From the Journals

App aims to detect respiratory failure in opioid overdoses


A new smartphone app under development seeks to detect the first moments of an overdose-related respiratory crisis and summon help before it’s too late.

Dr. Jacob E. Sunshine, University of Washington, Seattle

Dr. Jacob E. Sunshine

“We’re hoping a device that most people carry around could be transformed into technology that could save your life in an overdose,” said anesthesiologist Jacob (Jake) E. Sunshine, MD, an assistant professor with the University of Washington, Seattle, and coauthor of a study about the app’s development.

The ultimate goal is “to provide a harm reduction system that can automatically connect naloxone-equipped friends and family or emergency medical services to help prevent fatal overdose events,” Rajalakshmi Nandakumar, and her associates wrote in the study, published in Science Translational Medicine.

An estimated 70,000 people in the United States died from drug overdoses in 2017, according to a 2018 data brief from the Centers for Disease Control and Prevention. On an age-adjusted basis, the overdose death rate in 2017 was more than three times higher than in 1999.

The app, which builds on previous work aimed at detecting disordered breathing in sleep apnea, uses a “short-range active sonar system” to detect respiration in a person within the distance of about 3 feet. The approach is similar to the echolocation strategy used by a dolphin or bat, Dr. Sunshine said, and relies on sending out an audio tone that humans cannot hear.

The app’s microphone detects an “audio reflection” of the tone after it bounces off a nearby person’s body and then analyzes it to calculate the distance to the person’s chest. “We’re able to use those distances to measure when someone is taking a breath, and when they’re not taking a breath,” said Dr. Sunshine, who conceptualized the study.

If a disordered breathing pattern is detected, the app is designed to send a text message with a GPS-pinpointed location to a prespecified contact, Dr. Sunshine said. Or the app could be set to call 911.

In the study, the investigators tested the app’s algorithm at a supervised injection facility – a space designed to allow users to inject illicit drugs safely – in Vancouver. They tested the app on 94 drug users as they injected themselves; half of the users “experienced clinically important respiratory depression,” and two needed to be treated by clinic staff for overdose. Both users survived, reported Ms. Nandakumar, a PhD candidate at the University of Washington, Seattle; Shyamnath Gollakota, PhD, an associate professor at the university; and Dr. Sunshine.

The new study found that the app detected cessation of breathing for 10 seconds or longer 95.9% of the time (95% confidence interval, 86.0%-99.5%) with 97.7% specificity (95% confidence interval, 88.2%-99.9%). However, the app was less adept at identifying respiratory depression (respiratory rate equal to or less than 7 breaths per minute): The investigators reported 87.2% sensitivity (95% CI, 74.2%-95.1%) and 89.3% specificity (95% CI, 76.9%-96.4%).

Ms. Nandakumar and her associates also tested the app’s algorithm on patients undergoing anesthesia. It correctly detected disordered breathing in 19 of 20 patients.

It’s not clear how the app would work in environments full of breathing people and, potentially, breathing animals such as pets. And the app has clear limitations. Since it needs to be able to bounce audio signals off a user’s chest, it will not work if a phone is in a pocket or if a user is face down, turns around, or wanders off.

However, the app can detect sudden changes in motion, Dr. Sunshine said, and investigators are developing a way to require users to check in with the app in certain situations that might signal trouble.

“For harm reduction interventions to be efficacious, further studies with participant feedback and human factor testing are needed to ensure that the system meets the needs, values, and preferences of people who use opioids, in addition to establishing the system’s safety vis-à-vis its potential to encourage moral hazard,” the investigators wrote in the article.

The next steps are to refine the app’s user interface and figure out how to connect it to the 911 emergency-response system, Dr. Sunshine said. Meanwhile, researchers have created a company to develop the product. “We’re going to do additional development through that entity and seek [Food and Drug Administration] approval,” Dr. Sunshine said. The investigators do not plan to charge users for the product, which can be used on iPhones and Androids, he said.

The study was funded by the Foundation for Anesthesia Education and Research, the National Science Foundation, and the University of Washington’s Alcohol and Drug Abuse Institute. Dr. Sunshine, Ms. Nandakumar, and Dr. Gollakota are inventors on a provisional patient application related to the project, and all have equity stakes in a company that is developing the technology. Dr. Gollakota is a paid consultant to Jeeva Wireless and Edus Health.

SOURCE: Nandakumar R et al. Sci Transl Med. 2019 Jan 9;11(474). doi: 10.1126/scitranslmed.aau8914.

Next Article: