Four-module system architecture for AI-SPs: LLM core, affective computing, personality modulation, and safety guardrails¶

Design · All designs
Evidence · no claims cited

Description¶

The article describes a system architecture comprising four core modules. The LLM core generates "natural language responses consistent with clinical scenarios and patient roles." The affective computing module identifies "emotional cues in learner input (eg, tone, keywords, and empathic expression)." The personality and communication stance modulation module integrates the Big Five personality model and Satir communication stance theory. The safety guardrail and content filtering module ensures responses "do not contain inappropriate content, do not provide medical advice beyond the SP role."

Design Implications¶

Context¶

Requirements¶

Medical dialogue–specific prompt engineering to improve medical accuracy and role consistency
Human-in-the-loop mechanism for sensitive scenarios involving patient safety or ethical boundaries

Constraints¶

The description remains at the module level rather than complete technical specifications

Target Learners¶

Medical students

Learning Goals¶

Simulated clinical communication practice

Claims¶

Examples¶

-

Key Sources¶

Jia Chuanlin; Qi Lulu. (2026). Digital Standardized Patients: Conceptual Framework for Generative AI–Empowered Medical Education. JMIR Medical Education. https://doi.org/10.2196/91050

Page metadata

Field	Value
`type`	`design`
`id`	`ai-sp-four-module-system-architecture`
`title`	`Four-module system architecture for AI-SPs: LLM core, affective computing, personality modulation, and safety guardrails`
`description`	`The article describes a system architecture comprising four core modules.`
`status`	`review`
`generated`	by `process:settle-candidates` on `2026-11-04`
`trust`	unverified — no one has confirmed this page's content
`sources`	1 — `jia-chuanlin-qi-lulu-2026`

Frontmatter is the page's machine-readable record — see the Schema & Guide.