mHealth Data Adds Detail, Meaning to Population Health Programs

Smartphones have long been considered an ideal mHealth tool for personalized medicine, capable of collecting individual data and pushing out targeted reminders and information. Now that data is being used to power population health programs, with strong success.

When a school district superintendent learned that a majority of text messages regarding eating disorders was sent on Wednesday mornings, for example, he alerted his staff to make that an emphasis of education and outreach on that day. And in Louisville, Ky., data collected from smartphones used by asthma patients was combined with climate and sociological information to help city health officials pinpoint when and where air quality would trigger asthma attacks.

“Each message that comes in to us is a data point,” says Baylee Greenberg, chief operations officer of Crisis Text Line (, on online crisis hotline that has handled some 52 million messages since its founding in 2014 and uses those messages for both personal and population health outreach. And that data, she says, can help in tailoring effective outreach methods.

Greenberg was part of a panel discussion at last week’s Connected Health Conference ( in Boston that focused on using personal mHealth data for population health programs – the idea being that programs can collect and analyze enough data from individuals to spot trends and push information out to larger groups of people.

The tactic has been used in the past in public health campaigns targeting smoking, drug and alcohol abuse, flu and other virus outbreaks and prenatal and maternal health. In each case, programs like Text2Quit ( or Text4Baby ( identify the population targeted, developed messages and push them out through text or e-mail.

Now those programs are becoming more effective, thanks to the data being gathered by mHealth devices – including sensor-embedded smartphones – and artificial intelligence platforms that can crunch large amounts of data and spot trends.

One of the better examples of this is Propeller Health’s Air Louisville program (

The Wisconsin-based mHealth company, which develops smartphone-based care management tools for people with asthma and COPD, launched a two- year program in 2015 with city and federal health officials in Louisville to collect and analyze data from more than 1,100 patients. The project also collected data on environmental conditions - nitrogen dioxide, particulate matter, ozone, sulfur dioxide, pollen levels, temperature, humidity and wind speed – and combined that with Environmental Protection Agency air quality readings.

“We were crowdsourcing the data,” says Meredith Barrett, PhD, Propeller Health’s vice president of research. And in collecting that data, she said, “we did see really important improvements in clinical outcomes.”

According to city officials and Propeller Health, the Air Louisville program helped 82 percent of those with asthma or COPD see a reduction in inhaler use, thanks to targeted information pushed to their smartphones that identified dangerous air conditions and helped them to manage their health better. Almost 30 percent improved overall control of their asthma, officials said, and 19 percent reduced their nighttime distress, thus improving quality of sleep.

City and county officials used that data to determine what types of weather conditions cause the most distress - if ozone levels exceed 70 parts per billion (ppb), for instance, Jefferson County could expect to see more than 65,000 asthma rescue inhaler uses.

This level of medication use equated to healthcare costs of $129,000 in a single day. In 2016, ozone exceeded the 70 ppb limit on 19 days, which translated to roughly $2.4 million in healthcare costs, Air Louisville estimated.

Bennett said the data is being used by city and county officials to plan ahead for bad air days, and to plan strategies for improving air quality, such as planting more trees or reducing traffic congestion.

Propeller Health also used that data, she said, to create an Air Forecast mHealth tool, which enables users to predict the risk of asthma symptoms based on their location.

Aside from using data collected from smartphone-based sensors, other participants in last week’s panel talked about how texts and other types of smartphone use can be analyzed for public health concerns.

John Brownstein, PhD, chief innovation officer at Boston Children’s Hospital and a professor at Harvard Medical School, pointed out that health systems can mine data collected in Google searches to pinpoint where virus and disease outbreaks may be occurring or ready to occur – giving public health officials, for instance, a chance to head off or at least contain a flu outbreak before it strains the limits of local health systems.

Earlier this year, Boston Children’s put that theory to the test in a project in China, where they used an mHealth thermometer to help predict season al flu outbreaks (

In that project, data collected by the iThermometer, an FDA-approved temperature sensing patch worn under the arm, and online questionnaires delivered through an accompanying mHealth app called Thermia was used by the National Health and Family Planning Commission (NHFPC) of the People's Republic of China to identify trends ahead of influenza outbreaks.

"Delays in clinically reported data and lack of data availability contribute to the challenges of identifying outbreaks rapidly," Brownstein said in a May 2017 press release on the project, which was profiled in the American Journal of Public Health. “As a result, we have more and more opportunities to use real-time, low-cost digital solutions like Thermia to improve disease surveillance."

"Collectively we are still coming to terms with the data deluge from wearable devices, but it is imperative that we begin to generate value from this data," added the study's senior author, Jared Hawkins, PhD, director of informatics at Boston Children's Innovation and Digital Health Accelerator (IDHA). "From a public health perspective -- as we have shown with this latest study -- there is enormous potential for tapping this data for research, surveillance and influencing policy.”

The Truth Initiative, a national program targeting smoking, has used text messaging programs to target younger populations before they take their first puff and those looking to quit. Now they’re using data culled from past efforts to identify when those who have quit are most likely to feel stressed. They’ll use a smartphone’s location sensors, for example, to identify when a user inside a building is going outside, perhaps to sneak a cigarette.

Jesse Saul, PhD, director of client success and innovations for the Truth Initiative (, says the program is constantly modifying its tactics to make better use of data and to pinpoint the best way to reach its target audience. Today’s younger generations text more than they use the phone to make a call, she points out.

Sometimes data used in public health campaigns can be tuned around to aid individuals.

Greenberg said the language used in texts can be used to help crisis counselors connect with someone at risk of suicide.

By analyzing all the data collected in past texts, she said, Crisis Text Line has surmised that someone using the word “Ibuprofen” in texts is likely more serious about considering suicide than someone using the word “suicide.” And for counselors, asking a caller’s name works best between the fifth and eighth message.

The challenge, the panelists said, lies in taking all that data coming in and making it useful. Too much information, or information not presented in an attractive format, can push a certain population away from text messages or e-mails designed to improve their health and wellness.

“Artificial intelligence needs real intelligence as well,” pointed out the panel’s moderator, Alan Labrique, PhD, an associate professor at Johns Hopkins

University and director of the JHU Global mHealth Initiative.” 

Written by: Eric Wicklund

Published in: mHealthIntelligence