Scholarlink

<p>This Data Descriptor presents a fully open, multi-modal dataset for estimating vertical ground reaction forces (vGRF) from consumer-grade Apple Watch sensors with laboratory force plate ground truth. Ten healthy adults (aged 26–41 years) performed five activities: walking, jogging, running, heel drops and step drops, while wearing two Apple Watch devices positioned at the left wrist and waist. The dataset comprises 492 validated trials with time-aligned inertial measurement unit (IMU) recordings (≈100 Hz) and force plate vGRF (Force_Z sampled at 1000 Hz). The release includes raw and processed time series, trial-level metadata, quality-control flags, and machine-readable data dictionaries. Dataset integrity is ensured through trial-level matching manifests linking recordings across modalities using stable identifiers. The primary trial manifest contains 395 triad-complete trials (wrist + waist + force-plate data; 80.3% of total trials) suitable for comprehensive cross-sensor analyses and reproducible downstream model evaluation. Sensor availability across the full dataset is high: wrist data are available for 461 trials (93.7%), waist data for 451 trials (91.7%), and force plate data for 449 trials (91.3%). Dataset quality is characterised using a three-phase cross-sensor plausibility and consistency framework (waist → force plate, wrist → waist, wrist → force plate), alongside repeatability metrics for peak vGRF (intraclass correlation coefficient 0.871–0.990) and systematic checks of force ranges and trial completeness. A Monte Carlo sensitivity analysis indicates that correlation-based validation metrics are robust to single-sample timing perturbations at the IMU sampling resolution (mean absolute change correlation ≈0.010 for comparisons with |r| ≥ 0.2 under ±10 ms perturbations, equivalent to ±1 sample at ~100 Hz). All data are available under CC BY 4.0; analysis scripts are archived with the release and mirrored on GitHub. This resource supports reproducible research in wearable biomechanics, benchmarking machine learning vGRF estimation models, and systematic investigation of sensor placement effects using widely available consumer wearables.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Multi-Modal Dataset for Ground Reaction Force Estimation Using Consumer Wearable Sensors

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年4月10日

查看全文
  • Parvin Ghaffarzadeh,
  • Debarati Chakraborty,
  • Koorosh Aslansefat,
  • Ali Dostan,
  • Yiannis Papadopoulos

摘要

This Data Descriptor presents a fully open, multi-modal dataset for estimating vertical ground reaction forces (vGRF) from consumer-grade Apple Watch sensors with laboratory force plate ground truth. Ten healthy adults (aged 26–41 years) performed five activities: walking, jogging, running, heel drops and step drops, while wearing two Apple Watch devices positioned at the left wrist and waist. The dataset comprises 492 validated trials with time-aligned inertial measurement unit (IMU) recordings (≈100 Hz) and force plate vGRF (Force_Z sampled at 1000 Hz). The release includes raw and processed time series, trial-level metadata, quality-control flags, and machine-readable data dictionaries. Dataset integrity is ensured through trial-level matching manifests linking recordings across modalities using stable identifiers. The primary trial manifest contains 395 triad-complete trials (wrist + waist + force-plate data; 80.3% of total trials) suitable for comprehensive cross-sensor analyses and reproducible downstream model evaluation. Sensor availability across the full dataset is high: wrist data are available for 461 trials (93.7%), waist data for 451 trials (91.7%), and force plate data for 449 trials (91.3%). Dataset quality is characterised using a three-phase cross-sensor plausibility and consistency framework (waist → force plate, wrist → waist, wrist → force plate), alongside repeatability metrics for peak vGRF (intraclass correlation coefficient 0.871–0.990) and systematic checks of force ranges and trial completeness. A Monte Carlo sensitivity analysis indicates that correlation-based validation metrics are robust to single-sample timing perturbations at the IMU sampling resolution (mean absolute change correlation ≈0.010 for comparisons with |r| ≥ 0.2 under ±10 ms perturbations, equivalent to ±1 sample at ~100 Hz). All data are available under CC BY 4.0; analysis scripts are archived with the release and mirrored on GitHub. This resource supports reproducible research in wearable biomechanics, benchmarking machine learning vGRF estimation models, and systematic investigation of sensor placement effects using widely available consumer wearables.