江苏省苏州中学

Loneliness and social isolation have been a rising public health concern among older adults in recent years, especially after the COVID-19 pandemic. However, how the distinct pathways affect health and the effectiveness of existing interventions still need clearer review. By collecting and comparing data from large-scale studies and meta-analyses, the review systematically analyzes different impacts of loneliness and social isolation on cognitive function, biological health, frailty and mortality. In addition, it evaluates the effectiveness of technology-based and social interventions through systematic reviews and experiments on older adults' reflections to interventions. The results show that social isolation is a more robust predictor of dementia and mortality, while loneliness is more strongly connected to cognitive decline and frailty progression. The COVID-19 pandemic significantly intensified both conditions, which the prevalence rates remain above the pre-pandemic level. For intervention strategies, both technology-based and social intervention show positive effects on reducing loneliness and social isolation, but the evidence is limited due to the heterogeneity and methodological weaknesses. Specifically, information and communication technology (ICT) and videoconferencing demonstrate the most significant effect among technology intervention, while structured group-based intervention is more effective than other social interventions. Though most research show significant results in their studies, there are limitations in standardized measurements. The need for consistent outcome measurement across studies can improve the research validity, while with larger-scale randomized controlled trials can increase the reliability of studies. At the same time, the results of intervention effectiveness help build efficient policy interventions.

RNA viruses are very dangerous to global health. Because of their tendency to either undergo high rates of mutation or quickly evolve as the case of influenza, COVID-19, and Ebola depicts. Their etiology is not only the solution to a core mystery in the life sciences, but it also applies to prevention of the disease by viruses and to drug development. This paper compares three classical origin theories, including the Escape Hypothesis (fragments escape from cellular genetic material), the Regressive Hypothesis (degeneration of complex parasitic cells) and the Coevolution Hypothesis. By analyzing empirical data, logical and limitation of both theories, the paper shows the pluralistic nature of the origin of viruses: retroviruses have the Escape Hypothesis, giant viruses have the Regressive Hypothesis, and the majority of RNA viruses have the Coevolution Hypothesis. The Coevolution Hypothesis, which is backed by a set of Endogenous Viral Elements (EVEs) in the genomes of its hosts and a strong level of conservation of replicase genes in lineages of ancient RNA viruses, is the most clear umbrella explanatory force on the overall diversity of RNA viruses. The study contributes to the existing body of knowledge on viral evolution by confirming the complementary character of the three hypotheses. It offers theory foundations to antiviral guidelines. They are the development of broad-spectrum antiviral drugs against conserved viral replicase proteins, and predicting the risks of zoonotic infections depending on lineage evolution.

Digital platforms implement personalized pricing relying on massive user data and algorithm technologies, and big data driven "big data ripening" has continuously triggered consumer disputes, making algorithmic differential pricing a key challenge for digital economy governance. Taking algorithmic differential pricing as the research object, this paper draws on theories of price discrimination, behavioral pricing and regulatory economics and adopts theoretical analysis and logical deduction to explore its operational logic. Algorithmic differential pricing represents traditional price discrimination adapted to digital contexts. Supported by user data portraits, it realizes personalized "thousands of people, thousands of prices" and approximates first degree price discrimination. The outcome of algorithmic tiered pricing hinges on data advantages, user lock in and market competition, and its welfare effect is dual sided rather than simply beneficial or detrimental to markets. This paper establishes a hierarchical industry differentiated regulatory framework covering flexible supervision, stringent supervision and bottom line protection to reconcile industrial market features and consumer rights demands. It extends theoretical studies on algorithmic price discrimination and offers references for prudent regulatory practice for platform algorithmic pricing.

This paper treats regional warming in China as an unsupervised pattern-mining task on ERA5 spatio-temporal data for thirty-four provincial-level cities. First, about 1.8 million hourly values of six ERA5 variables for 2020 are reduced to five standardized features per city, and K-means groups the cities into five clusters with no predefined zoning labels. Second, the cluster labels are used to pool the 1975–2024 2-meter temperature series, and the regional decadal means are compared by cumulative warming and by the linear rate from ordinary least squares regression on decade midpoints. Every region warmed, though by different amounts. From 1975–1984 to 2015–2024 the Northwest gained 1.99 °C (0.51 °C per decade) and the Qinghai–Tibet Plateau gained 1.61 °C (0.40 °C per decade). South China had the smallest rise, 1.09 °C (0.24 °C per decade). The warming was also not steady, as the North–Northeast and South China regions cooled a little in 2005–2014 and then rose again, but the Northwest and the plateau kept warming. Mostly the clusters follow conventional geography, and only three boundary cities change region. Thus the partition learned without a zoning scheme separates warming paths of different size and timing, with faster warming in the north-western interior and on the plateau than on the southern coast.

This dissertation investigates and evaluates the accuracy and potential of data mining techniques for identifying traffic accident hotspots, which is a fundamental step of road accident analysis, identifying the accident-prone regions to bolster further accident research, and the proposal of prevention methods. In the discussion, a hotspot analysis based on data mining techniques is conducted to research the accidents that occurred in Birmingham, UK, in 2017. The identification result is compared with that of Kernel Density Estimation (KDE), a classical identification method, to assess the ability and accuracy of data mining’s identification. The finding underscores that data mining is a potentially effective method to identify the accident hotspot with an accurate estimation.