Effectiveness of Large Language Model for Anaesthesia and Procedural Consent (PEAR)

Evaluating the Effectiveness of Large Language Models in Anaesthesia and Procedural Consent: A Comparative Analysis With Traditional Patient Consent Methods

Patient understanding of anaesthesia risks remains inconsistent due to time constraints, language barriers, and variable clinician communication styles. Traditional verbal consent may not consistently ensure comprehension or reduce preoperative anxiety. PEAR (Patient Education of Anesthesia Risks) is a multilingual, AI-driven chatbot developed to enhance patient education and improve the quality of anaesthesia risk counselling.

Study Objective:

To compare PEAR's performance in delivering anaesthesia risk consent against the standard face-to-face verbal method.

Study Overview

• Status: Recruiting
• Conditions: Anesthesia; Consent Forms; Artificial Intelligence (AI)
• Intervention / Treatment: Other: PEAR

Detailed Description

This study evaluates the effectiveness of PEAR (Patient Education of Anaesthesia Risks), a conversational AI-based chatbot designed to deliver anaesthesia risk education to patients in a personalized, interactive, and multilingual format. The goal is to support informed consent by improving patient comprehension, satisfaction, and reducing anxiety, while also streamlining clinician workflow.

Participants undergoing elective surgery will be randomly assigned to either receive anaesthesia counselling via PEAR before their consultation with the anaesthetist (intervention group) or undergo the standard face-to-face verbal consent process (control group). The PEAR chatbot is accessed through a secure digital interface and presents information aligned with institutional anaesthesia protocols.

The study will be conducted at hospitals within the SingHealth cluster in Singapore. Following the consent process, patients will complete a short quiz to assess understanding, a survey to evaluate satisfaction, and an anxiety scale. Clinicians will record time taken and perceived workload.

All patients will still meet their anaesthetist, ensuring clinical oversight is maintained. This study does not alter standard care but evaluates a digital adjunct to enhance it. Data will be collected electronically, anonymised, and stored securely. Insights from this trial may inform the wider implementation of digital tools in perioperative patient education.

• Study Type: Interventional
• Enrollment (Estimated): 120
• Phase: Not Applicable

Contacts and Locations

• Study Contact: Name: Yuhe Ke, MMED (ANES); Phone Number: +6581022852; Email: yuhe.ke36@gmail.com

Study Locations

• Singapore
  • Singapore
    • Singapore, Singapore, Singapore, 249094
      • Recruiting
      • Singapore General Hospital
      • Contact: Yuhe Ke; Phone Number: 81022852; Email: yuhe.ke36@gmail.com
    • Singapore, Singapore, Singapore, 751126
      • Not yet recruiting
      • Singapore General Hospital
      • Contact: Ke Yuhe; Phone Number: 81022852; Email: guardian.angel898@gmail.com

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Adult; Older Adult
• Accepts Healthy Volunteers: Yes

Description

Inclusion Criteria:

- Adults (≥21 years old) undergoing elective surgery requiring anaesthesia

Classified as ASA Physical Status I to III

• Able to provide informed consent
• Able to communicate effectively in English, Chinese (Mandarin), Malay, or Tamil
• Willing and able to complete questionnaires and interact with the PEAR chatbot (intervention arm)

Exclusion Criteria:

• ASA Physical Status IV or above
• Cognitive impairment or psychiatric conditions that may limit comprehension or communication
• Non-literate patients or those unable to understand English, Chinese, Malay, or Tamil
• Emergency surgery cases
• Prior participation in the study (to prevent bias)

Study Plan

How is the study designed?

Design Details

• Primary Purpose: Other
• Allocation: Randomized
• Interventional Model: Parallel Assignment
• Masking: Single

Number of Arms

2

Arms and Interventions

Participant Group / Arm	Intervention / Treatment
No Intervention: Control; Participants in the control arm will receive the standard anaesthesia risk counselling conducted face-to-face by a licensed anaesthetist. This process follows institutional protocols and includes a verbal explanation of anaesthesia procedures, associated risks, benefits, and potential complications, tailored to the patient's specific surgical context. Patients are encouraged to ask questions and engage in discussion during the session. No digital tools or chatbot assistance will be used in this arm.
Experimental: PEAR; Participants in this arm will receive anaesthesia risk counselling through the PEAR (Patient Education of Anaesthesia Risks) chatbot, a multilingual, AI-powered conversational tool. The chatbot provides personalized education on anaesthesia risks, procedures, and post-operative expectations. Patients interact with the chatbot prior to meeting the anaesthetist, enhancing their understanding and preparing them for the face-to-face consultation.	Other: PEAR; Participants in the intervention arm will receive anaesthesia risk counselling through the PEAR (Patient Education of Anaesthesia Risks) chatbot prior to their face-to-face consultation with an anaesthetist. PEAR is a multilingual, AI-powered conversational tool designed to provide personalized, interactive education on anaesthesia-related procedures, risks, and safety information. The chatbot delivers content aligned with institutional guidelines and allows patients to explore topics at their own pace, ask questions in natural language, and revisit information as needed. After completing the chatbot interaction, patients proceed with their standard preoperative consultation, where any further questions are addressed by the anaesthetist. This approach is designed to enhance patient understanding, reduce anxiety, and optimize the in-person consultation by preparing patients in advance.

