Dataset Information

Predicting collapse of adaptive networked systems without knowing the network.

ABSTRACT: The collapse of ecosystems, the extinction of species, and the breakdown of economic and financial networks usually hinges on topological properties of the underlying networks, such as the existence of self-sustaining (or autocatalytic) feedback cycles. Such collapses can be understood as a massive change of network topology, usually accompanied by the extinction of a macroscopic fraction of nodes and links. It is often related to the breakdown of the last relevant directed catalytic cycle within a dynamical system. Without detailed structural information it seems impossible to state, whether a network is robust or if it is likely to collapse in the near future. Here we show that it is nevertheless possible to predict collapse for a large class of systems that are governed by a linear (or linearized) dynamics. To compute the corresponding early warning signal, we require only non-structural information about the nodes' states such as species abundances in ecosystems, or company revenues in economic networks. It is shown that the existence of a single directed cycle in the network can be detected by a "quantization effect" of node states, that exists as a direct consequence of a corollary of the Perron-Frobenius theorem. The proposed early warning signal for the collapse of networked systems captures their structural instability without relying on structural information. We illustrate the validity of the approach in a transparent model of co-evolutionary ecosystems and show this quantization in systems of species evolution, epidemiology, and population dynamics.

SUBMITTER: Horstmeyer L

PROVIDER: S-EPMC6985233 | biostudies-literature | 2020 Jan

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Predicting collapse of adaptive networked systems without knowing the network.

Horstmeyer Leonhard L Pham Tuan Minh TM Korbel Jan J Thurner Stefan S

Scientific reports 20200127 1

The collapse of ecosystems, the extinction of species, and the breakdown of economic and financial networks usually hinges on topological properties of the underlying networks, such as the existence of self-sustaining (or autocatalytic) feedback cycles. Such collapses can be understood as a massive change of network topology, usually accompanied by the extinction of a macroscopic fraction of nodes and links. It is often related to the breakdown of the last relevant directed catalytic cycle withi ...[more]

PMID: 31988357

Similar Datasets

Project description:BackgroundModels estimating risk for advanced proximal colorectal neoplasia (APN) may be used to select colorectal cancer (CRC) screening test, either prior to knowing distal colorectal findings or afterward. Current models have only fair discrimination and nearly all require knowing distal findings.ObjectiveDerive and test risk prediction models for APN with and without distal findings.SettingSelected endoscopy centers within central Indiana, USA.ParticipantsAverage-risk persons undergoing first-time screening colonoscopy.InterventionsDemographics, personal and family medical history, lifestyle factors and physical measures were linked to the most advanced finding in proximal and distal colorectal segments. For both models, logistic regression identified factors independently associated with APN on a derivation set. Based on equation coefficients, points were assigned to each factor, and risk for APN was examined for each score. Scores with comparable risks were collapsed into risk categories. Both models and their scoring systems were tested on the validation set.Main outcomeAPN, defined as any adenoma or sessile serrated lesion ≥1 cm, one with villous histology or high-grade dysplasia, or CRC proximal to the descending colon.ResultsAmong 3025 subjects in the derivation set (mean age 57.3 ± 6.5 years; 52% women), APN prevalence was 4.5%; 2859 (94.5%) had complete data on risk factors. Independently associated with APN were age, sex, cigarette smoking, cohabitation status, metabolic syndrome, non-steroidal anti-inflammatory drug use and physical activity. This model (without distal findings) was well-calibrated (P = 0.62) and had good discrimination (c-statistic = 0.73). In low-, intermediate- and high-risk groups that comprised 21, 58 and 21% of the sample, respectively, APN risks were 1.47% (95% CI, 0.67-2.77%), 3.09% (CI, 2.31-4.04%) and 11.6% (CI, 9.10-14.4%), respectively (P < 0.0001), with no proximal CRCs in the low-risk group and 2 in the intermediate-risk group. When tested in the validation set of 1455, the model retained good metrics (calibration P = 0.85; c-statistic = 0.83), with APN risks in low- (22%), intermediate- (56%) and high-risk (22%) subgroups of 0.62% (CI, 0.08-2.23%) 2.20% (CI, 1.31-3.46%) and 13.0% (CI, 9.50-17.2%), respectively (P < 0.0001). There were no proximal CRCs in the low-risk group, and two in the intermediate-risk group. The model with distal findings performed comparably, with validation set metrics of 0.18 for calibration, 0.76 for discrimination and APN risk (% sample) in low-, intermediate-, and high-risk groups of 1.1 (69%), 8.3 (22%) and 22.3% (9%).ConclusionThese models stratify large proportions of average-risk persons into clinically meaningful risk groups, and could improve screening efficiency, particularly for noncolonoscopy-based programs.

Project description:Establishing the invariance property of an instrument (e.g., a questionnaire or test) is a key step for establishing its measurement validity. Measurement invariance is typically assessed by differential item functioning (DIF) analysis, i.e., detecting DIF items whose response distribution depends not only on the latent trait measured by the instrument but also on the group membership. DIF analysis is confounded by the group difference in the latent trait distributions. Many DIF analyses require knowing several anchor items that are DIF-free in order to draw inferences on whether each of the rest is a DIF item, where the anchor items are used to identify the latent trait distributions. When no prior information on anchor items is available, or some anchor items are misspecified, item purification methods and regularized estimation methods can be used. The former iteratively purifies the anchor set by a stepwise model selection procedure, and the latter selects the DIF-free items by a LASSO-type regularization approach. Unfortunately, unlike the methods based on a correctly specified anchor set, these methods are not guaranteed to provide valid statistical inference (e.g., confidence intervals and p-values). In this paper, we propose a new method for DIF analysis under a multiple indicators and multiple causes (MIMIC) model for DIF. This method adopts a minimal [Formula: see text] norm condition for identifying the latent trait distributions. Without requiring prior knowledge about an anchor set, it can accurately estimate the DIF effects of individual items and further draw valid statistical inferences for quantifying the uncertainty. Specifically, the inference results allow us to control the type-I error for DIF detection, which may not be possible with item purification and regularized estimation methods. We conduct simulation studies to evaluate the performance of the proposed method and compare it with the anchor-set-based likelihood ratio test approach and the LASSO approach. The proposed method is applied to analysing the three personality scales of the Eysenck personality questionnaire-revised (EPQ-R).

Dataset Information

Predicting collapse of adaptive networked systems without knowing the network.

Publications

Predicting collapse of adaptive networked systems without knowing the network.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets