This prospective observational study aims to evaluate the ability of artificial
intelligence (AI) models to interpret arterial waveform analysis data obtained from a
hemodynamic monitoring system. The study will focus on assessing the accuracy of
ChatGPT-4 and Gemini 2.0 in detecting hemodynamic abnormalities, providing diagnostic
suggestions, and offering treatment recommendations based on arterial waveform data
collected from elective surgical patients.
Background and Rationale Arterial waveform analysis is a critical component of advanced
hemodynamic monitoring, providing real-time insights into cardiac output, vascular
resistance, and volume status. These parameters are essential for guiding perioperative
fluid management and optimizing hemodynamic stability in surgical and critically ill
patients. While automated monitoring systems generate large amounts of data, the
interpretation of these waveforms remains dependent on clinician expertise. The
integration of AI-based decision-support tools in this context could enhance real-time
clinical decision-making and reduce workload for healthcare providers.
Study Objectives
The primary objective of this study is to determine the ability of AI models to analyze
arterial waveform data and detect clinically significant hemodynamic abnormalities. The
secondary objectives are:
To assess the concordance between AI-generated diagnoses and expert anesthesiologist
assessments.
To evaluate the clinical appropriateness of AI-generated treatment recommendations.
To explore the potential role of AI in clinical decision support systems for hemodynamic
monitoring.
Study Design and Methodology
This study will be conducted at two tertiary-level healthcare institutions:
Health Science University İstanbul Kanuni Sultan Süleyman Education and Training Hospital
Başakşehir Çam and Sakura City Hospital The study will include adult patients undergoing
elective surgery who require intraoperative arterial waveform monitoring as part of
routine perioperative care.
Data Collection Process Hemodynamic data will be collected from participants using the
MostCare hemodynamic monitoring system, which is routinely used in perioperative
settings.
Data collection will take place at three time points:
Pre-anesthesia (baseline hemodynamic status before induction) Post-anesthesia induction
(after intubation, before surgical incision) Intraoperative period (during key surgical
events requiring hemodynamic intervention) If an intervention needs according to arterial
wave analysis we will also take data before and after intervention.
AI-Based Analysis
The collected arterial waveform data will be anonymized and processed by AI models
(ChatGPT-4 and Gemini 2.0) to provide:
Abnormality detection - Identifying any deviations from normal hemodynamic parameters.
Diagnostic suggestions - Providing likely clinical diagnoses based on the waveform
patterns.
Treatment recommendations - Suggesting possible interventions to optimize hemodynamic
status.
Expert Validation AI-generated results will be independently reviewed by experienced
anesthesiologists to assess their accuracy and clinical relevance.
The concordance between AI outputs and expert assessments will be statistically analyzed.
Outcome Measures
Primary Outcome:
Accuracy of AI models in detecting hemodynamic abnormalities compared to expert
assessments.
Secondary Outcomes:
Concordance between AI-generated diagnoses and anesthesiologist diagnoses. Clinical
appropriateness of AI-generated treatment recommendations compared to standard clinical
practice.
AI models' potential role in enhancing clinical decision-making in perioperative
hemodynamic management.
Ethical Considerations The study does not involve any additional interventions beyond
routine clinical monitoring.
No patient-identifiable data will be used in AI model analysis. Informed consent will be
obtained from all participants before enrollment. The study has been approved by the
relevant ethics committees at both participating institutions.
Study Timeline Planned study duration: 6 months Estimated start date: February 15, 2025
Estimated completion date: August 15, 2025 Potential Impact
This study will provide valuable insights into the role of AI in automated hemodynamic
monitoring and perioperative decision support. If successful, AI-driven analysis of
arterial waveform data could:
Enhance patient safety through early detection of hemodynamic abnormalities. Improve
efficiency by assisting anesthesiologists in data interpretation. Reduce workload for
perioperative and critical care teams. Support future AI-based clinical decision-support
tools for hemodynamic monitoring.