Truace
Subscribe

General-purpose AI chatbots giving psychological support for depression and anxiety in type 2 diabetes and chronic urticaria scenarios

Source article: Psychological support for depression and anxiety across chronic illness contexts: a comparative evaluation of three general-purpose AI chatbots

Background Digital mental health tools, including general-purpose artificial intelligence (AI) chatbots, may provide accessible and low-threshold psychological support for people living with chronic conditions. However, their value in this population depends not only on recognizing depression or anxiety, but also on understanding how psychological distress is shaped by disease-specific experiences and when self-guided support should transition toward professional care. We evaluated the appropriateness, contextua…

By the Truvace desk1 min read1 sourcePeer-reviewedTRV-2026-1713· cite & verify
Psychological support for depression and anxiety across chronic illness contexts: a comparative evaluation of three general-purpose AI chatbots
Seattle - Sound Mental Health 02 by Joe Mabel. CC BY-SA 3.0 · http://creativecommons.org/licenses/by-sa/3.0/
Trace impact readingNegative state
P 74The P score combines the specificity and measured human impact of the grounded problem claim with the strength of this Trace’s cited sources.

Both sides are scored from claims and sources, not community votes.

G 69The G score combines the specificity and measured human impact of the grounded gain claim with the strength of this Trace’s cited sources.

In brief

Researchers tested ChatGPT, Gemini, and Doubao with 20 simulated-patient scenarios covering type 2 diabetes and chronic spontaneous urticaria. Each scenario was submitted twice to each system, producing 120 responses that blinded raters scored for appropriateness, contextualization, referral, and safety.

The results suggest chatbots may serve as an accessible first layer of emotional support, but they were weaker when distress was moderate or tied to a specific disease. Because the test used simulated patients, how this performs with real users remains unresolved, and the authors argue for integration with professional care rather than substitution.

Main points

  1. Appropriateness differed significantly across systems (P = 0.002): 90.0% for ChatGPT, 75.0% for Gemini, and 55.0% for Doubao.
  2. All responses to mild presentations were appropriate, versus 58.3% for moderate presentations.
  3. Contextualization was less consistent for urticaria, especially for flare unpredictability, pruritus, sleep disruption, and illness uncertainty.
  4. Test-retest consistency was moderate and varied between systems.

The gain

In simulated chronic-illness patients, chatbots recognized distress and gave emotional support consistently well, with 90.0% of ChatGPT responses meeting the appropriateness threshold.

The problem

Support became less reliable for moderate presentations and for disease-specific context, with major clinical or safety errors in 10.0% of Gemini and 15.0% of Doubao first-round responses.

The rundown

The study used matched mild and moderate depression and anxiety presentations alongside condition-specific distress, such as diabetes self-management and uncertainty over urticaria flares. This allowed comparison across two contrasting chronic illness contexts.

Differences were small for recognition and emotional support but larger for chronic-illness contextualization (P = 0.001), psychological support and referral (P = 0.002), and psychosomatic balance and safety (P<0.001).

The authors conclude that implementation should emphasize context-sensitive guidance, severity-responsive referral, and clear integration with professional care.

What this doesn’t fix

The evaluation used standardized simulated patients and only 20 scenarios, and test-retest consistency was only moderate, so results may not reflect real-world use by actual patients.

Sources

  1. Peer-reviewedFrontiers in Psychiatry2026-10-16

The debate