Shadow Health-Related Data: Definition, Categorization, and User Perspectives

Yamane El-Zein, Kavous Salehzadeh Niksirat, Noé Zufferey, Mathias Humbert, Kévin HugueninEuroUSEC 2024 · DOI

TL;DR

This paper introduces the concept of "shadow health-related data" (shadow HRD) – health information generated or processed by individuals using general-purpose digital tools outside of professional healthcare systems. Through focus groups and a large-scale survey, the authors reveal the widespread creation and use of shadow HRD, users' limited awareness of associated privacy risks, and their minimal adoption of protective measures.

What Problem Does This Paper Address?

The proliferation of digital tools like fitness trackers, health apps, and AI chatbots has led to an explosion of health-related data (HRD). While HRD is recognized as sensitive and subject to strict privacy regulations (like GDPR and HIPAA), much of this data is generated and handled through everyday digital tools not specifically designed for healthcare. This "shadow HRD" – think health queries on search engines, medical appointment details in personal calendars, or even photos of symptoms – often flies under the radar of privacy regulations and user awareness. This oversight creates a significant blind spot in understanding and mitigating the privacy risks associated with our digital lives, potentially leading to discrimination or misuse of sensitive personal information. The paper aims to define, categorize, and understand user perspectives on this overlooked category of HRD.

Methods

The research employed a two-phase approach:

  1. Focus Groups: Two focus group sessions were conducted with a total of 23 participants (researchers and students). The goal was to identify and inventory a broad range of user behaviors that lead to the creation of shadow HRD. This qualitative data was then used to develop a classification of shadow HRD sources.
  2. Online Survey: A questionnaire was designed based on the focus group findings and deployed to 300 respondents in the US. This survey aimed to:
  3. Assess the prevalence of shadow HRD-creating behaviors.
  4. Gauge user awareness of and concerns about the privacy risks associated with shadow HRD.
  5. Understand the protective measures users employ (or don't employ).

Datasets

The study did not utilize pre-existing datasets for training or evaluation. Instead, it generated its own data through:

  • Focus Group Transcripts and Notes: Collected from 23 participants.
  • Online Survey Responses: Collected from 300 participants in the US.

Key Insights and Ideas

  • Ubiquity of Shadow HRD: The concept of shadow HRD is not a niche issue; it's deeply embedded in everyday digital activities.
  • Limited User Awareness: Many users are not fully aware that their common digital actions are generating sensitive health data.
  • Privacy Paradox: Users express concern about privacy from certain entities (hackers, advertisers) but show less concern about data access by intimate partners or healthcare providers, despite the potential for misuse.
  • Lack of Protective Measures: A significant gap exists between the creation of shadow HRD and the adoption of privacy-preserving behaviors.
  • Interdependent Privacy Risks: Shadow HRD creation extends beyond the individual, impacting family members and others.

Key Results

  • Prevalence of Behaviors: A high percentage of users engage in behaviors that create shadow HRD across various categories:
  • Productivity/Organization: 69% save healthcare provider contacts, 87% use calendar reminders for appointments.
  • Shopping/Finance: 78% purchase health-related items online, 90% use digital payments in health-related stores.
  • Communications: 73% communicate with healthcare providers via email, 67% via messaging/calling.
  • Information Seeking: 95% use general search engines for health info, 89% use health information websites.
  • User Awareness: Users show higher awareness of shadow HRD creation in categories like Productivity/Organization and Information Seeking compared to Shopping/Finance and Navigation/Transport.
  • Privacy Concerns: Users are most concerned about hackers (62.6% extremely concerned) and strangers on the internet (49.3% extremely concerned), but less so about intimate partners (59.2% not at all concerned) or healthcare providers (61.6% not at all concerned).
  • Low Adoption of Protections: For most categories, a majority of users reported using "none" of the proposed protection mechanisms. Even when data is synced to the cloud, users often don't employ specific privacy measures.
  • Interdependent Privacy: 26% of respondents reported engaging in behaviors that create shadow HRD about others at least "often."

Limitations and Future Work

  • Geographic Limitation: The survey was conducted in the US, limiting generalizability to other cultural and socioeconomic contexts. Future work aims to extend the survey to international respondents.
  • Focus on Technology Users: The study primarily focused on technology users and did not specifically investigate the shadow HRD-creating behaviors of healthcare practitioners. This is identified as a challenging but important area for future research.
  • In-depth Understanding: While the survey identified behaviors, in-depth interviews are planned to further investigate users' thought processes and the privacy-utility trade-offs they make.
  • Extensibility of Classification: The authors acknowledge that their classification of shadow HRD sources is not exhaustive and is intended to be extensible to incorporate new technologies.

Summary updated

BibTeX
@inproceedings{elzein2024shadow,
  author = {Yamane El-Zein and Kavous Salehzadeh Niksirat and Noé Zufferey and Mathias Humbert and Kévin Huguenin},
  title = {{Shadow Health-Related Data: Definition, Categorization, and User Perspectives}},
  booktitle = {EuroUSEC},
  year = {2024},
  doi = {10.1145/3688459.3688462},
}
Figure