We use a unique dataset containing 44 million records of almost all inpatient stays from 2003 to 2014 in Austria to investigate disease trajectories. We develop a new, multilayer disease network approach to quantitatively analyze how cooccurrences of two or more diagnoses form and evolve over the life course of patients. Nodes represent diagnoses in age groups of ten years; each age group makes up a layer of the comorbidity multilayer network. Inter-layer links encode a significant correlation between diagnoses (p < 0.001, relative risk > 1.5), while intra-layers links encode correlations between diagnoses across different age groups. We use an unsupervised clustering algorithm for detecting typical disease trajectories as overlapping clusters in the multilayer comorbidity network. We identify critical events in a patient's career as points where initially overlapping trajectories start to diverge towards different states. We identified 1260 distinct disease trajectories (618 for females, 642 for males) that on average contain 9 (IQR 2-6) different diagnoses that cover over up to 70 years (mean 23 years). We found 70 pairs of diverging trajectories that share some diagnoses at younger ages but develop into markedly different groups of diagnoses at older ages. The disease trajectory framework can help us to identify critical events as specific combinations of risk factors that put patients at high risk for different diagnoses decades later. Our findings enable a data-driven integration of personalized life-course perspectives into clinical decision-making.

New Forms of Collaboration Between the Social and Natural Sciences Could Become Necessary for Understanding Rapid Collective Transitions in Social Systems.

Human societies are complex systems and as such have tipping points. They can rapidly transit from one mode of operation to another and thereby change the way they function as a whole. Such transitions appear as financial or economic crises, rapid swings in collective opinion, political regime shifts, or revolutions. In physics collective transitions are known as phase transitions; for example, water exists in states of liquid, ice, and vapor. A few variables determine which state is realized: temperature, pressure, and volume. For social systems it is less clear what determines collective social states. A better understanding of social tipping points would allow us to tackle some of the big challenges more systematically, such as polarization, loss of social cohesion, fragmentation, or the green transition. The physics concept of universality might be key to understanding some tipping points in human societies and why agent-based models (ABMs) might make sense for identifying the transition points. If universality exists in social systems there is hope that relatively simple ABMs will be sufficient for understanding collective social systems in transition; if it does not exist, highly detailed computational models will be unavoidable. Both are possible. Both need new forms of collaboration between the social and natural sciences, and new types of data will be essential.

Forensic analysis of the Turkey 2023 presidential election reveals extreme vote swings in remote areas.

Concerns about the integrity of Turkey's elections have increased with the recent transition from a parliamentary democracy to an executive presidency under Recep Tayyip Erdogan. Election forensics tools are used to identify statistical traces of certain types of electoral fraud, providing important information about the integrity and validity of democratic elections. Such analyses of the 2017 and 2018 Turkish elections revealed that malpractices such as ballot stuffing or voter manipulation may indeed have played a significant role in determining the election results. Here, we apply election forensic statistical tests for ballot stuffing and voter manipulation to the results of the 2023 presidential election in Turkey. We find that both rounds of the 2023 presidential election exhibit similar statistical irregularities to those observed in the 2018 presidential election, however the magnitude of these distortions has decreased. We estimate that 2.4% (SD 1.9%) and 1.9% (SD 1.7%) of electoral units may have been affected by ballot stuffing practices in favour of Erdogan in the first and second rounds, respectively, compared to 8.5% (SD 3.9%) in 2018. Areas with smaller polling stations and fewer ballot boxes had significantly inflated votes and turnout, again, in favor of Erdogan. Furthermore, electoral districts with two or fewer ballot boxes were more likely to show large swings in vote shares in favour of Erdogan from the first to the second round. Based on a statistical model, it is estimated that these shifts account for 342,000 additional ballots (SD 4,900) or 0.64% for Erdogan, which is lower than the 4.36% margin by which Erdogan was victorious. Our results suggest that Turkish elections continue to be riddled with statistical irregularities, that may be indicative of electoral fraud.

Building an alliance to map global supply networks.

New firm-level data can inform policy-making.

A large-scale empirical investigation of specialization in criminal career.

We use a comprehensive longitudinal dataset on criminal acts over 6 years in a European country to study specialization in criminal careers. We present a method to cluster crime categories by their relative co-occurrence within criminal careers, deriving a natural, data-based taxonomy of criminal specialization. Defining specialists as active criminals who stay within one category of offending behavior, we study their socio-demographic attributes, geographic range, and positions in their collaboration networks relative to their generalist counterparts. Compared to generalists, specialists tend to be older, are more likely to be women, operate within a smaller geographic range, and collaborate in smaller, more tightly-knit local networks. We observe that specialists are more intensely embedded in criminal networks, suggesting a potential source of self-reinforcing dynamics in criminal careers.

