aceModerate evidence
More
Review of 85 sources finds care robots with AI chat are rarely tested for long-term safety
Study title: The HEART Framework for LLM-Enabled Socially Assistive Robots in Healthcare: A PRISMA-Informed Structured Review
- Evidence type
- Systematic review or meta-analysis
- Sample
- n = 85
- Sources
- 1 sourcePeer-reviewed
- Published
- 2 min read
- Truvace record
- TRV-2026-2564cite or check it
Study title: The HEART Framework for LLM-Enabled Socially Assistive Robots in Healthcare: A PRISMA-Informed Structured Review
A PRISMA-informed review of research on socially assistive robots powered by large language models in healthcare found that the evidence base leans heavily toward early-stage questions. Database searches returned 128 records, 110 of them unique, and supplementary retrieval left 85 substantive sources for analysis.
Read the study abstract
Large language models (LLMs) are expanding the capabilities of socially assistive robots (SARs) through natural dialogue, personalisation, multimodal reasoning, retained interaction context, and adaptive behaviour in healthcare. Integrating generative language models into robots, however, complicates evaluation because fluent output may exaggerate perceived competence and increase the risks of hallucination, overtrust, privacy exposure, relationship dependency, and unsafe reliance on advice or actions. This PRIS…
The problem
A structured review of 85 sources found studies of LLM-enabled healthcare robots focused on feasibility, usability and short-term engagement, while longitudinal safety, governance of retained interaction context and sustained healthcare value received limited attention.
In brief
Those studies mostly measured feasibility, usability, acceptability, dialogue quality and short-term engagement. Longitudinal safety, comparative effectiveness, workflow integration and sustained healthcare value got little attention, according to the review.
The authors argue that fluent output from language models may exaggerate how competent a robot seems, which raises risks of hallucination, overtrust, privacy exposure, relationship dependency and unsafe reliance on advice or actions. They propose a framework called HEART and call for longitudinal and comparative validation.
Main points
- Database searches returned 128 records, of which 110 were unique after deduplication.
- Supplementary retrieval and assessment yielded 85 substantive sources covering background mapping, primary analysis and governance.
- Most studies examined feasibility, usability, acceptability, dialogue quality and short-term engagement.
- Longitudinal safety, governance of retained interaction context, comparative effectiveness and workflow integration received limited attention.
- The authors say evaluation must account for physical presence, social role, interaction memory and potential actions, not conversational performance alone.
- HEART has five domains: Human-Centred Communication, Ethical and Trustworthy Deployment, Adaptive and Embodied Intelligence, Relationship Continuity, and Translational Healthcare Value.
- The framework uses non-additive deployment gates so that strengths in one area cannot mask critical safety or governance failures.
The full story
The review covers healthcare robotics, human-robot interaction, LLM-enabled systems, ethics, implementation and care delivery. Language models are adding natural dialogue, personalisation, multimodal reasoning, retained interaction context and adaptive behaviour to socially assistive robots.
The authors say these gains complicate evaluation. Fluent output may make a robot seem more competent than it is, and a robot with physical presence, a social role and memory of past interactions carries risks beyond those of a text chatbot.
HEART uses boundary rules, operational indicators, qualitative labels and non-additive deployment gates to separate evaluative domains, define assessable outcomes and summarise reported support.
The authors recommend validating HEART through longitudinal and comparative assessment of hallucination severity, language-to-action safety, long-term effects, equity and post-deployment monitoring.
Sources
1 on file- Applied Sciencesdoi:10.3390/app16167904
Talk it over
Who carries this cost, and what would have to change to prevent it?
Review of 85 sources finds care robots with AI chat are rarely tested for long-term safetyhttps://truvace.com/item/the-heart-framework-for-llm-enabled-socially-assistive-robots-in-healthcare-a-pr
Who carries this cost, and what would have to change to prevent it?
The problem A structured review of 85 sources found studies of LLM-enabled healthcare robots focused on feasibility, usability and short-term engagement, while longitudinal safety, governance of retained interaction context and sustained healthcare value received limited attention.
Answer in the threadHow do we know? What can this kind of evidence show, and what can’t it?
The evidence This rests on a systematic review or meta-analysis with a sample of 85, from Applied Sciences.
Answer in the threadWhat does it leave open? What else would you want to know before acting on it?
The record’s own caveat The HEART framework is a proposal that has not yet been validated; the authors say future research should test it through longitudinal and comparative assessment.
Answer in the threadHave you seen AI used this way where you study, work or live? Did it go the same way?
Answer in the thread
The questions are the same on every record. The details under them are taken from this record as written.
What readers say