arXiv subjects

Cai Wu

Publications and source records attributed to Cai Wu.

7 recordsLinked to original sources

arXiv2026

Urban Boundaries, Social Barriers: A Benchmark and Vision-Centric Framework for Mapping Gated Communities and Equity Implications

Communities are fundamental spatial units that shape urban form and social life. Whether a residential compound is spatially open or enclosed affects mobility, access to public services, and equity, yet studies of Chinese fengbi xiaoqu remain largely qualitative or small-scale, limiting reproducible city-scale analysis. We address this gap by introducing GBA-GCs, a metropolitan-scale multimodal benchmark for locally grounded gated/open community recognition in China's Greater Bay Area, covering 37,444 residential compounds with aligned boundary polygons, high-resolution satellite imagery, Chinese metadata, and structured attributes, together with expert-verified labels, inter-annotator reliability, and official evaluation splits. Built on this benchmark, we present Multimodal Classifier for Gated Community (MCGC), a vision-centric multimodal framework based on DINOv3-SAT that fuses imagery, text, and structured cues via modality-aware cross-attention and adaptive gating to mitigate modality imbalance. MCGC consistently outperforms strong unimodal and multimodal baselines. Finally, we apply the validated model to metropolitan-scale mapping and report equity-oriented findings including spatial clustering of GCs, privatized green space, and reduced pedestrian connectivity. The benchmark, code, and release documentation are available at https://github.com/MinweiZhao/GBA-GCs.

cs.CV↗

arXiv2025

Optimal and Efficient Sample Size Re-estimation: A Dynamic Cost Framework

Adaptive sample size re-estimation (SSR) is a well-established strategy for improving the efficiency and flexibility of clinical trials. Its central challenge is determining whether, and by how much, to increase the sample size at an interim analysis. This decision requires a rational framework for balancing the potential gain in statistical power against the risk and cost of further investment. Prevailing optimization approaches, such as the Jennison and Turnbull (JT) method, address this by maximizing power for a fixed cost per additional participant. While statistically efficient, this paradigm assumes the cost of enrolling another patient is constant, regardless of whether the interim evidence is promising or weak. This can lead to impractical recommendations and inefficient resource allocation, particularly in weak-signal scenarios. We reframe SSR as a decision problem under dynamic costs, where the effective cost of additional enrollment reflects the interim strength of evidence. Within this framework, we derive two novel rules: (i) a likelihood-ratio based rule, shown to be Pareto optimal in achieving smaller average sample size under the null without loss of power under the alternative; and (ii) a return-on-investment (ROI) rule that directly incorporates economic considerations by linking SSR decisions to expected net benefit. To unify existing methods, we further establish a representation theorem demonstrating that a broad class of SSR rules can be expressed through implicit dynamic cost functions, providing a common analytical foundation for their comparison. Simulation studies calibrated to Phase III trial settings confirm that dynamic-cost approaches improve resource allocation relative to fixed-cost methods.

stat.ME↗

arXiv2025

Futility Analysis under Scrutiny

This paper investigates the robustness of futility analyses in clinical trials when interim analysis population deviates from the target population. We demonstrate how population shifts can distort early stopping decisions and propose post-stratification strategies to mitigate these effects. Simulation studies illustrate the impact of subgroup imbalances and the effectiveness of naive, model-based, and hybrid post-stratification methods. We also introduce a permutation-based screening test for identifying variables contributing to population heterogeneity. Our findings support the integration of post-stratification adjustments using all available baseline data at the interim analysis to enhance the validity and integrity of futility decisions.

stat.ME↗

arXiv2025

From Street Form to Spatial Justice: Explaining Urban Exercise Inequality via a Triadic SHAP-Informed Framework

Urban streets are essential everyday health infrastructure, yet their capacity to support physical activity is unevenly distributed. This study develops a theory-informed and explainable framework to diagnose street-level exercise deprivation by integrating Lefebvre's spatial triad with multi-source urban data and SHAP-based analysis. Using Shenzhen as a case study, we show that while conceived spatial attributes have the strongest overall influence on exercise intensity, local deprivation mechanisms vary substantially across contexts. We identify a seven-mode typology of deprivation and locate high-demand but low-support street segments as priority areas for intervention. The study offers both a theory-grounded analytical framework and a practical diagnostic tool for promoting spatial justice in everyday physical activity.

