Hui Liu

Dementia prevention is an urgent global health priority in aging populations. Mild cognitive impairment (MCI) presents a key window for intervention, where

Geographic segmentation of health insurance creates administrative barriers that limit patients' access to out-of-network healthcare, particularly in underserved regions. We assess the impact of a reimbursement reform removing interregional administrative barriers on patient healthcare utilization. We conduct a quantitative quasi-experimental study using longitudinal administrative data from a representative city with limited high-tier medical resources. The causal effects of the reform are estimated based on patient visit records from January 2021 to November 2022 using a regression discontinuity of time (RDiT) design. The reform significantly increases intercity healthcare visits by up to 11.2%, especially among low-income patients and women. While patients with severe conditions benefit, increased utilization for minor ailments also occurs. This surge in demand strains medical resources, leading to increased readmission rates without reducing hospital stays. Removing administrative barriers enhances equity but may reduce healthcare efficiency if demand management is neglected. Policymakers should combine expanded access with investments in local healthcare capacity and demand-side controls to ensure sustainable and equitable service delivery.

Background: The management of chronic obstructive pulmonary disease (COPD) places a significant burden on health care systems worldwide. Home noninvasive positive pressure ventilation (NPPV) is an established treatment option associated with significant benefits for patients with COPD and hypercapnic chronic respiratory failure. Internet of Things (IoT)–based management may improve communication between patients and physicians and strengthen the integration and comprehensiveness of home NPPV telemonitoring. However, the economic value of such systems remains insufficiently understood. Objective: This study aimed to assess the cost-effectiveness of IoT-based management versus standard management of home NPPV in patients with COPD and hypercapnic chronic respiratory failure. Methods: A Markov decision analytic model was developed to simulate real-world COPD progression and predict health outcomes and costs associated with IoT-based and standard management of home NPPV. COPD progression consisted of 4 health states: stable period, nonserious exacerbation period, serious exacerbation period, and death. Efficacy and cost inputs were primarily sourced from a published multicenter, prospective, randomized controlled trial and supplemented by official Chinese databases where necessary. Quality-adjusted life years (QALYs) were used as effect indicators for this model, which were derived from COPD Assessment Test scores. The discounted lifetime cost per QALY gained was calculated from the Chinese health care payer perspective, and sensitivity analyses were conducted to test the robustness of model results across different assumptions. Results: Compared with standard NPPV, IoT-based NPPV increased costs by ¥3607.26 and improved QALYs by 0.24 per person across the lifetime horizon, resulting in an incremental cost-effectiveness ratio of ¥15,030.25 per QALY, with a 93.6% probability of being cost-effective at the given willingness-to-pay threshold (currency conversion to US dollars was based on the average exchange rate in 2019 [US $1=¥6.9]). Base case results were also robust to multiple one-way sensitivity analyses, with the main drivers being hospitalization costs for the standard and IoT-based NPPV groups during the serious exacerbation period. Conclusions: IoT-based NPPV was cost-effective compared with standard NPPV for patients with COPD and hypercapnic chronic respiratory failure.

Background: As artificial intelligence continues to play an expanding role in healthcare, ensuring its compliance with medical ethics is essential. However, the ethical performance of artificial intelligence in medical contexts remains insufficiently studied. This study aimed to evaluate the ability of ChatGPT to address questions related to medical ethics and to compare its performance with that of human experts. Methods: A Medical Ethics Evaluation dataset was developed, consisting of 465 single‐choice questions derived from a range of medical ethics standards. These questions were used to assess two artificial intelligence models, GPT‐3.5 and GPT‐4. Model responses were compared with those provided by two medical ethics experts. Each test was conducted independently twice to ensure consistency. Accuracy was calculated for each model and expert, and chi‐square tests were used to compare differences in performance. Results: GPT‐3.5 achieved an overall accuracy of 38.92%, while GPT‐4 achieved 27.10%. In comparison, two medical ethics experts achieved substantially higher accuracies of 86.23% and 78.32%, respectively. Both experts performed significantly better than GPT‐3.5 and GPT‐4. These findings indicate a substantial gap between artificial intelligence models and human experts in understanding and applying medical ethics principles. The relatively low performance of the models, compared with their reported strengths in diagnostic tasks, may reflect the complexity and nuance of ethical reasoning in medicine. Nevertheless, the large language models showed some ability to align with core medical ethics principles, particularly in ethical dilemma scenarios, and were also able to generate responses that addressed psychological needs. Conclusions: Artificial intelligence models currently show limited accuracy in medical ethics decision‐making compared with human experts. Although these models demonstrate some alignment with fundamental ethical principles, the performance is not yet sufficient for reliable use in ethically sensitive medical contexts. Further optimization is needed to improve their ability to meet the ethical demands of medical practice.

Recent copyright lawsuits against artificial intelligence (AI) and large language model (LLM) developers have ignited debates over how to balance technological innovation with the public interest. In the scientific research field, the performance and reliability of LLMs trained on scientific literature (SciLit-LLMs) depend heavily on access to comprehensive, up-to-date full-text sources. This paper argues that the current copyright framework, including the U.S. fair use doctrine, often regarded as a flexible solution for AI-related copyright issues, is ill-suited for SciLit-LLMs. First, the normative values emphasized in scientific—such as accuracy, transparency and interpretability—fundamentally conflict with the “transformative use” requirement central to copyright law. Second, the expression of scientific literature is intended to ensure scientific precision rather than to convey creative originality, remains insufficiently considered under current copyright law. Third, the fair use doctrine’s emphasis on limiting the proportion of use from a single copyrighted work contradicts the need for comprehensive training on information-dense scientific texts. Finally, commercial use restrictions impede the sustainable development of SciLit-LLMs and preclude a mutually beneficial model for researchers, publishers, developers, and the public. Imposing current copyright restrictions on these models is unjustified, unnecessary, and risks perpetuating scientific biases. We therefore propose reconstructing copyright exceptions for scientific literature and removing commercial use restrictions to better support scientific innovation.

