UCSF study: Smartphone geofencing can detect hospitalizations with 65 percent accuracy

By Jonah Comstock
Share

A new study from UCSF -- one part of the organizations's extensive Health eHeart study -- shows that geofencing (using cellphone triangulation and GPS data to determine a person's location) can be used to track hospitalization with 65 percent accuracy and reasonable sensitivity. The findings were recently published in the American Heart Association journal Circulation: Cardiovascular Quality and Outcomes.

While theoretically EHR records should make it trivial to know when a patient has been hospitalized, in fact a combination of lack of interoperability and data privacy concerns mean that it can be hard for healthcare providers to get this information, especially if a patient visits an unfamiliar facility in an emergency. UCSF researchers set out to find out whether an app on a patient's smartphone that automatically detects when they're in a hospital could improve the situation.

“As the prevalence of chronic disease increases with the aging population, there is a need for improved health care monitoring and more timely treatment between encounters with health care providers,” Senior Author Dr. Gregory Marcus, director of clinical research in the UCSF Division of Cardiology, said in a statement. “Our app only had moderate accuracy, but this approach could revolutionize not only the way we ascertain if someone is sick, but also could be relevant to geofencing any location for a number of health-care related studies or interventions.”

The study had two arms of participants: 22 in-person participants and 3443 remote participants. All the participants installed an app on their phone, created by Ginger.io, that would detect if they were in a hospital and send this information to researchers. Patients were sent a questionnaire to confirm that they had been admitted to a hospital. Researchers than checked the accuracy against EMR data -- data from UCSF's system for the in-person cohort and data retrieved by calling other hospitals for the remote cohort. Participants signed HIPAA release forms that allowed researchers to get this data from other hospitals.

The app accurately detected visits 65 percent of the time. There were several reasons the app might not be accurate -- some were technical, but others were logistical, like registering false positives for patients who worked at or near a hospital or who were visiting a friend or relative in the hospital. Privacy was also a concern and a reason a number of patients chose not to participate. Overall, though, researchers believe the technology is promising, that accuracy can be improved with usability upgrades and a learning algorithm to account for non-medical hospital visits, and that privacy concerns are somewhat mitigated by an app that reports only whether a patient is in a hospital and doesn't record or report their location all the time.

A few startups have already sprung up around using this technology with patients who are at risk for re-hospitalization. Boston-based PatientPing works with hospitals to detect –  using hospitals' checkin workflows – when patients enter their facility and send a message to that patient's regular provider. Position Health, a New York City startup, actually uses geofencing toward the same purpose.

The UCSF researchers believe that the app has potential not just for healthcare, but for community health – detecting when patients enter grocery stores, fast food restuarants, or pharmacies and sending them context-aware notifications that could promote healthy lifestyle choices.Air Jordan Trainer Essential