Measurement error and heart rate data quality explained

This article explains what Measurement error (%) means and how heart rate data quality affects the physiological metrics in Firstbeat Sports: why heart rate recordings contain errors, how Firstbeat Sports corrects them and how to interpret the percentage.

Why heart rate recordings contain errors

The Firstbeat Sports Sensor records heart rate from the electrocardiogram (ECG), the electrical signal of the heart. The Sensor detects each heartbeat by finding a specific spike in the signal, the R-peak. These factors can disturb the detection:

• Movement and muscle noise: the electrodes shift slightly, or nearby muscle activity creates electrical interference that looks like a heartbeat.
• Poor electrode contact: sweat, dry skin or a loose Sensor weakens the signal.
• Irregular heartbeats: naturally occurring irregular beats can be harder for the algorithm to classify correctly.

No heart rate recording is perfectly clean. Some noise and misread beats are a normal, expected part of every recording, not a sign that something is broken.

How Firstbeat Sports corrects errors

Firstbeat Sports applies artifact correction to every recording before any results are calculated. The artifact correction algorithm scans the raw data, identifies heartbeats that are unlikely to be physiologically real, and corrects or removes them. The correction uses information from the surrounding heartbeats and normal heart physiology. Artifact correction is the quality control between the raw data and the results, and it keeps heart rate monitoring reliable even when a recording contains some noise.

What Measurement error (%) means

Measurement error (%) is the proportion of heart rate data that Firstbeat Sports identifies as unreliable and corrects during the analysis. When several recordings are merged, the gaps between the recordings are also included in the percentage. The same metric can also appear as Missing or Poor Quality Data (%).

• Lower Measurement error means higher confidence in the results.
• Higher Measurement error means a greater risk that physiological metrics are affected.

Measurement error is a quality indicator, not a pass-or-fail assessment. Measurement error gives context for interpreting physiological metrics and shows how much confidence you can place in the results.

How to interpret Measurement error

• Good (green): less than 15%. Physiological metrics can generally be trusted with confidence.
• Moderate (yellow): 15–30%. Most metrics are only minimally affected, but high-intensity metrics may be affected more.
• Poor (red): more than 30%. Metrics related to training load and high-intensity exercise may be substantially affected. Evaluate the results case by case.

Resting measurements

Resting heart rate variability (HRV) analysis has stricter data quality requirements than exercise monitoring. HRV calculations rely on the precise timing of individual heartbeats, so corrected data cannot fully replace genuine heartbeats. For resting measurements, aim for less than 15% Measurement error whenever possible.

How data quality affects the results

Firstbeat studied the impact of heart rate data quality on about 1.4 million football training sessions. Recordings with less than 15% Measurement error produced very similar physiological results, while larger errors increasingly affected metrics related to training load and high-intensity exercise. When Measurement error exceeds 30%, the impact is most noticeable in metrics that depend on accurately capturing periods of high heart rate.

Metrics most affected when Measurement error exceeded 30% (average impact):

• Time in heart rate zone 5: about 27% lower
• TRIMP (training impulse): about 20% lower
• Aerobic and Anaerobic Training Effect: about 8–11% lower

Metric less sensitive to poor data quality:

• Average heart rate: about 3% lower

What to expect in daily use

• Easy and moderate sessions are usually less sensitive to data quality issues.
• Sprint training, intensive intervals and match play are more likely to be affected when Measurement error is high.
• Metrics based on time at very high heart rates are the most vulnerable to poor data quality.
• Data quality may be worse at the beginning of a measurement because of dry skin and clothes, especially if the chest strap is not moistened properly.
• A poor-quality "tail" can appear at the end of a measurement after the chest strap is already taken off. The tail raises the Measurement error percentage but has no practical impact. You can trim the tail off, as described in How to trim a measurement in Sports Cloud.
• If data quality drops suddenly during a session, a Sensor pin may have detached from the strap, or the chest strap may have slipped too low.

Key takeaways

• Every heart rate recording contains some errors, and Firstbeat Sports corrects them automatically with artifact correction.
• Measurement error (%) shows how much of the heart rate data was unreliable and corrected. Gaps between merged recordings are included.
• Less than 15% is good, 15–30% is moderate and more than 30% is poor.
• Above 30%, time in heart rate zone 5, TRIMP and Training Effect are affected the most; average heart rate is affected the least.
• For resting HRV measurements, aim for less than 15%.
• To improve data quality, follow the steps in High error percentage in Sensor data.