# SOTER: A Generative Time-Series Foundation Model for Wearable Human Physiological Signals

arXiv: https://arxiv.org/abs/2609.16804 · HTML: https://mlsanity.com/paper/2609.16804/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 18 of its 64 references that are in the map.
- First posted on arXiv 15 Sep 2026.

## Builds on (most cited first, 18 of 18)

- [Neural Ordinary Differential Equations](https://mlsanity.com/paper/1806.07366/) (2018)
- [A Time Series is Worth 64 Words: Long-term Forecasting with Transformers](https://mlsanity.com/paper/2211.14730/) (2022)
- [Neural Controlled Differential Equations for Irregular Time Series](https://mlsanity.com/paper/2005.08926/) (2020)
- [MOMENT: A Family of Open Time-series Foundation Models](https://mlsanity.com/paper/2402.03885/) (2024)
- [Time-MoE: Billion-Scale Time Series Foundation Models with Mixture of Experts](https://mlsanity.com/paper/2409.16040/) (2024)
- [Self-Supervised Contrastive Pre-Training For Time Series via Time-Frequency Consistency](https://mlsanity.com/paper/2206.08496/) (2022)
- [TiRex: Zero-Shot Forecasting Across Long and Short Horizons with Enhanced In-Context Learning](https://mlsanity.com/paper/2505.23719/) (2025)
- [PaPaGei: Open Foundation Models for Optical Physiological Signals](https://mlsanity.com/paper/2410.20542/) (2024)
- [Improving Medical Predictions by Irregular Multimodal Electronic Health Records Modeling](https://mlsanity.com/paper/2210.12156/) (2022)
- [Toward Foundation Model for Multivariate Wearable Sensing of Physiological Signals](https://mlsanity.com/paper/2412.09758/) (2024)
- [Aurora: Towards Universal Generative Multimodal Time Series Forecasting](https://mlsanity.com/paper/2509.22295/) (2025)
- [Self-Supervised Learning for Anomalous Channel Detection in EEG Graphs: Application to Seizure Analysis](https://mlsanity.com/paper/2208.07448/) (2022)
- [MIRA: Medical Time Series Foundation Model for Real-World Health Data](https://mlsanity.com/paper/2506.07584/) (2025)
- [Clustering Interval-Censored Time-Series for Disease Phenotyping](https://mlsanity.com/paper/2102.07005/) (2021)
- [PULSAR at MEDIQA-Sum 2023: Large Language Models Augmented by Synthetic Dialogue Convert Patient Dialogues to Medical Records](https://mlsanity.com/paper/2307.02006/) (2023)
- [T-Phenotype: Discovering Phenotypes of Predictive Temporal Patterns in Disease Progression](https://mlsanity.com/paper/2302.12619/) (2023)
- [TarDiff: Target-Oriented Diffusion Guidance for Synthetic Electronic Health Record Time Series Generation](https://mlsanity.com/paper/2504.17613/) (2025)
- [Self-explaining Hierarchical Model for Intraoperative Time Series](https://mlsanity.com/paper/2210.04417/) (2022)

## Built on by (most cited first, 1 of 1)

- [Training and Scaling Compute-Optimal Physiological Waveform Foundation Models](https://mlsanity.com/paper/2610.05649/) (2026)

## Abstract (arXiv)

Time-series foundation models have demonstrated strong cross-domain transfer, yet their common architectural assumptions remain poorly aligned with wearable physiological signals, which are multichannel, irregularly sampled, noisy, and governed by coupled continuous-time dynamics spanning distinct spectral scales. We present SOTER, a generative foundation model for wearable physiological time series that unifies cross-channel coupling, spectrum-guided expert specialization, and continuous-time latent evolution within a single pre-training framework. SOTER combines a spatial feature-aware backbone that models inter-signal dependencies, a power spectral density (PSD)-guided mixture-of-experts layer that routes representations to experts associated with fixed spectral bands through an inspectable, non-learned rule, and a neural controlled differential equation decoder that supports prediction and imputation at arbitrary timestamps. We pre-train SOTER on 226 billion time points from five public physiological datasets and evaluate the same pre-trained model across out-of-distribution zero-shot forecasting, frozen-encoder linear-probe classification, and continuous-time imputation on wearable benchmarks. SOTER achieves the best RMSE on 4 of 6 datasets and the best MAE on 5 of 6 in zero-shot forecasting, the highest average Macro-AUROC in classification, and the lowest imputation error on all six datasets at 75% missingness. It further remains robust to additive acquisition noise, matching or surpassing baselines evaluated on clean inputs even under the strongest corruption. These results indicate that domain-specialized foundation models for wearable physiology benefit from jointly modeling channel structure, spectral scale, and continuous-time dynamics.