OpenMHC LSM2 weekly-sparse
Reference checkpoint for OpenMHC. Window: weekly (10080 minute) sparse-decoder, 19 sensor channels.
A sparse-decoder extension of LSM2 for 7-day inputs. Instead of one transformer over the full week, this variant applies the frozen daily encoder (LSM2-daily) independently per-day and reconstructs the week through alternating day-local / cross-day decoder blocks (optionally calendar-aware via RoPE day embeddings). Lower compute than the dense weekly variant. Wrapped here for the OpenMHC imputation evaluation track.
Intended use
Intended use: research, methodological comparison, and reproduction of the OpenMHC benchmark.
Out of scope: diagnosis, screening, prognosis, prospective risk prediction, clinical decision support, or any other medical decision-making. OpenMHC's medical condition labels are self-reported and have no diagnosis dates, and benchmark performance does not establish clinical utility. Any diagnostic, prognostic, or screening claim requires independent validation using temporally aligned, clinically adjudicated outcomes in the target population.
Usage
from openmhc.imputers import LSM2WeeklySparseImputer
import openmhc
imp = LSM2WeeklySparseImputer.from_release("hf://MyHeartCounts/openmhc-lsm2-weekly-sparse")
results = openmhc.evaluate_imputation(imp, version="xs")
print(results.summary())
Requires the matching optional extras: pip install 'openmhc[lsm2,hf]'.
Pin a specific revision with the @ suffix:
LSM2WeeklySparseImputer.from_release("hf://MyHeartCounts/openmhc-lsm2-weekly-sparse@v1.0")
Provenance
- Origin: W&B artifact
MHC_Dataset/mhc-mae-ssl/mae-weekly-sparse-d4:v0 - Validation: val loss 0.2464
- Architecture hyperparameters are pinned in
openmhc_manifest.json.
License
Released under the BigScience OpenRAIL-M license. The use restrictions in its Attachment A apply to this checkpoint and any derivatives; among other things, they prohibit providing medical advice and medical results interpretation.
Citation
@misc{openmhc,
title = {OpenMHC: Accelerating the Science of Wearable Foundation Models},
author = {OpenMHC team},
url = {https://github.com/AshleyLab/OpenMHC}
}