We've joined our RTI Health Solutions colleagues under the RTI Health Solutions brand to offer an expanded set of research and consulting services.
Harnessing Digital Technology To Personalize Patient Experience
Personalization trend extends to healthcare
From television streaming services to music-playing apps, many people have come to expect and rely upon personalized content and individual customization.
Increasingly, the healthcare industry, too, is exploring the potential of "digital phenotyping," or using data from personal digital devices and other sources to better understand patient behaviors. This information can be used to assess conditions, customize approaches, and develop new interventions.
As interest—and artificial intelligence (AI) capacity—grows it's important to approach healthcare information gathering with caution, keeping in mind the numerous ethical and privacy considerations.
Gathering and analyzing digital footprints for health
As we navigate our day-to-day lives, smartphones and computers collect data on our actions, movements, and decisions. One app might monitor someone's step counts while another compiles data on computer usage patterns. The digital footprints we leave behind can reveal patterns and changes in behaviors. For example, a sudden jump in smartphone usage late at night might indicate sleeping problems.
In medicine, “digital phenotyping" refers to the practice of bringing together these digital traces from places such as smartphones and electronic health records (EHR) to create a better sense of a patient's distinct behaviors and needs. The hope is that the practice could lead to more customized, personal medical interventions driven by a better understanding of someone's distinct personality, behavior, and health.
“By combining 24/7 data—on location, movement, email and text communications, and social media—with brain scans, genetics, genomics, neuropsychological batteries, and clinical interviews, researchers will have an unprecedented amount of objective, individual-level data," notes researchers in a paper published in the Journal of Medical Internet Research. “Analyzing these data with ever-evolving artificial intelligence could one day include bringing interventions to patients where they are in the real world in a convenient, efficient, effective, and timely way. Yet, the road to this innovative future is fraught with ethical dilemmas as well as ethical, legal, and social implications."
Which aspects of health can digital tools capture?
With smartphone use at record highs, personal sensing offers tremendous potential in clinically managing and researching health conditions like depression.
Devices can gather data on someone's motion, heart rate, and other physiological variables. They can also capture behaviors that inform psychiatric assessments such as sociability, sleeping patterns, and physical activity. GPS apps can identify how long someone spends at home using location data, which may indicate social withdrawal or lack of energy.
In fact, remote sensing may even generate more objective and frequent measurements of mood and other markers, compared to relying on an individual to document and recall their previous experiences, notes the authors of a systematic review of these digital tools published in npj Digital Medicine.
Going beyond basic data capture for health data
These digital health platforms can generate tremendous amounts of user data, though not all of it is typically used. In an article on harnessing “your digital exhaust," Wellframe describes how health plans can use this excess data to improve member experiences. The company highlighted the following types of useful data:
- Biometric data: This includes measurements such as a user's body weight, blood pressure, physical activity, and blood glucose. The figures can also be sorted by age, location, or type of program, which could better educate specific population needs.
- Social Determinants of Health (SDoH) data: Gathering information on the non-medical factors that impact someone's health could influence how providers personalize care. This data might be gathered via digital surveys, or even through a secure chat function in which members might share lifestyle data, such as challenges seeking transportation for medical visits.
- Messaging and conversation data: Secure messaging often uncovers relevant information on factors influencing someone's health. Natural language processing (NLP) can be a vital tool in sifting through the data and flagging areas that could require care team intervention. In addition, a “sentiment analysis" might offer insights on how a member feels about care team interactions, identifying frustrations that could impact care plan adherence. This information could elucidate frustrations in their care plan that might influence adherence.
- Digital Assessments: Surveys can be sent via a health plan app, giving people the chance to complete health information at a time that's convenient and private for them. The collected information could lead to a patient sharing previously nondisclosed health concerns.
Connecting people, pinpointing diagnosis
Gathering and analyzing massive amounts of data through EHRs could also help with understanding how a condition might manifest in patients and whether there are gaps in diagnosing criteria. A study recently published in the Journal of Neurodevelopmental Disorders looked at the role of machine learning in autism spectrum disorder (ASD), a diagnosis that is rooted in observations and reports of an individual's characteristics.
In their study, researchers used NLP techniques to identify and curate more than 3,000 terms from raw clinical notes in EHRs for people with an autism diagnosis. Better understanding the prevalence and interplay of these terms could also be used in a diagnostic pipeline to differentiate people with autism from other psychiatric disorders.