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Patient self-reported understanding of anaesthesia risks	The primary outcome was assessed using a Patient-Reported Experience Measure (PREM) focused on subjective comprehension of anesthesia. This was measured via three validated 5-point Likert scale items (1 = strongly disagree, 5 = strongly agree) evaluating: (1) clarity of risks and procedures, (2) confidence in the anesthesia plan, and (3) self-reported ability to recall and explain key risks. While both groups completed these items following the clinician consultation, the intervention group underwent additional longitudinal assessments-at baseline and post-chatbot interaction-to facilitate a within-group analysis of the chatbot's independent educational impact.	Immediately post-interaction with the PEAR Chatbot

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Perceived Usefulness	Description: Evaluation of the user's subjective belief that using the PEAR chatbot enhanced their clinical experience. Measured using the mean score of three items: "The chatbot improved my understanding," "It was a useful addition to my consultation," and "It helped me make informed decisions." Unit of Measure: 5-point Likert scale (1 = strongly disagree, 5 = strongly agree). Time Frame: Immediately pre-consent and immediately post-consent (within the same clinic visit).	Immediately pre-consent and post-consent (within the same clinic visit)
Cost effectiveness	Participants were presented with a hypothetical choice scenario outlining explicit trade-offs between in-person clinic attendance (travel time 60-120 minutes, waiting time 60-180 minutes, travel cost $10-50 Singapore dollars, includes physical examination) versus chatbot use at home (zero travel time, zero waiting time, zero cost, no physical examination). Participants selected their preference, providing insight into patient values and priorities.	Immediately post-chatbot use (same clinic visit)
Perceived Ease of Use (PEOU)	Description: Evaluation of the degree to which the user believes that using the chatbot was free of effort. Measured using the mean score of four items: "Easy to use," "Easy to learn," "Comfortable navigating," and "Language was easy to understand." Unit of Measure: 5-point Likert scale (1 = strongly disagree, 5 = strongly agree).	Immediately pre-consent and immediately post-consent (within the same clinic visit).
Attitude Toward Using (ATT)	Description: Evaluation of the user's positive or negative feelings about performing the target behavior. Measured using the mean score of three items: "I enjoyed using the chatbot," "Using the chatbot was a good idea," and "I feel confident after using it." Unit of Measure: 5-point Likert scale (1 = strongly disagree, 5 = strongly agree).	Immediately pre-consent and immediately post-consent (within the same clinic visit).
Behavioral Intention to Use (BI)	Description: Evaluation of the user's likelihood to engage with the PEAR chatbot in future clinical scenarios. Measured using a 5-point Likert scale. Unit of Measure: 5-point Likert scale (1 = strongly disagree, 5 = strongly agree).	Immediately pre-consent and immediately post-consent (within the same clinic visit).

Collaborators and Investigators

• Sponsor: Singapore General Hospital

Publications and helpful links

General Publications

• Ke YH, Jin L, Elangovan K, Abdullah HR, Liu N, Sia ATH, Soh CR, Tung JYM, Ong JCL, Kuo CF, Wu SC, Kovacheva VP, Ting DSW. Retrieval augmented generation for 10 large language models and its generalizability in assessing medical fitness. NPJ Digit Med. 2025 Apr 5;8(1):187. doi: 10.1038/s41746-025-01519-z.

Study record dates

Study Major Dates

• Study Start (Actual): January 7, 2026
• Primary Completion (Actual): April 27, 2026
• Study Completion (Estimated): April 27, 2026

Study Registration Dates

• First Submitted: April 22, 2025
• First Submitted That Met QC Criteria: April 28, 2025
• First Posted (Actual): April 29, 2025

Study Record Updates

• Last Update Posted (Actual): May 7, 2026
• Last Update Submitted That Met QC Criteria: April 30, 2026
• Last Verified: April 1, 2026

More Information

Terms related to this study

• Keywords: Artificial Intelligence; Anaesthesia; Informed Consent; Large Language Model
• Other Study ID Numbers: 2025-04-22

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: YES

IPD Plan Description

De-identified individual participant data (IPD) that underlie the results reported in the publication (including primary and secondary outcomes, baseline characteristics, and questionnaire scores) will be made available to qualified researchers upon reasonable request.

Data will be shared beginning 6 months after publication and will be accessible for up to 3 years post-publication. Requests must include a methodologically sound proposal and be submitted to the Principal Investigator. A data access agreement will be required to ensure ethical use and protection of participant confidentiality.

• IPD Sharing Time Frame: 1st July 2025 - 1st july 2028
• IPD Sharing Access Criteria: Access to de-identified individual participant data (IPD) will be granted to qualified academic researchers, healthcare professionals, or institutions conducting methodologically sound research that aligns with ethical standards and scientific purpose.
• IPD Sharing Supporting Information Type: ANALYTIC_CODE

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No