Prospective Validation of an Artificial Intelligence Tool for Pre-Anesthetic Assessment

NCT07290647 | Not Yet Recruiting

• 1 other identifier: 85674025.4.0000.5530
• Study Type: observational
• Target: 270
• Locations: 1 country
• Sites: 1

Brief Summary

This prospective observational cohort study aims to validate an artificial intelligence (AI) tool designed for pre-anesthetic assessment in Portuguese, tailored to the Brazilian healthcare context. Conducted at a single tertiary hospital, the study will enroll 270 adult patients (aged \>18 years) scheduled for elective non-cardiac surgeries. Participants will use the AI tool to complete a self-assessment, generating general patient guidance and a detailed medical evaluation (the latter withheld from the anesthesiologist). A standard pre-anesthetic evaluation will then be performed by an anesthesiologist blinded to the AI results. A third blinded anesthesiologist will compare the assessments for accuracy, consistency, and risk identification (e.g., ASA classification and perioperative risk models). Primary outcome is concordance between AI and human assessments using Cohen's Kappa. Secondary outcomes include anesthesiologist perceptions of the tool's utility, impact on assessment quality, and patient usability challenges. The study poses minimal risks, with data collected over 24 months, and aims to enhance perioperative safety and efficiency in Brazil.

Conditions

Preoperative Care

Keywords

Artificial Intelligence; Large Language Model; Preoperative Evaluation; Perioperative Care; Risk Assessment; Observational Study

Outcome Measures

Primary Outcomes (1)

• Preoperative Risk Assessment

  The primary outcome is the level of concordance between the preoperative risk assessments performed by the artificial intelligence (AI) tool, based on a Large Language Model (LLM) in Portuguese, and those conducted by a human anesthesiologist. This concordance is evaluated in terms of the quality of information collected (e.g., completeness and relevance of patient data) and the precision of clinical judgment (e.g., identification of perioperative risks, American Society of Anesthesiologists (ASA) classification, and alignment with validated surgical risk models such as Ex-Care). Measurement Method: A third anesthesiologist, blinded to both the AI and human evaluations, will compare the assessments. The concordance will be quantified using the Cohen's Kappa coefficient for categorical variables (e.g., quality of information and clinical judgment).

  The assessments are conducted preoperatively for each participant, with data collected during their preoperative evaluation phase. The comparison is performed post-assessment, during the data analysis phase, which occurs after the 24-mo data collection.

Interventions

AI-based pre-anesthetic assessment OTHER

Patients use an artificial intelligence (AI) tool based on a Large Language Model (LLM) in Portuguese to complete a pre-anesthetic self-assessment. The tool collects patient information and generates two outputs: Generic Orientations: General instructions sent directly to the patient to aid in surgical preparation. Specific Assessment: A detailed evaluation, including recommendations and warning signs for medical use, which is not shared with the anesthesiologist performing the traditional evaluation.

Anesthesiologist-led pre-anesthetic evaluation OTHER

Each patient undergoes a standard pre-anesthetic evaluation conducted by an anesthesiologist, following routine clinical practice at both institutions. This evaluation is performed without access to the AI tool's results to ensure blinding. Purpose: To serve as the comparator for the AI-based assessment, allowing evaluation of concordance in risk assessment, quality of information collected, and clinical judgment. Details: The anesthesiologist conducts a clinical interview and review of medical records, assessing factors such as the American Society of Anesthesiologists (ASA) classification, perioperative risk models (for instance, Ex-Care model), and potential complications.

Eligibility Criteria

• Age: 18 Years+
• Sex: all
• Healthy Volunteers: No
• Age Groups: Adult (18-64), Older Adult (65+)
• Sampling Method: Non-Probability Sample

Study Population

The study population consists of adult patients (aged ≥ 18 years) scheduled for elective non-cardiac surgeries at two tertiary hospitals in Brazil. This population represents a typical cohort in a Brazilian public health system context, including individuals with varying comorbidities and surgical needs, such as those arising from the backlog of procedures in the Unified Health System (SUS) due to the COVID-19 pandemic. The focus is on patients requiring pre-anesthetic assessment for non-emergent, non-cardiac interventions, reflecting common surgical demands in digestive, genitourinary, circulatory, and upper airway/head/neck procedures. The estimated sample size is 270 participants to ensure adequate statistical power for evaluating the concordance between AI-based and human pre-anesthetic assessments.

You may qualify if:

• Patients aged 18 years or older
• Patients scheduled for elective non-cardiac surgeries

You may not qualify if:

• Patients undergoing diagnostic procedures with isolated sedation or local anesthesia

Sponsors & Collaborators

• Hospital Nossa Senhora da Conceicao: lead

Study Sites (1)

Hospital Nossa Senhora da Conceição (Grupo Hospitalar Conceição)

Porto Alegre, Rio Grande do Sul, 91787-400, Brazil

Location

Related Publications (3)

• Wongtangman K, Aasman B, Garg S, Witt AS, Harandi AA, Azimaraghi O, Mirhaji P, Soby S, Anand P, Himes CP, Smith RV, Santer P, Freda J, Eikermann M, Ramaswamy P. Development and validation of a machine learning ASA-score to identify candidates for comprehensive preoperative screening and risk stratification. J Clin Anesth. 2023 Aug;87:111103. doi: 10.1016/j.jclinane.2023.111103. Epub 2023 Mar 8.

  PMID: 36898279 RESULT

• Abdel Malek M, van Velzen M, Dahan A, Martini C, Sitsen E, Sarton E, Boon M. Generation of preoperative anaesthetic plans by ChatGPT-4.0: a mixed-method study. Br J Anaesth. 2025 May;134(5):1333-1340. doi: 10.1016/j.bja.2024.08.038. Epub 2024 Nov 14.

  PMID: 39547871 RESULT

• Yoon HK, Yang HL, Jung CW, Lee HC. Artificial intelligence in perioperative medicine: a narrative review. Korean J Anesthesiol. 2022 Jun;75(3):202-215. doi: 10.4097/kja.22157. Epub 2022 Mar 29.

  PMID: 35345305 RESULT

Central Study Contacts

Andre P. Schmidt, MD, PhD

CONTACT

+5551996412212; aschmidt@ghc.com.br

Study Design

• Study Type: observational
• Observational Model: COHORT
• Time Perspective: PROSPECTIVE
• Sponsor Type: OTHER
• Responsible Party: PRINCIPAL INVESTIGATOR
• PI Title: MD, PhD (Anesthesiologist - Department Chair).

Study Record Dates

• First Submitted: December 5, 2025
• First Posted: December 18, 2025
• Study Start: March 1, 2026
• Primary Completion (Estimated): June 30, 2027
• Study Completion (Estimated): June 30, 2028
• Last Updated: December 18, 2025

Record last verified: 2025-10

Data Sharing

• IPD Sharing: Will not share

Locations

BR (1)