Crowdsourcing delivers responses that are asynchronous and incomplete, making offline aggregators that assume complete response sets impractical. Prior online methods often either require per-step completeness or repeatedly reload historical responses, which is storage- and privacy-unfriendly and susceptible to forgetting. We present OLA-Incomplete, an online label-aggregation framework designed for incomplete response streams. It integrates a variational-inference aggregator with a generative replay module that preserves historical information without reloading prior responses and explicitly models unknown worker reliability. At each update step, the generator replays cumulative responses and side information for previously observed instances to mitigate catastrophic forgetting, while the aggregator infers current truths by maximizing the evidence lower bound over a mixture of replayed and newly received labels. Across three public datasets—Duck, RTE, and PostSent—OLA-Incomplete attains final accuracies of 90.74%, 92.50%, and 95.99%, respectively, delivering at least 7.79% relative improvement over the strongest baseline. The approach further exhibits strong instantaneous online accuracy and robustness across response-chunk sizes and arrival orders, underscoring its practical utility for real-world crowdsourcing workflows.

What is already known about this topic? The Sustainable Development Goals target a 50% reduction in anemia among women of reproductive age by 2025 and the elimination of all forms of malnutrition by 2030. However, robust evidence documenting temporal changes in anemia prevalence remains scarce. What is added by this report? Drawing on large-scale national health examination data, this report demonstrates overall progress in reducing anemia among urban women in China between 2019 and 2024. However, it also reveals increasing prevalence in several provinces and a growing burden of moderate-to-severe anemia specifically among women aged 40–49 years. What are the implications for public health practice? Risk-stratified and targeted anemia prevention and control strategies are urgently needed. Priority should be given to women aged 40–49 years, women of reproductive age residing in high-burden provinces, and geographic areas where anemia prevalence either exceeds 20% or demonstrates an upward trend.

The association between metabolic risk factors and human papillomavirus (HPV) infection is under debate. This study aimed to investigate the associations between metabolic risk factors and HPV infection, while also providing updated data on the prevalence of HPV infection in the post-vaccine era in China. This cross-sectional study involved 684,632 adult women who had undergone a standard health screening in 2019. HPV infection prevalence was estimated. Linear and nonlinear associations between HPV infection and metabolic risk factors were assessed, such as body mass index (BMI), systolic blood pressure (SBP), fasting plasma glucose (FPG), and low-density lipoprotein (LDL). The standardized prevalence of overall and high-risk HPV infection was 16.2% (95% confidence interval [95%CI]: 15.1–17.3) and 12.8% (95%CI: 11.8–13.7), respectively. HPV infection was negatively associated with BMI, with women in the highest quartile having the lowest odds of infection compared to those in the lowest quartile (odds ratio [OR] = 0.84, 95%CI: 0.78–0.91, P < 0.0001), indicating a significant linear trend (Ptrend<0.0001). Nonlinear associations were observed for SBP (Pnon−linearity<0.0001), FPG (Pnon−linearity<0.0001) and LDL (Pnon−linearity=0.008). The overall HPV risk was observed to decrease with increasing SBP, especially when SBP was below 114 mmHg. Both FPG and LDL were positively associated with HPV risk. ORs were more pronounced when FPG was less than 4.9 mmol/L and LDL was between 2.7 and 3.5 mmol/L. The prevalence of HPV infection in China remains high, and metabolic risk factors might be associated with an increased susceptibility to HPV. Longitudinal studies are needed to validate the association between metabolic risk factors and HPV infection.

What is already known about this topic? Female breast nodules represent the most frequently detected lesions during breast ultrasound screening. Notably, nodules classified as Breast Imaging Reporting and Data System (BI-RADS) 4 or 5 indicate an elevated risk of breast cancer. Nevertheless, the detection rate and BI-RADS classification of female breast nodules in China remain largely undocumented. What is added by this report? In 2021, breast nodules were detected in 27.9% of urban women in China. Among women with breast nodules marked with BI-RADS classification information, 95.9% were categorized as BI-RADS 2–3, while 4.0% were classified as BI-RADS 4–5. Age, geographic location, per capita gross domestic product (GDP), body mass index (BMI), high triglyceride (TG), high low-density lipoprotein cholesterol (LDL-C), and diabetes were identified as risk factors for BI-RADS 4–5. What are the implications for public health practice? This study highlights the importance of managing high-risk women with breast nodules through BI-RADS classification. Women with older age, high TG, high LDL-C, or diabetes demonstrate higher detection rates of BI-RADS 4–5, underscoring the need for targeted health interventions for high-risk populations while accounting for regional and socioeconomic disparities.

What is already known about this topic? Breast cancer has emerged as the most prevalent cancer among women globally and is increasingly affecting younger populations. However, the relationship between individual socioeconomic status (SES) and breast cancer risk remains incompletely understood. What is added by this report? This population-based cohort study revealed a breast cancer incidence rate of 48.9 per 100,000 person-years. Women with high SES demonstrated a significantly elevated risk of breast cancer compared to those with low SES [hazard ratio (HR)=1.42, 95% confidence interval (CI): 1.05, 1.92]. Self-perceived SES appeared to moderate this association, with an increased breast cancer risk particularly evident among women who had both low objectively assessed and self-perceived SES. What are the implications for public health practice? These findings underscore the need for tailored breast cancer screening programs and targeted health education initiatives that account for differences across SES groups.