Harvard's Onnela Lab needed a research-grade iOS app that could passively capture smartphone sensor data — GPS, accelerometer, call and text metadata — while delivering ResearchKit surveys to study participants across multi-month clinical trials. We built Beiwe.
JP Onnela coined the term "digital phenotyping" — using the sensors already inside a smartphone to measure human behavior for clinical research. The insight: phones are always with patients, and they already capture the signals clinicians wish they had between appointments.
Traditional clinical studies rely on patients self-reporting symptoms during periodic office visits. Days or weeks of real behavior go unrecorded — mood shifts, sleep disruption, social withdrawal — all invisible to the study.
GPS traces map mobility patterns. The accelerometer captures physical activity. Call and text metadata reveal social interaction frequency. Screen on/off timestamps track sleep. None of it requires the participant to do anything.
Beiwe needed to run continuously across multi-week and multi-month studies, meet IRB privacy requirements, work across different clinical protocols — depression, schizophrenia, bipolar disorder, post-surgical recovery — and never drain the battery.
"Rocket Farm Studios was highly engaged throughout the project and offered a true collaborative design and development experience."
Building a passive data collection app that runs for months without the participant thinking about it is a fundamentally different engineering problem from building a consumer app someone opens for five minutes a day.
The core technical challenge was continuous background sensor collection on iOS without killing battery life. GPS polling, accelerometer sampling, and metadata logging all compete for power — and iOS aggressively throttles background processes. We had to find the right sampling intervals and batching strategies so the app could run for weeks without the participant noticing a battery hit or the phone's OS terminating the process.
On top of passive collection, the app needed to deliver active data collection through customizable surveys built on Apple's ResearchKit framework. Consent flows, survey scheduling, branching logic, and response encryption all had to work inside the same app that was quietly capturing sensor data in the background. The two modes — passive and active — could not interfere with each other.
Every piece of data had to meet the privacy and security standards demanded by academic Institutional Review Boards. That meant end-to-end encryption, no raw GPS coordinates leaving the device without obfuscation controls, and a backend architecture that kept participant identity separated from sensor streams. This wasn't a feature request — it was a prerequisite for the studies to be approved at all.
Beiwe ships as a native iOS app with three tightly integrated layers: passive sensor collection, an active survey engine powered by ResearchKit, and a secure backend that ties the data together for researchers.
GPS, accelerometer, call/text metadata, and screen state — captured continuously in the background without participant interaction. Battery-optimized sampling intervals keep the app invisible while running for months.
Customizable surveys delivered on configurable schedules using Apple ResearchKit. Consent flows, branching logic, and response validation — all built so researchers can define new study protocols without a code change.
End-to-end encrypted data pipeline separating participant identity from sensor streams. Built to satisfy IRB requirements across multiple institutions and study protocols running simultaneously.
"Rocket Farm Studios was unique in their deep understanding of the subtleties of application design, and they did a fantastic job developing a sophisticated and highly functional backend to the research app."
Beiwe didn't ship and sit on a shelf. It became a core piece of the Onnela Lab's research infrastructure — actively collecting data across multiple ongoing clinical studies, from mental health to post-surgical recovery.
A native iOS app combining passive sensor data collection and active ResearchKit surveys — built to run silently for months on participants' phones while meeting the strictest academic privacy requirements.
Beiwe powers studies across multiple clinical domains — depression, schizophrenia, bipolar disorder, post-surgical recovery — generating the continuous behavioral data that traditional methods simply cannot capture.
The app didn't just launch — it became a permanent part of the lab's toolkit. Multiple ongoing studies rely on the platform we built, and it continues to be the foundation for the Onnela Lab's digital phenotyping research.
"The Beiwe app continues to be used in multiple ongoing studies and has become a core piece of our lab's digital phenotyping research infrastructure."
More healthcare builds from the Rocket Farm portfolio — devices, networks, and patient ops.