Shaoyang University

A heated debate has been ongoing regarding the potential overlap in language disorders exhibited by children with developmental language disorder (DLD) and those with autism spectrum disorder plus language impairment (ALI). This study compared the production of Chinese negative structures by these two populations, alongside typically developing age-matched (TDA) children. Nineteen children with DLD, thirty-two children with ALI and twenty-eight TDA children participated in a video task and a game task on two Chinese negative structures (Structure B: V-bu-C; Structure M: mei-V-C). The results reflected a general resemblance between the two groups (DLD vs. ALI) in their most prevalent non-target responses in the two structures. However, children with ALI evinced impairment in both structures, while children with DLD only presented significantly lower-than-TDA performance in Structure M. Moreover, only the ALI group performed better in the game task than in the video task when producing Structure B, and the DLD and ALI groups produced some distinctive non-target responses. The findings indicate that the DLD group’s poor performance may be related to the impairment in the assignment of negative scope and the ALI’s difficulties may be attributed to the deficits in pragmatics. This study supports the idea that while children with DLD and children with ALI share some common symptoms in language, their language difficulties may result from different impairments. Labels should not be attached to children without an in-depth analysis of children’s responses in language assessments.

Regarding the difficulty in parameter determination for conventional maximum power point tracking (MPPT) algorithms, common issues such as sluggish tracking response and persistent oscillations near the maximum power point often arise under non-ideal operating conditions, such as environmental variations and partial shading. This paper investigates an exponential sliding-mode MPPT algorithm for photovoltaic cells in a quasi-Z-source H-bridge configuration that utilises reinforcement learning for parameter self-tuning. Initially, a mathematical model of the photovoltaic quasi-Z-source H-bridge is established to derive the equivalent sliding mode control. Subsequently, an exponential sliding-mode reaching law is designed, with its parameters self-tuned based on the distance to the sliding surface, enabling real-time MPPT control. Simulation experiments using a Simulink model compared the proposed algorithm with traditional MPPT methods under scenarios such as system startup and environmental changes. The proposed algorithm achieves fast, accurate, and stable maximum power point tracking for photovoltaic power generation systems.

Federated Graph Learning (FGL) enables collaborative training of Graph Neural Networks (GNNs) across decentralized clients while preserving data locality for privacy. However, data heterogeneity—non-IID distributions of topology, features, and labels across clients—severely degrades model performance. Existing methods struggle with graph-specific heterogeneity, and naive noise perturbation often worsens this issue. To address this, we propose an Adaptive Noise Perturbation (ANP) mechanism that repurposes noise as an active mechanism against heterogeneity. We first quantify client-specific heterogeneity, then design a scheduler that adaptively injects calibrated noise into graph structures and features. Clients with higher heterogeneity receive stronger perturbations to encourage generalized representations, while others retain more faithful local learning. Extensive experiments under non-IID settings show ANP achieves competitive gains in node classification over state-of-the-art baselines.

Given that Acer sutchuenense Franch., an endangered maple endemic to China, severely threatened by habitat degradation and climate fluctuations, understanding its spatiotemporal dynamics is crucial for formulating conservation strategies. Herein, climatic, topographic and soil variables were employed to simulate historical, present, and future distribution patterns of A. sutchuenense using the optimized MaxEnt model. Our results indicated that Mean Temperature of Driest Quarter (Bio9) and Temperature Seasonality (Bio4) were the key environmental drivers. Since the Last Interglacial, A. sutchuenense had experienced a continuously reduction in its suitable area, though the mountains surrounding the Sichuan Basin functioned as vital glacial shelters. Although the potential suitable habitat was distributed in a ring shape, A. sutchuenense occurs only on the east and west sides of the Sichuan Basin, probably due to the terrain complexity and limited dispersal ability. In the future, A. sutchuenense faces a westward contraction and a migration lag behind climate velocity due to dispersal constraints. Overall, we recommend a multi-dimensional conservation framework that prioritizes in situ conservation in core refugia, urgently establishes ecological corridors to facilitate eastward migration under climate change, implements ex situ conservation through germplasm collection for vulnerable southwestern populations, and enhances long-term monitoring to ensure species persistence.

We study rapid geometric transfer for physics-informed neural networks (PINNs) in buoyancy-driven cavity flows with an inner heated cylinder. Small eccentricities of the cylinder reshape the thermal layer and the rising plume, which makes wall heat transfer a stringent test. We propose a geometry-aware adapter augmented PINN (GeoAda-PINN) that conditions a frozen backbone on explicit geometric descriptors and signed distance features while updating only compact residual adapters during transfer. Training uses a hybrid strong and weak objective with plume-aware collocation. For vertical and horizontal offsets, we compare computational fluid dynamics (CFD), a Vanilla-PINN retrained per geometry, and GeoAda-PINN. Temperature and velocity fields and mean Nusselt numbers on inner and outer walls indicate that the Vanilla-PINN most closely matches CFD on the hardest inner wall cases, while the GeoAda-PINN achieves competitive accuracy with markedly fewer updated parameters and shorter fine-tunes. An accuracy and time frontier shows the GeoAda-PINN reaching within about 5% mean absolute percent error of the inner wall Nusselt in substantially less wall clock time. These results support parameter-efficient and geometry-conditioned adapters as a practical path to scalable PINN surrogates in multi-query convection problems.

Background Chest wall invasion is a relatively kind of infrequent direct tumor extension in non-small cell lung cancer (NSCLC) with a poor survival outcome. Risk factors that impact overall survival (OS) and cancer-specific survival (CSS) remain unclear. We aimed to explore prognostic factors and construct predictive nomograms to predict both OS and CSS in NSCLC patients with chest wall invasion. Methods We extracted a total of 2091 patients between 2010 and 2015 from the SEER database. The total patients were divided into the training cohort (1463 patients) and the validation cohort (628 patients). Univariate and multivariate Cox regression analyses were applied to distinguish the independent prognostic factors. Two prognostic nomograms for OS and CSS were established. The concordance index (C-index), receiver operating characteristic curves (ROC), calibration curves, and decision curve analysis (DCA) curves were applied to assess the performance of these two nomograms. Results After analysis, age, sex, histology, grade, N stage, M stage, surgery, and chemotherapy were identified as independent prognostic factors for OS, meanwhile, age, histology, grade, N stage, M stage, surgery, and chemotherapy for CSS. The C-index for OS in the training and validation cohorts was 0.711 and 0.716, respectively. The C-index for CSS was 0.721 and 0.726, respectively. The ROC curves, calibration curves, DCA curves, and K–M survival curves also exhibited good predictive performance. Conclusion Two nomograms provide a useful tool to predict both OS and CSS in NSCLC patients with chest wall invasion.