arXiv:2512.07136v1 Announce Type: cross Abstract: Multimodal human action recognition (HAR) leverages complementary sensors for activity classification. Beyond recognition, recent advances in large language models (LLMs) enable detailed descriptions and causal reasoning, motivating new tasks: human action understanding (HAU) and human action reasoning (HARn). However, most LLMs, especially large vision language models (LVLMs), struggle with non-RGB modalities such as depth, IMU, and mmWave due to the lack of large-scale data-caption resources. Existing HAR datasets mainly provide coarse data-label annotations, which are insufficient to capture fine-grained action dynamics needed for HAU and HARn. We consider two ground-truth pair types: (1) data label (discrete category) and (2) data caption (textual description). Naively generating captions from labels often lacks logical and spatiotemporal consistency. We introduce CUHK-X, a large-scale multimodal dataset and benchmark suite for HAR, HAU, and HARn. CUHK-X contains 58,445 samples covering 40 actions performed by 30 participants across two indoor environments. To improve caption consistency, we propose a prompt-based scene creation method that leverages LLMs to generate logically connected activity sequences, followed by human validation. CUHK-X includes three benchmarks with six evaluation tasks. Experiments report average accuracies of 76.52% (HAR), 40.76% (HAU), and 70.25% (HARn). CUHK-X aims to enable the community to apply and develop data-intensive learning methods for robust, multimodal human activity analysis. Project page and code: https://openaiotlab.github.io/CUHK-X/ and https://github.com/openaiotlab/CUHK-X.

Background: In the digital health era, telemedicine has become a key driver of health care reform and innovation globally. Understanding the factors influencing residents’ choices of telemedicine services is crucial for optimizing service design, enhancing user experience, and developing effective policy measures. Objective: This study aims to explore the key factors influencing Chinese residents’ choices of telemedicine services, including consultation fee, physician qualifications, appointment waiting time, scope of services, privacy protection, and service hours. The study also analyzes preference heterogeneity among residents with different demographic characteristics to provide scientific evidence for optimizing telemedicine services in the digital health era. Methods: This study used a mixed methods design combining qualitative interviews and a discrete choice experiment. Interviews identified key telemedicine attributes, informing the discrete choice experiment scenarios. Preferences and willingness to pay were analyzed using mixed logit and latent class models. Results: Residents’ preferences for telemedicine services were primarily shaped by the scope of services, appointment waiting time, and privacy protection, with substantial willingness to pay for more comprehensive, secure, and timely services. The optimal telemedicine services configuration—offering consultation plus prescription, high privacy, immediate access, 24-hour availability, and expert physicians—yielded a maximum willingness to pay of RMB 661.6 (a currency exchange rate of US $1=RMB 7.1803 is applicable). Latent class analysis revealed pronounced heterogeneity: while privacy and service scope remained universally prioritized, older, male, rural, and less-educated residents favored broader coverage, easier platforms, and lower costs; younger, female, and highly educated groups preferred faster, higher-quality, and more privacy-sensitive services. Conclusions: This study reveals key drivers and significant demographic heterogeneity in Chinese residents’ preferences for telemedicine services. Residents demonstrated a high willingness to pay for comprehensive services (eg, “consultation + prescription”), enhanced privacy protection, and shorter appointment waiting times. Additionally, the study innovatively identified 3 distinct resident profiles: “Diverse-Service-Oriented,” “Utility-Oriented,” and “Value-Oriented,” and proposed differentiated optimization strategies to effectively address diverse resident needs, thereby promoting equitable access and efficient adoption of telemedicine services.