Camille Nebeker, EdD, MS, a professor at the UC San Diego Herbert Wertheim School of Public Health and Human Longevity Science and an affiliate of both the Design Lab and Qualcomm Institute, with decades of experience conducting empirical research on digital health research ethics, argues that there needs to be a rethink about informed consent in the digital health era.
When researchers ask people to join a health study, they often hand them a dense document written at a college reading level and call it “informed consent.”
At University of California San Diego, Nebeker leads a research team with Brian J. McInnis, Ph.D., Ramona Pindus, BSPH, Daniah Kareem, BSPH, Savannah Gamboa, BSPH, Maryam Gholami, Ph.D., Daniela G. Vital, MPH, and Eric B. Hekler, Ph.D. One line of research examines why traditional informed consent fails and what can be done instead.
Why informed consent matters and where it fails
Digital health studies often involve wearables, mobile apps and AI systems that collect continuous personal data across commercial and academic platforms.
Because informed consent is the bedrock of ethical research, before anyone volunteers to participate in a study, they should understand what they are signing up for. In practice, applying the principle of “respect for persons” sounds straightforward, yet it rarely is.
Consent forms are long, complex and rarely allow space for questions or real dialogue. Explaining in plain language, to a diverse public, what data will be collected, how it will be used, who might see it and what the risks are, is challenging.
What the research found
Across five peer-reviewed studies—from hands-on consent redesign to surveys, focus groups and design workshops—the team has built up a detailed picture of what makes consent work and what holds it back.
Plain language is necessary but not sufficient
The first study, published in Frontiers in Digital Health (2021), rewrote an actual Institutional Review Board (IRB)-approved consent form guided by the Digital Health Checklist for Researchers.
The original document was written at a 12th-grade reading level, well above the sixth-to-eighth grade standard that IRBs recommend. After an iterative process involving multiple researchers and a professional readability consultant, it was reduced to a ninth-grade reading level, and the word count dropped from 6,424 to 679.
That improvement mattered but revealed missing information and failed to address reasonable questions about privacy, third-party access and what happens to data after a study ends.
The consent form is not a form problem. It is a communication problem that requires listening to the people on the other side of the table.
What prospective participants want
The second study, published in the Journal of Empirical Research on Human Research Ethics (2024), went directly to prospective research participants. Nineteen people joined the focus group in which six major themes emerged.
People wanted to understand what they were agreeing to before volunteering to participate in research. They wanted to know who had access to their data and what happened if things went wrong. And they also wanted a chance to connect with the research team, not just a link to click. And they wanted options for how information was delivered and for how much control they retained over their data over time.
Participants were asking for a relationship, not a simpler form.
One size does not fit all
A 2025 survey study in the Journal of Medical Internet Research asked: Do people prefer simpler, shorter consent language? And does that preference hold across different groups?
Researchers presented 79 prospective participants with original IRB-approved language alongside simplified rewrites. Overall, there was slight preference for the original, detailed versions, but that pattern broke down along demographic lines in important ways.
- Age matters: Participants aged 40–49 were nearly twice as likely to prefer detailed original text compared to younger participants.
- Risk sections are different: When consent described study risks specifically, people were twice as likely to prefer the simplified version, suggesting risk communications need special attention.
- Questions not in the text: 55% of participants who provided feedback raised questions that neither version of the consent form answered.
- Hispanic/Latino participants: Participants identifying as Hispanic or Latino were three times more likely to say both versions needed improvement, a strong signal that language access is a gap.
Different people need different things.
Rethinking consent as a service, not a form
The Proceedings of the ACM on Human-Computer Interaction study (2024) applied “service design” thinking to the consent process. Researchers asked: What would the experience of consenting to research feel like if it were designed the way a good service is with the participant’s journey at the center?
Three priorities surfaced: people want the ability to withdraw their data at any time; they want visual tools showing data flow; and they want ongoing control over what they have agreed to share.
This suggests consent should be a dynamic relationship between researchers and participants.
Pictures worth a thousand words
The most recent study, published in Journal of the American Medical Informatics Association (2025), asked whether dataflow diagrams—visual maps showing how data moves through a research system—helped participants understand the complex system of digital health research.
Across eight focus groups with 34 participants, they found that these diagrams helped by supplementing text-based consent. Seeing where their data would go prompted conversations about risks that plain text had not surfaced.
Participants were clear that diagrams must be simple and participant-centered. They also drew a firm line around artificial intelligence (AI), rejecting the idea of AI evaluating research risks on their behalf, though most were comfortable with AI being used to help format and simplify a diagram that humans had created and vetted.
Why this matters beyond the research lab
The implications of this work go beyond academic research studies. It applies in clinical settings, public health programs or commercial digital health products. As AI systems take on more roles in health care, and as wearables become ubiquitous, the gap between perceived and actual consent will widen.
In 2023, the National Institutes of Health issued a request asking for ideas on how to improve informed consent for digital health research. This body of work offers some concrete answers.
At the heart of all five studies is a deceptively simple premise. The goal of informed consent is not a signed form. It is an informed person. Getting there requires treating consent not as a compliance exercise, but as a genuine act of respect for the people who volunteer to advance science.
Key medical concepts
Clinical categories
The content is provided for information purposes only.