This study presents a practical extension of prior work on skeleton-based nursing activity recognition by introducing a framework for evaluating nursing procedures, specifically Gastrostomy Tube Feeding (GTF). Previous studies using fixed-length time windows for activity segmentation face challenges in GTF procedures where action durations vary significantly: short micro-actions (e.g., Close the clamp) are often overshadowed by long macro-phases (e.g., Adjust the infusion rate), leading to reduced recognition accuracy for brief critical steps. To address this temporal heterogeneity, we adopted a BiLSTM with Attention mechanism incorporating variable time-step modeling, which accommodates diverse action durations without rigid segmentation. Additionally, we integrated a workflow-based assessment framework that evaluates nursing performance from the perspectives of Sequential Constraints, Safety Step Completion, and Precondition Checking, enabling structured and interpretable assessment of procedural correctness and safety compliance. Under the same label setting (including “Others”), the proposed variable-length temporal modeling improves recognition accuracy from 48% to 64%. Furthermore, by removing the ambiguous “Others” class and applying class-balanced focal loss, recognition accuracy increases to 71% on held-out subjects. The proposed scoring-based evaluation system then quantifies nursing skills via workflow adherence and safety validation, providing quantitative scores and structured violation feedback to support formative assessment in nursing education.

View full abstract

Diffusion models are increasingly being utilised to create synthetic tabular and time series data for privacy-preserving augmentation. Tabular Denoising Diffusion Probabilistic Models (TabDDPM) generate high-quality synthetic data from heterogeneous tabular datasets but assume independence between samples, limiting their applicability to time-series domains where temporal dependencies are critical. To address this, we propose a temporal extension of TabDDPM, introducing sequence awareness through the use of lightweight temporal adapters and context-aware embedding modules. By reformulating sensor data into windowed sequences and explicitly modeling temporal context via timestep embeddings, conditional activity labels, and observed/missing masks, our approach enables the generation of temporally coherent synthetic sequences. Compared to baseline and interpolation techniques, validation using bigram transition matrices and autocorrelation analysis shows enhanced temporal realism, diversity, and coherence. On the WISDM accelerometer dataset, the suggested system produces synthetic time-series that closely resemble real world sensor patterns and achieves comparable classification performance (macro F1-score 0.64, accuracy 0.71). This is especially advantageous for minority class representation and preserving statistical alignment with real distributions. These developments demonstrate that diffusion based models provide effective and adaptable solutions for sequential data synthesis when they are equipped for temporal reasoning. Future work will explore scaling to longer sequences and integrating stronger temporal architectures.

View full abstract