Equivalence of information production and generalised entropies in complex processes.

Complex systems with strong correlations and fat-tailed distribution functions have been argued to be incompatible with the Boltzmann-Gibbs entropy framework and alternatives, so-called generalised entropies, were proposed and studied. Here we show, that this perceived incompatibility is actually a misconception. For a broad class of processes, Boltzmann entropy -the log multiplicity- remains the valid entropy concept. However, for non-i.i.d. processes, Boltzmann entropy is not of Shannon form, -k∑ipi log pi, but takes the shape of generalised entropies. We derive this result for all processes that can be asymptotically mapped to adjoint representations reversibly where processes are i.i.d. In these representations the information production is given by the Shannon entropy. Over the original sampling space this yields functionals identical to generalised entropies. The problem of constructing adequate context-sensitive entropy functionals therefore can be translated into the much simpler problem of finding adjoint representations. The method provides a comprehensive framework for a statistical physics of strongly correlated systems and complex processes.

Diabetes incidence in Austria: The role of famines on diabetes and related NCDs.

Undernutrition in early life associates with increased risk for type 2 diabetes in later life. Whether similar associations hold for other diseases remains unclear. We aim to quantify how perinatal exposure to famines relates to the risk of becoming incident with type 2 diabetes in later life. Using population-wide medical claims data for Austrians aged >50y, yearly diabetes incidence was measured in an epidemiological progression model. We find incidence rates that increase from 2013 to 2017 and observe two famine-related birth cohorts of 5,887 patients with incidence rate increases for diabetes of up to 78% for males and 59% for females compared to cohorts born two years earlier. These cohorts show increased risks for multiple other diagnoses as well. Public health efforts to decrease diabetes must not only focus on lifestyle factors but also emphasize the importance of reproductive health and adequate nutrition during pregnancy and early postnatal life.

Shock propagation from the Russia-Ukraine conflict on international multilayer food production network determines global food availability.

Dependencies in the global food production network can lead to shortages in numerous regions, as demonstrated by the impacts of the Russia-Ukraine conflict on global food supplies. Here we reveal the losses of 125 food products after a localized shock to agricultural production in 192 countries and territories using a multilayer network model of trade (direct) and conversion of food products (indirect), thereby quantifying 108 shock transmissions. We find that a complete agricultural production loss in Ukraine has heterogeneous impacts on other countries, causing relative losses of up to 89% in sunflower oil and 85% in maize via direct effects and up to 25% in poultry meat via indirect impacts. Whereas previous studies often treated products in isolation and did not account for product conversion during production, the present model considers the global propagation of local supply shocks along both production and trade relations, allowing for a comparison of different response strategies.

Obesity as pleiotropic risk state for metabolic and mental health throughout life.

Obesity, a highly prevalent disorder and central diagnosis of the metabolic syndrome, is linked to mental health by clinical observations and biological pathways. Patients with a diagnosis of obesity may show long-lasting increases in risk for receiving psychiatric co-diagnoses. Austrian national registry data of inpatient services from 1997 to 2014 were analyzed to detect associations between a hospital diagnosis of obesity (ICD-10: E66) and disorders grouped by level-3 ICD-10 codes. Data were stratified by age decades and associations between each pair of diagnoses were computed with the Cochran-Mantel-Haenszel method, providing odds ratios (OR) and p values corrected for multiple testing. Further, directions of the associations were assessed by calculating time-order-ratios. Receiving a diagnosis of obesity significantly increased the odds for a large spectrum of psychiatric disorders across all age groups, including depression, psychosis-spectrum, anxiety, eating and personality disorders (all pcorr < 0.01, all OR > 1.5). For all co-diagnoses except for psychosis-spectrum, obesity was significantly more often the diagnosis received first. Further, significant sex differences were found for most disorders, with women showing increased risk for all disorders except schizophrenia and nicotine addiction. In addition to the well-recognized role in promoting disorders related to the metabolic syndrome and severe cardiometabolic sequalae, obesity commonly precedes severe mental health disorders. Risk is most pronounced in young age groups and particularly increased in female patients. Consequently, thorough screening for mental health problems in patients with obesity is urgently called for to allow prevention and facilitate adequate treatment.

Trends in diabetes incidence in Austria 2013-2017.

