Aspect-Aware Patient Feedback Analytics and Edge-Accelerated Time-Series Demand Forecasting for Remote Healthcare Delivery Platforms

• Billy Elly

  Lautech

  Author

The proliferation of remote healthcare delivery platforms has generated unprecedented volumes of unstructured patient feedback and high-frequency demand signals, yet existing analytics frameworks treat these data streams in isolation, limiting their utility for operational decision-making. This study addresses the research gap by proposing an integrated framework that combines aspect-aware sentiment analysis of patient feedback with edge-accelerated time-series demand forecasting. A DeBERTa-v3-based aspect classification model was developed to extract fine-grained sentiment across service quality dimensions from patient comments, achieving an accuracy of 89.4% on a corpus of 25,000 annotated patient reviews. Concurrently, a hybrid CNN-LSTM architecture was deployed at the edge to forecast service demand across 15-minute intervals, attaining 94.7% accuracy in predicting request volumes up to 48 hours in advance . The framework's key contribution lies in demonstrating that aspect-level sentiment patterns serve as leading indicators for demand fluctuations, enabling proactive resource allocation. Practical implications include reduced wait times, improved patient satisfaction, and bandwidth-optimized edge-cloud orchestration.

• pdf

Copyright (c) 2026 Billy Elly (Author)

This work is licensed under a Creative Commons Attribution 4.0 International License.