Mobile phones and other connected devices are an increasingly common tools for researchers examining a range of medical and public health issues. Some, such as Bluetooth connected blood pressure cuffs or blood glucose monitors record and analyze information about medical conditions in the hopes of tracking patient outcomes over time and developing new interventions. Others, like such as personal breathalyzers and apps that use GPS and accelerometer sensors to track driving behavior, are working to cut down on the increasing number of fatalities and injuries caused by unsafe and often illegal driving behaviors.
Mobile technology provides researchers with new cost-effective ways to gain a much more nuanced understanding of risky behaviors. These technologies also provide an opportunity to develop personalized prompts or nudges and deliver them to participants in real time when they are most likely to prevent a risky behavior. For example, there are smartphone apps that can detect when a user might be at risk for drinking and driving, and deliver prompts encouraging them to use a rideshare service instead. However, the “big data” collected from these technologies in real time also opens up a plethora of ethical questions and challenges that plague health care professionals and medical ethicists alike.
“I wasn’t planning to examine ethical issues around having access to all this data when I went into this, but it was something I faced every time I presented a new study idea or project,” said M. Kit Delgado, MD, MS, an assistant professor of Emergency Medicine and Epidemiology, whose research focuses on leveraging smartphones, connected devices, and insights from behavioral economics for injury prevention. “The ethical questions we’re facing may not be readily apparent, but once you realize the amount of data you can gather and how granular it can get, it’s easy to see that the issues are significant and extremely challenging.”
Today, Delgado is working to develop smartphone based interventions aimed at reducing cellphone use while drinking and driving, but last year, we profiled his Way to Health study in which teens and their parents used smartphone apps to monitor and report cell phone use while driving over the course of 10 weeks. Data from the Way to Health study is pending publication, but Delgado says the data allows researchers to see which participants were using their phone while driving, how often they used their phone, and how fast the car was moving at the time.
“We look at some of the data from these participants and, we notice that some only used their phone a few times and when the car was moving very slowly or was stopped, but others are using their phones constantly while driving, and are doing so at high speeds,” Delgado said last week at a speaker seminar hosted by the department of Medical Ethics & Health Policy. “These participants – whether it’s the teens or the parents – are putting themselves and other motorists in harm’s way. I ask myself if I should do something with this knowledge beyond the planned interventions. Is this person going to hurt themselves or someone else?”
This question of duty to react to potential harm is one Delgado’s team faces often. Similar questions surface in Delgado’s newer work aimed at curbing drinking and driving.
“If we can see that someone has a blood alcohol level past the legal limit, how responsible are we for stepping in and preventing that person from falling and hurting themselves, or getting in a car and potentially hurting other people?”
One solution, Delgado said, is to plan workflows aimed at reducing harm when possible, such as the aforementioned prompts to use a ridesharing app in the case of participant who may be thinking about drinking and driving. Another option is to create a notification that would be sent to a previously designated friend or family member when a study participant appears to be endangering him or herself. Frequent audits of the data collected may also provide insights on high-risk participants who could benefit from additional interventions.
For now, Delgado says currently his team only pulls and reviews collected data on a weekly or monthly basis, making it impossible for researchers to step in in real time. However, as smartphone based interventions are programmed to respond to the collected data in real-time, more consideration will be need to be devoted to these challenges. For the time being at least, there are other issues around privacy and confidentiality that are perhaps even more pervasive.
For example, GPS tracking apps, such as the ones that are increasingly being used by auto-insurance companies to determine whether a customer is a safe driver, provide far more information about users, including where they’ve been. While the objective of these studies using GPS tracking apps is to better understand and ultimately encourage safe behavior, the data collected can be very sensitive and must be carefully handled and protected. For example, it may be possible to literally trace a participant’s daily path, potentially providing detailed insights into not only participant’s lives, but also the locations of people the participant interacts with. Having access to data at this level of granularity, Delgado says, raises major issues around privacy and confidentiality, with regards to the participant as well as others who have not consented to participate.
Best practices for this kind of information dictate that studies need to be designed in a way that ensures maximum protection for participants, and that means collecting the minimum amount of detail necessary, de-identifying data as much as possible and only publishing aggregate results, and being open and explicit with participants about what data will be collected, how it will be analyzed and what can be inferred.” For example, when using a smartphone app for research, the study team can create de-identified user accounts so the app vendor and other unauthorized entities cannot access the data or identify research participants. A further challenge, Delgado says, is that all these data on risky and illegal behaviors could potentially be subpoenaed in court. This issue alone makes it difficult to recruit people to participate in these studies. Delgado’s team and others working with these types of sensitive data obtain an NIH Certificate of Confidentiality which protects the researchers from disclosing any participant or study data from third parties, even if subpoenaed in court.
Depending on the study and information gathered, Delgado says there may be additional considerations, as well. For example, when conducting remote monitoring research with minors, Delgado says it’s important that researchers be honest with the participants about what their parents will see and have access to. Recruiting teens directly instead of having parents involved in that process also reduces the risk of coercion.
“Smartphone remote monitoring and interventions has major potential for tackling dangerous and pervasive public health issues, but researchers need to be thoughtful about how these studies are being designed and conducted,” Delgado said. “Transparency is crucial, but we also need to be careful stewards of the data that’s being collected, and make sure the potential benefits of collecting the data outweigh any potential risks.”