arXiv subjects

Heming Zheng

Publications and source records attributed to Heming Zheng.

3 recordsLinked to original sources

arXiv2025

Evaluating the Challenges of LLMs in Real-world Medical Follow-up: A Comparative Study and An Optimized Framework

When applied directly in an end-to-end manner to medical follow-up tasks, Large Language Models (LLMs) often suffer from uncontrolled dialog flow and inaccurate information extraction due to the complexity of follow-up forms. To address this limitation, we designed and compared two follow-up chatbot systems: an end-to-end LLM-based system (control group) and a modular pipeline with structured process control (experimental group). Experimental results show that while the end-to-end approach frequently fails on lengthy and complex forms, our modular method-built on task decomposition, semantic clustering, and flow management-substantially improves dialog stability and extraction accuracy. Moreover, it reduces the number of dialogue turns by 46.73% and lowers token consumption by 80% to 87.5%. These findings highlight the necessity of integrating external control mechanisms when deploying LLMs in high-stakes medical follow-up scenarios.

cs.CL↗

arXiv2025

Breast-Rehab: A Postoperative Breast Cancer Rehabilitation Training Assessment System Based on Human Action Recognition

Postoperative upper limb dysfunction is prevalent among breast cancer survivors, yet their adherence to at-home rehabilitation exercises is low amidst limited nursing resources. The hardware overhead of commonly adopted VR-based mHealth solutions further hinders their widespread clinical application. Therefore, we developed Breast-Rehab, a novel, low-cost mHealth system to provide patients with out-of-hospital upper limb rehabilitation management. Breast-Rehab integrates a bespoke human action recognition algorithm with a retrieval-augmented generation (RAG) framework. By fusing visual and 3D skeletal data, our model accurately segments exercise videos recorded in uncontrolled home environments, outperforming standard models. These segmented clips, combined with a domain-specific knowledge base, guide a multi-modal large language model to generate clinically relevant assessment reports. This approach significantly reduces computational overhead and mitigates model hallucinations. We implemented the system as a WeChat Mini Program and a nurse-facing dashboard. A preliminary clinical study validated the system's feasibility and user acceptance, with patients achieving an average exercise frequency of 0.59 sessions/day over a two-week period. This work thus presents a complete, validated pipeline for AI-driven, at-home rehabilitation monitoring.

cs.HC↗

arXiv2025

Improving vision-language alignment with graph spiking hybrid Networks

To bridge the semantic gap between vision and language (VL), it is necessary to develop a good alignment strategy, which includes handling semantic diversity, abstract representation of visual information, and generalization ability of models. Recent works use detector-based bounding boxes or patches with regular partitions to represent visual semantics. While current paradigms have made strides, they are still insufficient for fully capturing the nuanced contextual relations among various objects. This paper proposes a comprehensive visual semantic representation module, necessitating the utilization of panoptic segmentation to generate coherent fine-grained semantic features. Furthermore, we propose a novel Graph Spiking Hybrid Network (GSHN) that integrates the complementary advantages of Spiking Neural Networks (SNNs) and Graph Attention Networks (GATs) to encode visual semantic information. Intriguingly, the model not only encodes the discrete and continuous latent variables of instances but also adeptly captures both local and global contextual features, thereby significantly enhancing the richness and diversity of semantic representations. Leveraging the spatiotemporal properties inherent in SNNs, we employ contrastive learning (CL) to enhance the similarity-based representation of embeddings. This strategy alleviates the computational overhead of the model and enriches meaningful visual representations by constructing positive and negative sample pairs. We design an innovative pre-training method, Spiked Text Learning (STL), which uses text features to improve the encoding ability of discrete semantics. Experiments show that the proposed GSHN exhibits promising results on multiple VL downstream tasks.

cs.CV↗