PAGE: Domain-Incremental Adaptation with Past-Agnostic Generative Replay for Smart Healthcare

TL;DR

PAGE addresses domain-incremental disease detection in wearable-based healthcare by eliminating the need to store past-domain data. It combines past-agnostic generative replay with a Gaussian Mixture Model to synthesize informative data from the new domain and replay it with real data, preserving learned knowledge without data retention. The approach is extended with Extended Inductive Conformal Prediction to provide confidence and credibility for each detection, achieving competitive performance while reducing clinical workload. The work demonstrates scalability, privacy preservation, and practical interpretability, offering a viable path for continual learning in smart healthcare across multiple domains.

Abstract

We propose PAGE, a domain-incremental adaptation strategy with past-agnostic generative replay for smart healthcare. PAGE enables generative replay without the aid of any preserved data or information from prior domains. When adapting to a new domain, it exploits real data from the new distribution and the current model to generate synthetic data that retain the learned knowledge of previous domains. By replaying the synthetic data with the new real data during training, PAGE achieves a good balance between domain adaptation and knowledge retention. In addition, we incorporate an extended inductive conformal prediction (EICP) method into PAGE to produce a confidence score and a credibility value for each detection result. This makes the predictions interpretable and provides statistical guarantees for disease detection in smart healthcare applications. We demonstrate PAGE's effectiveness in domain-incremental disease detection with three distinct disease datasets collected from commercially available WMSs. PAGE achieves highly competitive performance against state-of-the-art with superior scalability, data privacy, and feasibility. Furthermore, PAGE can enable up to 75% reduction in clinical workload with the help of EICP.

Figures (7)

• Figure 1: The schematic diagram of our PAGE strategy.
• Figure 2: The top-level flowchart of our PAGE strategy.
• Figure 3: The flowchart of EICP.
• Figure 4: The training loss of each training data instance in the CovidDeep dataset covidD across 150 epochs.
• Figure 5: The DNN architecture used in experiments.