Individualized and Interpretable Sleep Forecasting via a Two-Stage Adaptive Spatial-Temporal Model

TL;DR

A hierarchical architecture, consisting of parallel 1D convolutions with varying kernel sizes and dilated convolution, which extracts multi-resolution temporal patterns-short kernels capture rapid physiological changes, while larger kernels and dilation model slower trends highlights its practical utility for real-world applications.

Abstract

Sleep quality impacts well-being. Therefore, healthcare providers and individuals need accessible and reliable forecasting tools for preventive interventions. This paper introduces an interpretable, individualized adaptive spatial-temporal model for predicting sleep quality. We designed a hierarchical architecture, consisting of parallel 1D convolutions with varying kernel sizes and dilated convolution, which extracts multi-resolution temporal patterns-short kernels capture rapid physiological changes, while larger kernels and dilation model slower trends. The extracted features are then refined through channel attention, which learns to emphasize the most predictive variables for each individual, followed by bidirectional LSTM and self-attention that jointly model both local sequential dynamics and global temporal dependencies. Finally, a two-stage adaptation strategy ensures the learned representations transfer effectively to new users. We conducted various experiments with five input window sizes (3, 5, 7, 9, and 11 days) and five prediction window sizes (1, 3, 5, 7, and 9 days). Our model consistently outperformed time series forecasting baseline approaches, including LSTM, Informer, PatchTST, and TimesNet. The best performance was achieved with a three-day input window and a one-day prediction window, yielding an RMSE of 0.216. Furthermore, the model demonstrated good predictive performance even for longer forecasting horizons (e.g., with a 0.257 RMSE for a three-day prediction window), highlighting its practical utility for real-world applications. We also conducted an explainability analysis to examine how different features influence sleep quality. These findings proved that the proposed framework offers a robust, adaptive, and explainable solution for personalized sleep forecasting using sparse data from commercial wearable devices.

Figures (6)

• Figure 1: Distribution of 23 input features and the sleep score target using violin plots with embedded box plots. Features span five categories: activity, heart rate, respiration, sleep stages, and stress. Blue: input features; Orange: prediction target (sleep score).
• Figure 2: Domain shifts for features from different subjects by PCA.
• Figure 3: Flowchart of our adaptive spatial-temporal model with optional two-stage domain adaptation
• Figure 4: True values vs predictions in validation and test set for subjects. Sleep quality prediction results across subjects representing typical patterns showing real trends (solid blue lines) compared to validation predictions (orange dashed) and test predictions (red dashed) with uncertainty bands (shaded regions). Color-coded bars indicate trend direction accuracy (green = correct, blue = incorrect).
• Figure 5: Comparison of ALL participants' , MSE, and by different models in terms of three-day input and one-day prediction window size.