ECHO: A Locally-Deployable Agentic Health Assistant with Temporal Memory, Safety Guardrails, and Speech Assessment

2026-08-06

AI Summary

This work addresses the lack of localized, highly secure, and multimodal health assessment–capable intelligent assistants for chronic disease management by proposing a system deployable on consumer-grade devices. The system innovatively integrates temporal knowledge graphs with a hybrid safety mechanism combining rule-based constraints and graph neural networks, enabling a ReAct agent endowed with cross-session memory and clinical tool invocation capabilities. It further incorporates an acoustic–textual cross-attention module for voice-based emotion assessment. Experimental results demonstrate a 94.9% tool execution success rate; on a Turkish health dataset, the safety module achieves 88.8% accuracy and 90.6% hazard recall; and the emotion assessment module attains a macro F1-score of 0.652, collectively ensuring strong privacy preservation, safety, and multidimensional health awareness.

Abstract

This paper presents ECHO (Enhanced Care \& Health Observer), a locally-deployable conversational health assistant for long-term chronic care management. ECHO integrates three complementary software modules developed under shared supervision as a unified system. The core module is an agentic chatbot built on a ReAct loop orchestrated via LangGraph, equipped with 17 clinical tools and a temporal knowledge graph for persistent cross-session memory; it achieves a 94.9\% tool-execution pass rate across a 59-scenario benchmark with GPT-5 Mini. A two-stage hybrid safety layer intercepts all incoming queries: a rule-based layer handles explicit crisis signals and jailbreak attempts in under 1ms, while a signed graph neural network (GNN) with APPNP-style propagation classifies boundary cases by clinical intent, achieving 88.8\% accuracy and 90.6\% unsafe recall on a 2,537-query annotated Turkish health dataset while outperforming zero-shot LLM baselines including Llama 3.3 70B. A multimodal speech assessment module combining Whisper acoustic encoding and BERT text encoding with cross-attention fusion estimates emotion, depression, and pain, reaching a mean macro F1 of 0.652. The full system is implemented as a web application that can run entirely on consumer hardware, with no patient data transmitted to external services, supporting compliance with GDPR and KVKK.

Problem

• health assistant
• chronic care management
• temporal memory
• safety guardrails
• speech assessment

Innovation

• agentic health assistant
• temporal knowledge graph
• hybrid safety layer
• signed GNN
• multimodal speech assessment