LLM + Wearable Personal Health Systems (2024-2026)
positioning
ClaimLLM + wearable systems (2024-2026) reduce the risk that the AgentSee architecture is infeasible but do not address agency-as-objective, human-as-controller topology, or caring governance.
Status: NEAREST ENGINEERING NEIGHBORS for technical feasibility. Critical difference in objective, topology, and governance.
What exists
Since 2024, several systems combine language models with wearable sensor data for coaching, stress monitoring, or health prediction:
- Stress chatbots integrating physiological state recognition with LLM interaction (Dongre et al. 2024)
- Wearable-triggered stress interventions via LLM chatbot (Neupane et al. 2025)
- LLM systems for wearable-grounded health prediction or insight generation (Kim et al. 2024; Merrill et al. 2026)
[VERIFIED: All four citations independently verified against primary literature on 2026-03-03.]
What this changes
These systems reduce the risk that the AgentSee architecture is infeasible -- LLM + sensing integration is being explored and is engineering-tractable. They also narrow the novelty claim.
What remains novel
The novelty is not "LLM + wearables exist." It is the specific combination of:
- Agency capacity as terminal objective rather than stress reduction, adherence, or coaching quality
- Human-as-controller topology with machine-as-observer/stabilizer rather than machine-as-coach driving behavior change
- Caring invariants and anti-dependency tests as success conditions rather than engagement or retention