Researchers from Florida International University have shown that smartwatches from leading brands overestimate calorie burn by up to 25%, with the margin of error increasing alongside the user's level of obesity.

The study evaluated the Apple Watch Series 8, Garmin Forerunner 955, Samsung Galaxy Watch 5, and Fitbit Sense 2, comparing their readings against medical-grade equipment. Garmin showed an average discrepancy of 69 kcal, Samsung 57 kcal, and Apple 22 kcal. Fitbit had an average deviation of 3 kcal but also exhibited extreme fluctuations and data loss.

The algorithms were calibrated primarily using lean individuals, which explains the greater discrepancies observed in overweight users.

Conclusion: Commercial smartwatches remain insufficiently accurate for precise individual calorie budgeting or rigorous clinical/research monitoring; users should treat caloric metrics as relative general trends rather than absolute values.

• Shift to Software and Firmware: The major differences in modern smartwatches and health trackers are increasingly driven by software and firmware updates rather than hardware improvements [00:00:08]. Hardware has matured to a point where new generations often show little change in direct performance [00:03:21].
• Three Levels of Health Metrics:
  • First Order: Direct measurements calculated from raw sensor data, such as heart rate and GPS tracking [00:00:35].
  • Second Order: Derived metrics calculated by combining first-order data, such as sleep stages and sleep apnea detection [00:00:43].
  • Third Order: Complex predictions requiring massive amounts of data and advanced AI, including disease, injury, and recovery forecasting [00:10:36].
• Algorithm vs. Hardware Impact: Video data tracks how major firmware updates can vastly alter or improve metrics (e.g., Oura Ring's Sleep Staging 2.0 algorithm significantly altered deep sleep data and lowered day-to-day variance, whereas the Oura Ring 4 hardware launch yielded virtually no metric changes) [00:02:42], [00:03:21].
• AI Expertise and Performance: Major tech companies like Apple and Google (Pixel) outperform dedicated sports watch brands in live heart rate tracking [00:07:39]. This is attributed to their superior AI expertise and software infrastructure to filter out noisy raw PPG sensor data [00:08:01].
• The Rise of Foundation Models: The future of health tracking relies on "Foundation Models" (similar to the AI architectures behind ChatGPT), which analyze wearable data over time to predict the long-term likelihood of developing diseases like heart disease or Alzheimer's [00:11:05], [00:12:11].
• Socio-Ethical Concerns: The emergence of third-order metrics introduces complex challenges regarding data privacy, accuracy, security, and potential societal gaps between who can and cannot afford this technology [00:12:58].