This study aims to quantify whether age and sex groups in Austrian regions are equally affected by the rise of type 2 diabetes. Population-wide medical claims data was obtained for citizens in Austria aged above 50 year, who received antihyperglycemic treatments or underwent HbA1c monitoring between 2012 and 2017. Diabetes incidence was measured using an epidemiological diabetes progression model accounting for patients who discontinued antihyperglycemic therapy; the erratic group. Out of 746,184 patients, 268,680 (140,960 females) discontinued their treatment and/or monitoring for at least one year. Without adjusting for such erratic patients, incidence rates increase from 2013 to 2017 (females: from 0·5% to 1·1%, males: 0·5% to 1·2%), whereas they decrease in all groups after adjustments (females: - 0·3% to - 0·5%, males: - 0·4% to - 0·5%). Higher mortality was observed in the erratic group compared to patients on continued antihyperglycemic therapy (mean difference 12% and 14% for females and males, respectively). In summary, incidence strongly depends on age, sex and place of residency. One out of three patients with diabetes in Austria discontinued antihyperglycemic treatment or glycemic monitoring for at least one year. This newly identified subgroup raises concern regarding adherence and continuous monitoring of diabetes care and demands further evaluation.

Systematic population-wide ecological analysis of regional variability in disease prevalence.

The prevalence of diseases often varies substantially from region to region. Besides basic demographic properties, the factors that drive the variability of each prevalence are to a large extent unknown. Here we show how regional prevalence variations in 115 different diseases relate to demographic, socio-economic, environmental factors and migratory background, as well as access to different types of health services such as primary, specialized and hospital healthcare. We have collected regional data for these risk factors at different levels of resolution; from large regions of care (Versorgungsregion) down to a 250 by 250 m square grid. Using multivariate regression analysis, we quantify the explanatory power of each independent variable in relation to the regional variation of the disease prevalence. We find that for certain diseases, such as acute heart conditions, diseases of the inner ear, mental and behavioral disorders due to substance abuse, up to 80% of the variance can be explained with these risk factors. For other diagnostic blocks, such as blood related diseases, injuries and poisoning however, the explanatory power is close to zero. We find that the time needed to travel from the inhabited center to the relevant hospital ward often contributes significantly to the disease risk, in particular for diabetes mellitus. Our results show that variations in disease burden across different regions can for many diseases be related to variations in demographic and socio-economic factors. Furthermore, our results highlight the relative importance of access to health care facilities in the treatment of chronic diseases like diabetes.

Atrial fibrillation and cancer: prevalence and relative risk from a nationwide study.

Background: Atrial fibrillation (AF) is an increasingly recognized codiagnosis in patients with cancer. Objectives: This study aimed to provide a robust and contemporary estimate on the coprevalence and relative risk of AF in patients with cancer. Methods: We conducted a nationwide analysis, utilizing diagnosis codes from the Austrian Association of Social Security Providers dataset. Estimates of the coprevalence of cancer and AF and the relative risk of AF in patients with cancer compared with individuals without cancer were obtained as point prevalences with binomial exact confidence intervals and summarized across age groups and cancer types with random-effects models. Results: Overall, 8,306,244 persons were included in the present analysis, of whom 158,675 (prevalence estimate, 1.91%; 95% CI, 1.90-1.92) had a cancer diagnosis code and 112,827 (1.36%; 95% CI, 1.35-1.36) an AF diagnosis code, respectively. The prevalence estimate for AF in patients with cancer was 9.77% (95% CI, 9.63-9.92) and 1.19% (95% CI, 1.19-1.20) in the noncancer population. Conversely, 13.74% (95% CI, 13.54-13.94) of patients with AF had a concurrent cancer diagnosis. The corresponding age-stratified random-effects relative risk ratio for AF in patients with cancer compared with no cancer diagnosis was 10.45 (95% CI, 7.47-14.62). The strongest associations between cancer and AF were observed in younger persons and patients with hematologic malignancies. Conclusion: Cancer and AF have a substantial coprevalence in the population. This finding corroborates the concept that cancer and AF have common risk factors and pathophysiology.

Homophily-Based Social Group Formation in a Spin Glass Self-Assembly Framework.

Homophily, the tendency of humans to attract each other when sharing similar features, traits, or opinions, has been identified as one of the main driving forces behind the formation of structured societies. Here we ask to what extent homophily can explain the formation of social groups, particularly their size distribution. We propose a spin-glass-inspired framework of self-assembly, where opinions are represented as multidimensional spins that dynamically self-assemble into groups; individuals within a group tend to share similar opinions (intragroup homophily), and opinions between individuals belonging to different groups tend to be different (intergroup heterophily). We compute the associated nontrivial phase diagram by solving a self-consistency equation for "magnetization" (combined average opinion). Below a critical temperature, there exist two stable phases: one ordered with nonzero magnetization and large clusters, the other disordered with zero magnetization and no clusters. The system exhibits a first-order transition to the disordered phase. We analytically derive the group-size distribution that successfully matches empirical group-size distributions from online communities.

Simvastatin therapy in higher dosages deteriorates bone quality: Consistent evidence from population-wide patient data and interventional mouse studies.

