Wearables: privacy‑first integrations
- Developers are building privacy-first iOS apps that use HealthKit for on-device recovery scores and biological‑age estimates. - Radek Józefowicz showcased apps like Metrya and Capacity Gauge that keep processing local and preserve user privacy. - Broader interest in extracting dozens of mental and metabolic biomarkers from wearables, including Google CoDaS research, is driving richer multimodal integrations (x.com) (x.com).
A small crop of iPhone apps is turning Apple Health data into recovery and “biological age” scores without sending that data to outside servers. (developer.apple.com) (apps.apple.com) Apple’s HealthKit framework stores health and fitness data locally on Apple devices, and Apple says apps must get user permission for each data type they read or write. Apple also says the HealthKit store is encrypted when the device is locked. (developer.apple.com 1) (developer.apple.com 2) (developer.apple.com 3) That setup lets developers build apps that compute scores on the phone itself, using heart rate variability, sleep, resting heart rate, blood oxygen, workouts, and other signals already synced into Apple Health. Metrya, listed on the App Store by developer Radoslaw Jozefowicz, says its daily recovery score is calculated on-device from heart rate variability against a personal baseline, resting heart rate, sleep duration, sleep-stage quality, and a three-day trend. (apps.apple.com) (developer.apple.com) Capacity Gauge, another Jozefowicz app, applies the same local-processing pitch to work readiness rather than fitness. Its App Store listing says it combines Apple Health sleep and heart rate variability data with calendar load to produce a 0-100 morning score, and says “no data ever leaves your iPhone.” (apps.apple.com) The appeal is straightforward: wearables produce a steady stream of raw measurements, but most people want one number or a short explanation. Developers are using local models and rules to turn those measurements into plain-language summaries while avoiding the extra privacy risks that come with cloud uploads. (developer.apple.com) (apps.apple.com 1) (apps.apple.com 2) The same pattern is showing up in biological-age apps. My Bio Age says it computes a biological-age estimate on-device from 12 Apple Health metrics, including VO₂ max, heart rate variability, resting heart rate, walking heart rate, sleep, and active energy, while Biological Age Insight says its age and stress-resilience calculations stay on the phone. (apps.apple.com 1) (apps.apple.com 2) Research groups are pushing the signal count higher. Google Research said in July 2025 that its SensorLM wearable models were trained on 60 million hours of sensor data, and in November 2024 it described “wearable foundation models” aimed at learning from noisy consumer health data at scale. (research.google) (research.google) Google has also published work that tries to infer specific health risks from everyday devices. In 2025 it reported models for predicting insulin resistance from wearable data plus routine blood biomarkers, and in January 2026 it said smartwatches could estimate advanced walking metrics in a large validation study. (research.google) (research.google) Those projects point to a broader shift in wearables software: more apps are trying to combine movement, sleep, heart signals, temperature, labs, and calendar context into a single personal model. Apple’s rules still require explicit permission and user-facing disclosures, but the newest apps are increasingly selling privacy itself as a feature rather than a footnote. (developer.apple.com) (developer.apple.com) (research.google)
