New York University Shanghai

Navigating human-populated environments without causing discomfort is a critical capability for socially-aware agents. While rule-based approaches offer interpretability through predefined psychological principles, they often lack generalizability and flexibility. Conversely, data-driven methods can learn complex behaviors from large-scale datasets, but are typically inefficient, opaque, and difficult to align with human intuitions. To bridge this gap, we propose RLSLM, a hybrid Reinforcement Learning framework that integrates a rule-based Social Locomotion Model, grounded in empirical behavioral experiments, into the reward function of a reinforcement learning framework. The social locomotion model generates an orientation-sensitive social comfort field that quantifies human comfort across space, enabling socially aligned navigation policies with minimal training. RLSLM then jointly optimizes mechanical energy and social comfort, allowing agents to avoid intrusions into personal or group space. A human-agent interaction experiment using an immersive VR-based setup demonstrates that RLSLM outperforms state-of-the-art rule-based models in user experience. Ablation and sensitivity analyses further show the model’s significantly improved interpretability over conventional data-driven methods. This work presents a scalable, human-centered methodology that effectively integrates cognitive science and machine learning for real-world social navigation.

Age-group specific disparities for dentalcare use persist in the United States. The COVID-19 led to delays in non-urgent dentalcare. We provide national estimates on dentalcare use and influencing factors for the U.S. population before and during the COVID-19. We used nationally representative Medical Expenditure Panel Survey for over pre-COVID-19 years (2018-2019) and COVID-19 years (2020-2021) We estimated yearly survey-weighted trends in mean non-zero dental visits by age followed Poisson regression, controlling for a comprehensive set of confounders across five domains of influence. Dentalcare visits were defined as visits to any dentalcare provider. Overall analytic sample included non-institutionalized community living persons (unweighted n=6518, weighted N∼320 million) grouped as ages 0-17, 18-44, 45-64, 65-74 and 75+ present in all four years The prevalence ratio (PR) for dental visits was slightly higher for ages 75+ in comparison to ages 65-74 across years 2018-2021 and increased from 1.73 (95% CI: 1.4, 2.1) to 1.84 (95% CI: 1.5, 2.3) to 2.13 (95% CI: 1.7, 2.7) from 2018 to 2020 but rebounding to near pre-pandemic level in 2021 to 1.66 (95% CI, 1.3, 2.0). Consistent factors during COVID-19 pandemic years 2020-2021 that increased dental visits included dental insurance, high income, and having a usual source of care (p < 0.01). Dentalcare use rebounded for older adults in 2021 but remained below pre-pandemic levels. Increasing dentalcare visits across ages remains a key policy priority. Continued monitoring of dentalcare use trends beyond COVID-19 among older adults is critical to improve their oral health.

Self-disclosure is a fundamental aspect of human communication, particularly important for adolescents. As adolescents navigate the gradual transition from parental reliance to individual autonomy, they increasingly turn to their peers as trusted confidants. Although self-disclosure has been associated with positive psychological outcomes among adolescents, research remains limited, inconsistent, and outdated, particularly in the Majority World. This study examined the direct relations between self-disclosure and psychological well-being among 192 Chinese adolescents and explored the indirect associations via self-perception. Adolescents (Mage = 13.18 years, SD = 1.1, 57.3% girls) reported their self-disclosure behaviors with best friends, self-perception, and psychological well-being. Results showed that Chinese adolescents reported a medium to high level of disclosure with their best friends, with no significant sex differences. Adolescents disclosed the most on school-related topics and the least about boy-girl relationships, with girls perceiving certain topics as more intimate than boys. Consistent with existing literature, self-disclosure was positively associated with overall psychological well-being (particularly the subfactors of self-acceptance, positive relationships, and personal growth). The indirect relation between self-disclosure and psychological well-being via perceived social competence was significant. These findings underscore the important role of self-disclosure in supporting to foster adolescents' mental well-being and reveal enhanced social competence as one potential pathway linking the two. Overall, the study confirms the links between self-disclosure and youth well-being and adds to our understanding of disclosure behaviors among Chinese adolescents.

The Green Fluorescent Protein (GFP) is widely used in imaging organisms at the sub-cellular level. However, upon exposure to UV or intense visible light, GFP undergoes irreversible reactions, altering its photocycle, which are believed to precede via photooxidation of the chromophore. The mechanism of this process is not well understood, even in the gas phase, with competing interpretations of photoelectron experiments on the isolated chromophore arguing either for nonadiabatic decay or vibrational energy relaxation and autoionization from an initially populated S3 shape resonance. To address the controversy, we simulated the excited-state dynamics and time-resolved photoelectron spectroscopy (TRPES) of the GFP chromophore with ab initio multiple spawning and on-the-fly multiconfigurational electronic structure using our dynamically weighted complete active space self-consistent field method. Our simulations show excellent agreement with experimental TRPES and reveal that S3-S2 non-adiabatic transitions do occur on an ultra-fast timescale that can compete with autoionization; however, the conversion between the shape and Feshbach states occurs primarily adiabatically on the S2 state. Furthermore, the threshold energies of the shape and Feshbach resonances are very similar, with a low barrier separating these regions of the PES, leading to both states being populated and reversibly inter-converting within 50 fs of the initial photoexcitation. As a result, rapid autoionization still precedes via the shape resonance. The picture that emerges thus reconciles the two competing views of the GFP chromophore's UV response: both internal conversion and vibrational energy relaxation from the shape resonance state are operative. Our findings of the involvement of the Feshbach state suggest new strategies to engineer FP chromophores with tailored photostabilities.

This study investigates exact solitary wave solutions of the modified Camassa–Holm (mCH) and modified Kuramoto–Sivashinsky (mKS) equations, which are fundamental in fluid dynamics, nonlinear optics, and quantum mechanics. Solutions are obtained using the [Formula: see text]-expansion neural network analytical method, a hybrid approach that integrates the symbolic capability of neural networks (NNs) with [Formula: see text]-expansion method, enabling the direct construction of analytical exact solutions without relying on classical transformations. The method yields a rich variety of soliton structures in trigonometric, hyperbolic, and rational forms, including periodic, bright, dark, V-shaped, kink, and singular kink solitons. The dynamics are illustrated through 2D, 3D with density surface, and polar plots, providing clear physical insights. The results demonstrate the efficiency and versatility of the method in capturing complex nonlinear behaviors and suggest potential applications in nonlinear wave propagation in fluid dynamics, plasma physics, optical communications, and biological systems.