BACKGROUND: Combining mouse experiments with big data analysis of the Austrian population, we investigated the association between high-dose statin treatment and bone quality. METHODS: The bone microarchitecture of the femur and vertebral body L4 was measured in male and ovariectomized female mice on a high-fat diet containing simvastatin (1.2 g/kg). A sex-specific matched big data analysis of Austrian health insurance claims using multiple logistic regression models was conducted (simvastatin 60-80 mg/day vs. controls; males: n = 138,666; females: n = 155,055). RESULTS: High-dose simvastatin impaired bone quality in male and ovariectomized mice. In the trabecular femur, simvastatin reduced bone volume (µm3: ♂, 213 ± 15 vs. 131 ± 7, p < 0.0001; ♀, 66 ± 7 vs. 44 ± 5, p = 0.02) and trabecular number (1/mm: ♂, 1.88 ± 0.09 vs. 1.27 ± 0.06, p < 0.0001; ♀, 0.60 ± 0.05 vs. 0.43 ± 0.04, p = 0.01). In the cortical femur, bone volume (mm3: ♂, 1.44 ± 0.03 vs. 1.34 ± 0.03, p = 0.009; ♀, 1.33 ± 0.03 vs. 1.12 ± 0.03, p = 0.0002) and cortical thickness were impaired (µm: ♂, 211 ± 4 vs. 189 ± 4, p = 0.0004; ♀, 193 ± 3 vs. 169 ± 3, p < 0.0001). Similar impairments were found in vertebral body L4. Simvastatin-induced changes in weight or glucose metabolism were excluded as mediators of deteriorations in bone quality. Results from mice were supported by a matched cohort analysis showing an association between high-dose simvastatin and increased risk of osteoporosis in patients (♂, OR: 5.91, CI: 3.17-10.99, p < 0.001; ♀, OR: 4.16, CI: 2.92-5.92, p < 0.001). CONCLUSION: High-dose simvastatin dramatically reduces bone quality in obese male and ovariectomized female mice, suggesting that direct drug action accounts for the association between high dosage and increased risk of osteoporosis as observed in comparable human cohorts. The underlying pathophysiological mechanisms behind this relationship are presently unknown and require further investigation.

Supporting COVID-19 policy-making with a predictive epidemiological multi-model warning system.

BACKGROUND: In response to the SARS-CoV-2 pandemic, the Austrian governmental crisis unit commissioned a forecast consortium with regularly projections of case numbers and demand for hospital beds. The goal was to assess how likely Austrian ICUs would become overburdened with COVID-19 patients in the upcoming weeks. METHODS: We consolidated the output of three epidemiological models (ranging from agent-based micro simulation to parsimonious compartmental models) and published weekly short-term forecasts for the number of confirmed cases as well as estimates and upper bounds for the required hospital beds. RESULTS: We report on three key contributions by which our forecasting and reporting system has helped shaping Austria's policy to navigate the crisis, namely (i) when and where case numbers and bed occupancy are expected to peak during multiple waves, (ii) whether to ease or strengthen non-pharmaceutical intervention in response to changing incidences, and (iii) how to provide hospital managers guidance to plan health-care capacities. CONCLUSIONS: Complex mathematical epidemiological models play an important role in guiding governmental responses during pandemic crises, in particular when they are used as a monitoring system to detect epidemiological change points.

During the SARS-CoV-2 pandemic, health authorities make decisions on how and when to implement interventions such as social distancing to avoid overburdening hospitals and other parts of the healthcare system. We combined three mathematical models developed to predict the expected number of confirmed SARS-CoV-2 cases and hospitalizations over the next two weeks. This provides decision-makers and the general public with a combined forecast that is usually more accurate than any of the individual models. Our forecasting system has been used in Austria to decide when to strengthen or ease response measures.

Evolving collaboration, dependencies, and use in the Rust Open Source Software ecosystem.

Open Source Software (OSS) is widely spread in industry, research, and government. OSS represents an effective development model because it harnesses the decentralized efforts of many developers in a way that scales. As OSS developers work independently on interdependent modules, they create a larger cohesive whole in the form of an ecosystem, leaving traces of their contributions and collaborations. Data harvested from these traces enable the study of large-scale decentralized collaborative work. We present curated data on the activity of tens of thousands of developers in the Rust ecosystem and the evolving dependencies between their libraries. The data covers eight years of developer contributions to Rust libraries and can be used to reconstruct the ecosystem's development history, such as growing developer collaboration networks or dependency networks. These are complemented by data on downloads and popularity, tracking dynamics of use, visibility, and success over time. Altogether the data give a comprehensive view of several dimensions of the ecosystem.