In smart healthcare systems, Image data of critical patients is essential in controlling and diagnosing the disease development. To acquire the medical images, traditional methods encountered the difficulty of generating cost-effective data. This research work introduces a novel and innovative approach to collect high-quality image data from individuals with atypical clinical presentations. Initially, a new Internet of Medical Things (IoMT) image collection architecture is introduced. This design uses edge intelligence and motion-static synergy to make it easier to record both coarse-grained and fine-grained patient images. This study introduces an image acquisition technique that leverages edge intelligence and collaborative static-dynamic monitoring, exemplified in intensive care units, to improve the efficiency and data value of image acquisition in healthcare IoMT settings. This approach revolves around the three distinct steps. To begin with, an advanced YOLO-based clinical abnormality detection is implemented by the edge server to identify patients affected by abnormal physiological conditions. The images from affected patients are captured by static monitoring nodes. In the next phase, coordinate calculation methods for the localization of abnormal patients and quantification techniques for severity assessment are introduced. The final step involves the intervention of a path optimization algorithm for mobile medical assistive robots using severity metrics and principles of ant colony optimization. Ultimately, algorithmic performance evaluations at every phase indicate that acquisition efficiency and image data value surpass traditional methodologies.

Dialogue State Tracking (DST) is a critical component of task-oriented spoken dialogue systems (SDS), tasked with maintaining an accurate representation of the conversational state by predicting slots and their corresponding values. Recent advances leverage Large Language Models (LLMs) with prompt-based tuning to improve tracking accuracy and efficiency. However, these approaches often incur substantial computational and memory overheads and typically address slot extraction implicitly within prompts, without explicitly modeling the complex dependencies between slots and values. In this work, we propose PUGG, a novel DST framework that constructs schema-driven prompts to fine-tune GPT-2 and utilizes its tokenizer to implement a memory encoder. PUGG explicitly extracts slot values via GPT-2 and employs Graph Attention Networks (GATs) to model and reason over the intricate relationships between slots and their associated values. We evaluate PUGG on four publicly available datasets, where it achieves state-of-the-art performance across multiple evaluation metrics, highlighting its robustness and generalizability in diverse conversational scenarios. Our results indicate that the integration of GPT-2 substantially reduces model complexity and memory consumption by streamlining key processes. Moreover, prompt tuning enhances the model’s flexibility and precision in extracting relevant slot-value pairs, while the incorporation of GATs facilitates effective relational reasoning, leading to improved dialogue state representations.