Evidotes: Integrating Scientific Evidence and Anecdotes to Support Uncertainties Triggered by Peer Health Posts

Paper Title

Evidotes: Integrating Scientific Evidence and Anecdotes to Support Uncertainties Triggered by Peer Health Posts

Publication Info

• Topic area: Health information systems for uncertainty management in online communities.
• Keywords: health information, uncertainty management, peer anecdotes, scientific evidence, Reddit, browser extension, information synthesis, emotional regulation, user lenses, mixed-methods study.

Background and Problem

• Problem / challenge: Existing health information systems focus on improving content quality and relevance but fail to address the emotional and informational uncertainties triggered by peer health anecdotes. Users lack tools to integrate scientific evidence with peer experiences effectively.
• Significance: Managing health uncertainties is critical for patients navigating illness, as it influences decision-making and emotional well-being. Current systems leave users overwhelmed or confused, hindering productive information-seeking behaviors.
• Motivation and related work: Prior research highlights the dual role of anecdotes as both valuable and anxiety-inducing. Existing systems prioritize credibility and filtering but neglect the emotional and cognitive needs of users. This paper addresses the gap by proposing a system that integrates scientific and anecdotal evidence to support uncertainty management.

Solution

• Proposed approach: Evidotes, a Chrome browser extension, augments Reddit health posts with synthesized scientific and anecdotal information presented through three user-selectable lenses: Dive Deeper, Focus on Positivity, and Big Picture.
• Novelty:
  1. Introduction of user-selectable lenses to tailor information retrieval and framing based on uncertainty management strategies.
  2. Integration of scientific evidence and peer anecdotes to create "information symbiosis" that enhances both accessibility and contextual relevance.
  3. Mixed-methods evaluation demonstrating improvements in information satisfaction and emotional well-being.
  4. Design principles for health information systems that prioritize user agency and explicit control over algorithmic inference.
• Procedure and key techniques:
  1. Evidotes adds a button to Reddit posts, allowing users to select one of three lenses.
  2. The system retrieves relevant scientific articles (via PubMed) and peer anecdotes (via Reddit) using a Retrieval-Augmented Generation (RAG) pipeline with Large Language Models (LLMs).
  3. Synthesized claims are presented in a panel with expandable sections for scientific and anecdotal sources.
  4. Claims are validated for citation faithfulness and topical relevance, with transparency mechanisms for unsupported claims.

Results

• Concrete findings:
  • Information satisfaction increased from 3.2 to 4.6 (+43%, p < .001).
  • Emotional stress reduced from 3.4 to 1.9 (-44%, p < .001).
  • 96.7% of citations were faithfully attributed; 98.2% of claims were topically relevant.
• Advantage over baselines:
  • Evidotes enabled bidirectional navigation between scientific and anecdotal evidence, resolving the specificity-generalizability tradeoff.
  • Users reported reduced cognitive demand and emotional overwhelm compared to baseline browsing experiences.
• Experiments / evaluation:
  • Mixed-methods study with 17 chronic illness patients.
  • Participants interacted with 77 Reddit posts, triggering 130 lens interactions.
  • Measures included pre- and post-study surveys on information satisfaction, stress, and cognitive demand, complemented by qualitative interviews.
• Limitations and future work:
  • Small sample size (n=17) limits generalizability.
  • Focused on Reddit; future work should test across other platforms.
  • Need for longitudinal studies to assess sustained impact on uncertainty management.
  • Potential risks of over-reliance on specific lenses or unverified claims.

Summary

Evidotes is a browser extension designed to address the uncertainties triggered by peer health posts by integrating scientific evidence and anecdotes through user-selectable lenses. A mixed-methods study with chronic illness patients demonstrated that Evidotes improved information satisfaction (+43%) and reduced emotional stress (-44%) compared to baseline browsing. The system enabled "information symbiosis," where scientific and anecdotal evidence complemented each other, enhancing accessibility and contextual relevance. Evidotes offers a novel approach to health information systems, emphasizing user agency and explicit control, with potential applications across diverse platforms and conditions.