Work

Coexistence of AI Nurse and AI-Simulated Patient in Nursing Education Using a Bidirectional Fine-Tuning Pipeline: Development and Content Analysis Study

Year: 2026
Type: article
Abstract: Background: Simulated patient systems are essential in nursing education but face actor scarcity, instructor burden, cost, and disrupted continuity where in-person simulation cannot be sustained. Existing large language model (LLM)-based systems address these issues but treat the dialogic partner-the AI nurse-as a static prompt or unspecialized model, compromising data collection quality and pedagogical applicability. Objective: We report (1) the construction of 2 specialized fine-tuned models, NURSE-ONE (AI nurse) and PATIENT-ONE (AI-simulated patient), via a 5-step bidirectional fine-tuning pipeline; (2) their evaluation using Berelson content analysis validated by independent human coders; (3) an exploratory third-party evaluation by nursing faculty; and (4) the design of a coexistence simulation ecosystem supporting 3 operating modes, including offline disaster preparedness applicability. Methods: On a single consumer-grade graphics processing unit (NVIDIA RTX 4090; 24 GB video random access memory), the multimodal base model gemma4-E4B-it was fine-tuned via low-rank adaptation in five steps: (1) a 251-scenario base from 29 institutionally authored nursing care plans and 20 items from the 115th Japanese National Nursing Examination; (2) a provisional dialogue system (AutoGen v18, maxmsg=25) to collect training dialogues; (3) NURSE-ONE fine-tuning across 3 training volume conditions (245, 735, and 1225 cases); (4) 4-condition Berelson comparison; and (5) PATIENT-ONE fine-tuning on NURSE-ONE-generated dialogues with a 3-condition comparison. Interrater reliability was confirmed by 2 independent nursing experts (196 blinded utterances, κ=0.770) and corroborated by 2 LLM raters (κ=0.742). Ten nursing faculty evaluated 3 coexistence modes in a working web user interface (21-item rubric; item-level content validity index=1.00). Results: =.30) showed no significant category shift but the lowest response length variability (SD 24.5) and highest information-providing rate (47.9%) at 735 cases, adopted as PATIENT-ONE. The 2 adapters appeared to operate on different axes-qualitative questioning style on the nurse side and quantitative response stability on the patient side-a preliminary observed pattern we term functional asymmetry, pending replication. Third-party evaluation rated the AI-simulated patient, AI nurse, and dyadic interaction at 90.3%, 69.0%, and 84.8%, respectively, with role maintenance at ceiling (mean 5, SD 0). All 3 modes met the prespecified 3-second latency threshold. Conclusions: A 5-step bidirectional fine-tuning pipeline produced 2 specialized models whose joint deployment constitutes a design pattern for nursing education simulation. The preliminarily observed functional asymmetry between qualitative nurse-side variation and quantitative patient-side stabilization appears to underwrite the system's coherence, supporting 3 operating modes on consumer-grade hardware and deployable offline in disaster preparedness contexts. Third-party evaluation confirmed exploratory educational applicability, with all evaluators rating the system as usable, providing a foundation for future learner outcome studies.
Language: English
Subfield: Education
Open Access status: gold
Estimated APC paid: $1,500
OpenAlex ID: W7212863361
FWCI:
Cites: 33
Cited by:
Related to: 9