Paper, code and model

Resources

Everything needed to read, reproduce and build on Sensori, collected in one place.

Paper

The manuscript describing the model, evaluation and implications for passive health monitoring.

Read the paper

Code

Preprocessing, model inference and downstream evaluation code for research use.

Browse the code

Model

Pretrained weights and documentation for generating minute- and day-level representations.

Download weights

Citation

Cite Sensori

If you find this paper or code useful in your research, please consider citing our paper.

@misc{wang2026learning,
  title         = {Learning Human Health and Diseases from 24-hour Wrist Movement},
  author        = {Wang, Yong and McGagh, Dylan and Broomberg, Katya and Zhang, Zizheng and Carter, Jonathan and Naushad, Junayed and Brocklebank, Laura and Sun, Yang and Nicholson, George and Sun, Dianjianyi and Yu, Canqing and Lv, Jun and Barnard, Maxim and Lam, Hubert and Steptoe, Andrew and Eyre, David W. and Li, Liming and Chen, Zhengming and Wray, Naomi and Denaxas, Spiros and Collins, Gary S. and Du, Huaidong and Doherty, Aiden and Yuan, Hang},
  year          = {2026},
  eprint        = {2608.29494},
  archivePrefix = {arXiv},
  primaryClass  = {cs.LG},
  url           = {https://arxiv.org/abs/2608.29494}
}

Data access

Built across four population studies.

UK Biobank and China Kadoorie Biobank data are available to approved researchers through their respective access platforms. ELSA is available through the UK Data Service; NHANES and the human activity recognition benchmarks are publicly available from their original sources.