Back to The Quantified Scientist

Apple Watch Ultra 4 - First Scientific Test!

The Quantified ScientistSeptember 18, 202613m
In a Nutshell

The Apple Watch Ultra 4 delivers near-perfect heart rate accuracy for indoor cycling (r=1.00) and weightlifting (r=0.99), but struggles during running with initial detection delays, dropouts, and a lower correlation (r=0.94) that ranks it below average compared to other devices. GPS tracking shows significant deviations during runs, unlike the Ultra 3, while step counting remains accurate with a median error of just 5 steps. The main upgrade is power efficiency enabling continuous 5-second heart rate monitoring rather than improved tracking accuracy over the Ultra 3.

AI-Generated Notes

These notes were generated by AI and may contain inaccuracies.

The test began with indoor cycling using the Polar H10 ECG chest strap as reference. The correlation value between the Apple Watch Ultra 4 and Polar H10 reached a rounded value of 1.00. The red line of the Apple Watch Ultra 4 tracked super close to the Polar H10 throughout the session with multiple intervals.

Running results showed a correlation of 0.94. During the first 5 minutes of the run, the Apple Watch Ultra 4 had issues detecting heart rate, resulting in points below the blue line indicating too low heart rate detection. There was also a deviation showing too high heart rate at one point. Dropouts occurred where no heart rate was recorded. Two different exports were tested - one with Straa and another app - both showing the same patterns. The Apple Fitness app displayed interval ranges that appeared between the Apple Watch measurement and the reference value.

The new heart rate sensor on the Apple Watch Ultra 4 has a different appearance from the previous version. The Ultra 4 is heavier than the Series 12, which may contribute to tracking issues. The new sensor may require optimization for signal processing across a larger population.

Weightlifting showed a correlation of 0.99. In the higher heart rate range, almost all points aligned with or were close to the blue line. The Apple Watch sometimes fully detected heart rate peaks but occasionally stopped detecting at the very peak. No severe dropouts occurred, though minor dropouts were possible. This performance ranked better than 94% of devices tested for weightlifting.

Indoor cycling performance ranked better than 98% of tested devices. Running performance ranked better than only 45% of devices due to the initial issue. Weightlifting performance ranked better than 94% of devices. Running emerged as the outlier in performance.

The Apple Watch Ultra 3 was tested with Watch OS 27 beta version over the previous 2 weeks. Similar performance was observed for spinning and weightlifting between Ultra 3 and Ultra 4. Biking comparison was not possible as no biking test was conducted for the Ultra 4.

Step counting was tested by manually counting approximately 5,000 steps, recording counts every 1,000 steps. The Ultra 4 results showed: first segment 1,000 manual vs 999 Apple Watch, second segment 6 overcount, third segment 5 overcount, fourth segment 8 overcount, fifth segment 4 overcount. Median difference was 5 steps. The Ultra was worn on the left arm while the manual counter alternated between right and left arms.

GPS tracking during running showed major deviations with signals not overlapping properly despite running in a rectangle shape at roughly 1 meter distance. The watch went through hedges and showed big deviations, including on the bottom right. This issue was not observed with the Apple Watch Ultra 3. The user was on the current non-beta version of Watch OS.

The main improvement in the Ultra 4 is not heart rate tracking quality but more continuous data through a more power efficient system. This enables continuous heart rate monitoring every 5 seconds and improved heart rate variability measurements throughout the day. The exercise tracking quality is expected to remain roughly the same as the Ultra 3, with potential improvements in removing dropouts. More weightlifting sessions with different exercise types are needed for thorough comparison. The step counting true positives appear good, though false positive testing remains pending. GPS performance may be affected by the specific unit, current firmware, or particular run conditions.

Keep The Quantified Scientist in your library

Save the videos and channels worth coming back to, and find them again in one place.