Phytochemical investigation of the root extract of Phyllanthus reticulatus Poir. (Phyllanthaceae) led to the isolation and identification of one previously undescribed ent-cleistanthane-type diterpenoid glucoside, 19hydroxycleistanthol 19-O-beta-D-glucopyranoside (1), together with ten known compounds (2-11). The structure of the new compound was established by HRESIMS, 1D and 2D NMR spectroscopy, ECD calculations, and GIAO NMR/DP4+ analysis. The known compounds were identified by comparison of their NMR data with those reported in the literature. Among these, compounds 6, 7 and 9 are reported for the first time from this species. In addition, compounds 1-10 were evaluated for their in vitro cytotoxic activities against MDA-MB-231, A549, MCF-7, and HCT-116 cell lines. As a result, phyllanthusmin C (5), an arylnaphthalene lignan, exhibited the most potent inhibitory activity against HCT-116 with IC50 value of 5.73 +/- 0.34 mu M, comparable to that of paclitaxel (5.03 +/- 0.69 mu M). The chemotaxonomic significance of all compounds was also discussed.
Adequate situation awareness for fault location plays a key role in preserving resilience during emergencies. To this end, many fault indicators (FIs) have been installed in the distribution system and extensive research has been conducted on the problem of optimal FI allocation. These methods have performed well in traditional distribution systems with unidirectional fault currents. However, with the development of smart grids, the integration level of distributed generations (DGs) has increased and network topologies have become more flexible and reconfigurable. Meanwhile, the increasing frequency of extreme events has led to a higher likelihood of multiple simultaneous faults. All these factors render traditional methods inapplicable. This paper first proposes a fictitious current (FC) model, which can simulate fault current distribution under multiple faults, DG integration, and flexible, reconfigurable network topologies. Based on the FC model, a tri-level FI allocation method for resilience enhancement of distribution systems is proposed: scenario-based stochastic programming is adopted to account for network topology uncertainty, while robust optimization is used to address the uncertainty of fault locations and fault current directions. Case studies on the IEEE 123-node system verify the effectiveness of the proposed method for FI allocation.
Abstract To address the limitations of traditional single-perspective risk analysis in tackling the complex safety governance challenges of urban underground space, this study analyzes core safety risks based on 128 accident cases and constructs an integrated analytical framework incorporating static structural analysis, dynamic propagation analysis, and risk chains assessment. The research was conducted by applying association rule mining, complex network analysis, and temporal propagation models. Firstly, 36 standardized risks were extracted from four dimensions: human, machine, environment, and management. The effective strong association rules were mined using the Apriori algorithm, and the confidence values of the rules were used as the corresponding edge weights to construct a directed weighted risk network. Then, the network centrality metrics integrated through principal component analysis determined the static core risks. At the same time, based on the small-world network theory, a temporal propagation model was established to depict the dynamic propagation characteristics and extract the dynamic core risks. Through the comparative analysis from two perspectives, the core safety risks were finally determined. Then, the risk propagation paths were analyzed using depth-first search, the risk chains were evaluated through multi-index quantification, and the high-risk chains were identified. The results show that adverse geotechnical conditions, inaccurate geological surveys, safety leadership and duty deficiencies, insufficient or malfunctioning emergency equipment, internal water accumulation, and extreme rainfall and flooding are the key risks with high structural importance and propagation criticality. High-risk chains are primarily triggered by geological hazards, propagating through environmental and equipment nodes before converging on management defects; management risks act as critical convergent nodes in most risk chains. Based on these findings, this study proposes a targeted intervention strategy that integrates source control and chain interruption, providing methodological support and decision-making references for the safety governance of urban underground space.
Accurately predicting mortality risk in intensive care unit (ICU) patients is critical for clinical decision-making. Large language models (LLMs) are increasingly explored for clinical prediction using structured medical data, but their outputs may exhibit demographic disparities. Mitigating such disparities without degrading predictive performance remains challenging. We systematically investigate demographic bias in LLM-based ICU mortality prediction and propose Case-Based Prompting (CAP), a training-free framework designed to improve the empirical balance between predictive performance and subgroup fairness. CAP retrieves clinically similar historical misprediction and demographic-sensitive cases with known outcomes and incorporates them as case-level contextual evidence. We further evaluate model discrimination, subgroup fairness, and feature-dependence consistency. On MIMIC-IV, CAP improved AUROC from 0.806 to 0.873 and AUPRC from 0.497 to 0.694 compared with the Baseline Prompt. CAP also reduced several subgroup disparities, particularly for sex and White-Black comparisons, although age-related disparities remained non-negligible. Controlled demographic-information ablations showed that removing demographic information did not uniformly improve both predictive performance and subgroup fairness, indicating distinct performance-fairness patterns rather than a uniform benefit from demographic suppression. Feature-dependence analysis showed generally consistent expressed clinical factor patterns across demographic subgroups. These findings suggest that CAP is a promising inference-time strategy for LLM-based clinical prediction under the evaluated retrospective setting. Further validation across additional LLMs, external datasets, and prospective clinical scenarios is required before deployment.
Visual-inertial odometry (VIO) can be viewed as a structured multimodal fusion problem, where heterogeneous sensory streams with time-varying reliability must be integrated to estimate ego-motion. Most learning-based VIO methods implicitly assume that all modalities are equally reliable at inference time and perform unconditional or softly gated fusion. However, when reliability becomes asymmetric or intermittently uncertain, such strategies may propagate corrupted information across modalities and degrade estimation stability. In this work, we propose WormVIO, a reliability-aware deep VIO framework motivated by compact sensorimotor decision processes, which formulates adaptive inference as a hierarchical discrete fusion problem. At its core, the Instinct-Bias Module is a hierarchical reliability-aware fusion mechanism that explicitly disentangles two decision factors: (i) fusion participation, which determines whether multimodal aggregation is beneficial at a given time step, and (ii) modality dominance, which determines which modality should guide estimation when reliability is imbalanced. These discrete decisions are implemented using a compact Neural Circuit Policy (NCP) combined with differentiable Gumbel-Softmax sampling, enabling end-to-end optimization of structured fusion control. Experiments on the KITTI benchmark demonstrate that the proposed decoupled fusion mechanism improves trajectory stability under intermittent visual and inertial perturbations while preserving competitive performance under nominal conditions. Furthermore, without fine-tuning, WormVIO transfers more consistently to the EuRoC MAV dataset, indicating that disentangling fusion participation and modality dominance enhances robustness to reliability shifts and cross-domain motion variations. The source code is available at: https://github.com/zRiverBird/worm_vio.git.
