Back to the library

Key question New

AI-processed · Author-approved

How do patient privacy concerns vary across different types of health data in digital mental health apps?

A Within-Person Privacy Calculus Perspective for Patient-Generated Health Data in Mental Health Care among Emerging Adults
Philipp Reindl-Spanner, Barbara Prommegger · International Conference on Information Systems (ICIS) · 2026

19 views

Illustration for: How do patient privacy concerns vary across different types of health data in digital mental health apps?

Loads Spotify's player; Spotify may set cookies. This device remembers your choice.

Key findings

  • Privacy concern varied significantly within individuals across the 21 health data types, with suicidal thoughts, communication tracking, and symptom tracking showing the highest privacy concerns, whereas external and step data showed the lowest.
  • Beyond general privacy dispositions (CFIP: b = 0.200, p < .001), perceived treatment relevance was positively associated with privacy concern within individuals (b = 0.248, p < .001), showing that data seen as more clinically useful are also perceived as more sensitive.
  • General performance expectancy had a negative main effect on privacy concerns (b = -0.123, p = .016) but positively moderated the relevance-concern relationship (b = 0.046, p = .016), indicating stronger relevance-driven privacy differentiation among users expecting higher app benefits.
  • Robustness checks confirmed that prior depression diagnoses, current depressive symptoms, and prior app use did not significantly alter or moderate these privacy assessments.

What it means for you

CIO / IT Executive

Stop using blanket consent modals and direct the engineering team to implement granular, data-type-specific permission toggles for mental health apps, separating high-sensitivity inputs like suicidal thoughts and communication tracking from step data.

IT Manager

Audit your mobile app's onboarding data collection flows to ensure that users are explicitly prompted to authorize or restrict access to high-risk categories like symptom tracking and communication logs individually.

Business Strategist

Revise the app's value proposition messaging; because users who expect high app utility actually exhibit greater privacy sensitivity for clinically relevant data, update in-app disclosures to explicitly link how specific sensitive data inputs directly drive clinical outcomes.

Researcher

Design follow-up experimental studies that investigate why perceived treatment relevance positively correlates with heightened privacy concerns in digital mental health, controlling for the specific clinical context.

Policymaker

Draft regulatory compliance guidelines that prohibit blanket consent mechanisms in digital health platforms, legally requiring apps to obtain granular, tiered consent for sensitive mental health data types such as suicidal ideation.

Summary

This study examines whether privacy concerns differ across specific types of patient-generated health data and how treatment usefulness relates to those concerns. The authors conducted an online survey with 439 emerging adults aged 18 to 29 who evaluated 21 data types using a within-person design. Data were analyzed using linear mixed-effects models to separate general privacy tendencies from data-type-specific assessments.

Problem

Digital health applications frequently rely on blanket consent models that treat all patient data as uniformly sensitive. In mental health contexts, individuals may view certain data categories as far more revealing than others, creating uncertainty about how perceived medical relevance shapes data-specific privacy risks.

Transcript
Host: Welcome to A.I.S. Insights — powered by Living Knowledge. I am Anna Ivy Summers. Today we examine an upcoming study from the International Conference on Information Systems 2026, titled "A Within-Person Privacy Calculus Perspective for Patient-Generated Health Data in Mental Health Care among Emerging Adults," authored by Philipp Reindl-Spanner and Barbara Prommegger. Joining me is our analyst, Alex Ian Sutherland. Alex, what is this study investigating?

Expert: Hello, Anna. This study explores how individuals weigh privacy risks against medical usefulness when generating health data on smartphones and wearables for mental health care. Specifically, it examines whether privacy concerns vary within the same individual across different types of data, rather than reflecting only a single, generalized stance toward disclosure.

Host: Let us look at the underlying challenge. Many digital health applications rely on blanket consent models that ask users to approve data collection all at once. Why is that problematic in mental health care?

Expert: Patient-generated health data encompasses a wide spectrum, from physical measurements like step counts and pulse rates to behavioral indicators like communication patterns, mood logs, and suicidal thoughts. When platforms treat all patient data as uniformly sensitive, they fail to account for how individuals evaluate different categories. In mental health contexts, certain data streams are perceived as far more revealing than others, creating uncertainty about how perceived medical relevance shapes data-specific privacy risks.

Host: How did the researchers investigate this dynamic?

Expert: They conducted an online survey with 439 emerging adults aged 18 to 29. Each participant evaluated 21 distinct types of patient-generated health data, rating both the perceived relevance of each type for mental health treatment and their privacy concern regarding that data. The authors analyzed the data using linear mixed-effects models, which allowed them to isolate within-person evaluations across data categories from general, between-person privacy dispositions.

Host: What did the analysis reveal about these assessments?

Expert: First, privacy concern varied significantly across the 21 health data types. Categories such as suicidal thoughts, communication tracking, and symptom tracking generated the highest privacy concerns, whereas external contextual data and step counts showed the lowest. Second, beyond general privacy dispositions, perceived treatment relevance was positively associated with privacy concern within individuals. When a participant rated a data type as more useful for treatment, their privacy concern regarding that specific data type also increased.

Host: Under traditional privacy calculus models, higher benefits typically offset privacy risks. Why did higher treatment relevance correspond to stronger privacy concern here?

Expert: In mental health care, treatment relevance functions not only as a utility signal but also as a sensitivity signal. The data types that offer the deepest clinical insights into a patient's condition—such as mood shifts or social interactions—are precisely the ones that expose intimate aspects of their personal life. Consequently, greater clinical value directly increases perceived inference risk.

Host: The study also evaluated performance expectancy—the overall expectation that digital health tools improve care. How did that influence the outcome?

Expert: Performance expectancy had a negative main effect on overall privacy concerns, meaning that users with higher general expectations reported lower baseline concern. However, it also positively moderated the relationship between relevance and privacy concern. Participants with higher performance expectancy differentiated more sharply between high-relevance and low-relevance data. In addition, robustness checks confirmed that prior depression diagnoses, current depressive symptoms, and past app use did not significantly alter these evaluations.

Host: What are the key practical implications for digital health developers and healthcare providers?

Expert: For technology developers, the findings demonstrate that applications should avoid blanket opt-ins and instead implement granular, data-type-specific consent options. Users should be able to share physiological data like sleep without being forced to share communication or thought logs. For healthcare providers, explaining the clinical relevance of data is necessary but insufficient; clinicians must also address privacy protections directly, recognizing that patients view the most useful data as the most sensitive. Finally, platform operators need tiered data governance policies that assign stricter access controls and retention limits to high-sensitivity categories.

Host: In summary, managing patient-generated health data requires differentiated transparency rather than uniform consent policies. Thank you, Alex, for sharing these insights, and thank you to our listeners for tuning in to A.I.S. Insights — powered by Living Knowledge.
Artificial Intelligence & Automation Information Systems & Management Human-Computer Interaction Data Analytics & Big Data Digital Transformation & Strategy Cybersecurity & Privacy Cloud & Infrastructure E-Commerce & Digital Platforms Operations & Supply Chain Sustainability & Green IS

Patient-generated health data, Privacy calculus, Privacy concerns, Mental health care, Perceived treatment relevance, Performance expectancy