cs.CY↗

arXiv2025

UST-GNN: A Unified Spatial--Topological Graph Neural Network Framework for Urban Analytics--Demonstrated through a Case Study on Urban Health Prediction

Understanding how social, demographic, environmental, and spatial factors jointly shape urban outcomes is essential for sustainable urban development and evidence-based policy. Traditional statistical approaches often struggle to capture complex non-linear relationships, while many machine learning methods overlook the joint roles of spatial autocorrelation and network topology in urban systems. Recent advances in GeoAI have addressed these challenges only partially, often treating spatial effects, graph structure, evaluation, and interpretability separately. We present \textbf{UST-GNN}, a unified spatial--topological graph neural network framework that integrates neighbourhood connectivity, heterogeneous urban features, and positional/locational embeddings into a single representation. Using the MedSAT dataset, which contains over 150 environmental and socio-demographic variables and six prescription outcomes across 4,835 neighbourhoods in Greater London, UST-GNN outperforms strong statistical, geographically enhanced, and graph Machine Learning baselines, improving out-of-sample $R^2$ by 8.4--13.2\% under strict spatial cross-validation. We further introduce a lightweight principal-component module to interpret learned node embeddings geographically and relate them to policy-relevant covariates. The resulting analyses recover established patterns, offer new perspectives on debated associations, and reveal novel predictors warranting further causal investigation. Together, these findings demonstrate the value of graph-based spatial machine learning for urban health analytics, environmental inequality assessment, and evidence-based urban policy. Beyond predictive gains, UST-GNN provides a unified GeoAI analytical pipeline that can be embedded into urban digital twin workflows for scenario testing, monitoring, and data-informed decision-making for healthier, more sustainable cities.

cs.LG↗

arXiv2019

Dynamic Prediction of Competing Risk Events using Landmark Sub-distribution Hazard Model with Multiple Longitudinal Biomarker

The cause-specific cumulative incidence function (CIF) quantifies the subject-specific disease risk with competing risk outcome. With longitudinally collected biomarker data, it is of interest to dynamically update the predicted CIF by incorporating the most recent biomarker as well as the cumulating longitudinal history. Motivated by a longitudinal cohort study of chronic kidney disease, we propose a framework for dynamic prediction of end stage renal disease using multivariate longitudinal biomarkers, accounting for the competing risk of death. The proposed framework extends the landmark survival modeling to competing risks data, and implies that a distinct sub-distribution hazard regression model is defined at each landmark time. The model parameters, prediction horizon, longitudinal history and at-risk population are allowed to vary over the landmark time. When the measurement times of biomarkers are irregularly spaced, the predictor variable may not be observed at the time of prediction. Local polynomial is used to estimate the model parameters without explicitly imputing the predictor or modeling its longitudinal trajectory. The proposed model leads to simple interpretation of the regression coefficients and closed-form calculation of the predicted CIF. The estimation and prediction can be implemented through standard statistical software with tractable computation. We conducted simulations to evaluate the performance of the estimation procedure and predictive accuracy. The methodology is illustrated with data from the African American Study of Kidney Disease and Hypertension.

q-bio.QM↗

arXiv2017

Quantifying and Estimating the Predictive Accuracy for Censored Time-to-Event Data with Competing Risks

This paper focuses on quantifying and estimating the predictive accuracy of prognostic models for time-to-event outcomes with competing events. We consider the time-dependent discrimination and calibration metrics, including the receiver operating characteristics curve and the Brier score, in the context of competing risks. To address censoring, we propose a unified nonparametric estimation framework for both discrimination and calibration measures, by weighting the censored subjects with the conditional probability of the event of interest given the observed data. We demonstrate through simulations that the proposed estimator is unbiased, efficient and robust against model misspecification in comparison to other methods published in the literature. In addition, the proposed method can be extended to time-dependent predictive accuracy metrics constructed from a general class of loss functions. We apply the methodology to a data set from the African American Study of Kidney Disease and Hypertension to evaluate the predictive accuracy of a prognostic risk score in predicting end-stage renal disease (ESRD), accounting for the competing risk of pre-ESRD death.

stat.ME↗