“Our ASD phenotype ontology can assist clinicians and researchers in characterizing individuals with ASD, facilitating automated diagnosis, and subtyping individuals with ASD to facilitate personalized therapeutic decision-making," authors wrote.
Wellness personalization could yield better outcomes
Increased customization also offers tremendous potential in the wellness field. Too often, digital health tools—such as wearable devices that track physical activity—falter because they use a one-size-fits all approach, explains an article in the Harvard Business Review.
“We know that each patient is different—not only in terms of their medical history but also their personality, motivations, and values—and those differences can be amplified when it comes to making decisions about their health," write authors X. Shirley Chen and Mitesh S. Patel.
Experiment highlights why healthcare customization matters
The authors point to a study that focused on overweight and obese adults who participated in a 6-month program that tested gamification strategies. They used a remote-monitoring platform to track participants' daily step counts, randomly assigning people to 1 of 4 groups with a daily step goal.
While the control group received feedback but no interventions, the other 3 experienced a “gamification" approach that used behavioral nudges and social incentives. All 3 groups that used this approach performed better than the control group. But the only version that led to sustained changes during the 3-month follow-up was the group that used competition as a driver.
Including personality traits changed results
Those findings could have led to a universal approach, with the competition game driving future models. Instead, though, researchers created a follow-up study in which they surveyed people on their personality traits, risk preferences, and social support. This time, the findings were even more striking: different approaches clearly motivated people differently, depending upon factors such as how extraverted or social they were.
“Just as other industries tailor their digital offerings, healthcare providers could improve digital health efforts by delivering experiences that are more personalized and precisely tailored for each patient," Chen and Patel wrote.
What if you don't want to know?
While digital phenotyping efforts offer much promise, there are also patient privacy implications with big data. In an article looking at the “data shadow" of digital phenotyping, authors explore what this means for patients with Alzheimer's disease, for example. The article, published in Big Data & Society, describes the digital phenotyping practices in the dementia field as “alternatively empowering, enabling, and threatening."
Data from sources ranging from search engines to home devices could analyze someone's “unique signature" to detect early signs of neurodegenerative diseases.
The question, “Would you want to know?" looms large in this and similar policy discussions surrounding early detection of disease, authors note. This is especially relevant since these early data clues might not necessarily represent one's current health state.
Consistency in digital phenotyping presents a challenge
There are logistical challenges to consider as well, such as establishing standards in reporting consistency.
That was the most pressing recommendation from researchers who undertook a systematic review of digital tools used to passively monitor depression, findings published in npj Digital Medicine. Differences in reporting standards, methodologies, and study design can create difficulty gleaning insights that can be confidently reproduced and generalized. Among the other recommendations: increasing the diversity of study populations by recruiting people from different ethnic backgrounds and ages.
Procced cautiously with patient protections in mind
Protecting participants should be at the forefront of digital phenotyping efforts, write the authors of an article published in General Hospital Psychiatry. Ethical issues regarding wearables and sensors are still unsettled, according to the article, which highlights the following concerns:
- The current tools in use typically have a number of security or privacy concerns and “are invasive by nature." Similarly, there is a lack of adherence to security and privacy regulations among some apps.
- There are questions about the analytical methods and approaches, such as whether the sample size should be larger-scale epidemiological scale or smaller—and potentially highly vulnerable—patient populations.
- There are questions about how such technologies are integrated into the healthcare system.
Compiling an ethics checklist for digital phenotyping
Meanwhile, other research is hoping to narrow some of these gaps by establishing an ethics checklist. In an “viewpoint" article published in the Journal of Medical Internet Research, authors offered an ethics checklist to promote careful design and research.
Existing ethical guidance and legal regulations aren't sufficient for deep phenotyping research in the field of psychiatry, the researchers wrote. They created a checklist with 20 questions related to topics like informed consent; equity, diversity and access; privacy; regulation and law; return of results; and duty to warn and report.
RTI can help
RTI can help you balance the potential benefits of digital phenotyping efforts with privacy and other ethical challenges. From AI to machine learning and wearable devices, our diverse team of experts can help you explore a range of digital tools that can transform and enhance how healthcare is delivered.
Stay up-to-date on our latest thinking. Subscribe to receive blog updates via email.