Modeling heart rate (HR) dynamics to reconstruct missing HR data in trail running.

Area of Science:

• Exercise Physiology
• Biomedical Engineering
• Wearable Technology

Background:

• Heart rate (HR) is a key metric for exercise intensity, but outdoor recordings suffer from artifacts and gaps.
• Existing reconstruction methods like linear interpolation (LI) are inadequate for trail running due to frequent intensity changes and ignored HR dynamics.
• A novel approach leveraging Global Navigation Satellite System (GNSS) data and HR dynamics is proposed for improved reconstruction.

Purpose of the Study:

• To develop and evaluate a new heart rate reconstruction method (HRMD) that integrates GNSS-derived energy expenditure and HR dynamics.
• To compare the accuracy and bias of HRMD against traditional linear interpolation (LI) during simulated trail running conditions.

Main Methods:

• 12 recreational trail runners performed trail running bouts, with simulated HR gaps (1-800 beats) at different running phases.
• GNSS data estimated energy expenditure, modeled HR dynamics using a first-order differential equation with individualized parameters.
• HRMD, including drift correction, was compared to LI and a basic HR model (HRM) using RMSE, mean error (ME), and limits of agreement (LoA).

Main Results:

• HRMD and LI showed similar low median RMSE (approx. 2 bpm), outperforming HRM (approx. 3 bpm).
• HRMD significantly outperformed LI in non-steady conditions (onset, switch) for gaps of 200+ beats, reducing mean error by up to 14 bpm.
• LI exhibited length-dependent bias (ME up to 14.6 bpm, LoA up to 20.3 bpm) at onset, while HRMD maintained ME near zero with LoA < 10 bpm.

Conclusions:

• Linear interpolation is insufficient for long, non-steady HR gaps common in trail running.
• The proposed HRMD method effectively reduces error and bias in challenging trail running scenarios by integrating GNSS data and HR dynamics.
• HRMD offers a practical solution for accurate HR reconstruction using readily available wearable technology, though generalizability requires further study.