An interventional study of Device: AI Clinical Decision Support System and Procedure: Standard of Care Anesthesia in Intraoperative Hypotension and Hemodynamic Instability, sponsored by Beijing Tsinghua Chang Gung Hospital. Not yet recruiting at 2 sites in China. Open to participants aged 18 Years to 85 Years. Per ClinicalTrials.gov, last updated 2026-04-06.
Sponsored by Beijing Tsinghua Chang Gung Hospital · Not applicable, Interventional, and Treatment
This proof-of-concept randomized controlled trial evaluates a reinforcement learning (RL)-based clinical decision support system for intraoperative hemodynamic management during non-cardiac surgery.
Background: Intraoperative hypotension is common during general anesthesia and is associated with adverse outcomes including acute kidney injury, myocardial injury, and increased mortality. Current hemodynamic management relies on the individual anesthesiologist's clinical judgment, which can vary in consistency and timeliness. An RL-based system that learns optimal vasoactive agent dosing strategies from clinical data may help standardize and improve real-time hemodynamic decision-making.
Purpose: The primary objective is to evaluate whether the RL-based decision support system can learn intraoperative hemodynamic management decisions comparable to those of experienced anesthesiologists, as measured by the mean absolute error (MAE) between RL-recommended and clinician-executed vasoactive agent doses. The secondary objective is to assess whether RL-guided management improves clinical hemodynamic outcomes, including the time-weighted average of hypotension and the percentage of time with mean arterial pressure within the target range.
Participants: Adult patients (aged 18 to 85 years, ASA I-IV) scheduled for elective non-cardiac surgery under general anesthesia with continuous invasive arterial blood pressure monitoring.
Procedures: Participants will be randomly assigned (1:1) to one of two groups. In the RL-guided group, the anesthesiologist will receive real-time vasoactive agent dosing recommendations from the decision support system displayed on a bedside screen; the anesthesiologist retains full clinical autonomy over all final decisions. In the standard care group, the anesthesiologist will manage hemodynamics according to institutional standard practice without input from the system. The patient and the outcomes assessor will be masked to group assignment. Data collection covers the intraoperative period and 30-day postoperative follow-up.
Rationale: Intraoperative hypotension, commonly defined as mean arterial pressure (MAP) below 65 mmHg, occurs in up to 60% of patients undergoing general anesthesia and is independently associated with acute kidney injury, myocardial injury, stroke, and 30-day mortality. Current approaches to hemodynamic management are reactive and rely on individual clinician judgment. Machine learning-based prediction systems (e.g., the Hypotension Prediction Index) have shown potential in reducing hypotension burden, but do not provide actionable dosing recommendations. Reinforcement learning (RL) offers a fundamentally different approach: learning optimal sequential decision-making policies from clinical data. The RL-PRAIS system (Reinforcement Learning-based Perioperative Real-time Anesthesia Intelligent System) consists of a Transformer-based patient state encoder and a model-based RL framework (patient model plus policy model) that generates real-time vasoactive agent dosing recommendations.
Hypothesis: The RL-PRAIS system can learn intraoperative hemodynamic management decisions that approximate those of experienced anesthesiologists and, when deployed as a decision support tool, can reduce the burden of intraoperative hypotension compared to standard care.
Study Design: This is a prospective, dual-center, parallel-group, randomized, controlled, proof-of-concept superiority trial. The study follows a four-phase stepwise validation framework: (1) model development using retrospective electronic health records (n = 7,216), (2) retrospective validation (n = 75), (3) prospective deployment study (n = 40), and (4) proof-of-concept randomized controlled trial (n = 40, 20 per arm). This registration covers Phase 4 (the RCT).
Objectives: Primary Objective: To evaluate the concordance between RL-recommended vasoactive agent dosing and attending anesthesiologist-executed dosing, quantified by mean absolute error (MAE) in norepinephrine-equivalent units. Secondary Objectives: To compare intraoperative hemodynamic outcomes between the RL-guided and standard care groups, including time-weighted average of MAP below 65 mmHg (TWA-MAP \< 65), percentage of time with MAP in the target range of 65-100 mmHg (MAP TIR), incidence of hypotensive events, cumulative vasoactive agent consumption, MAP variability, and clinician acceptance rate of RL recommendations. Exploratory Objectives: To collect pilot data on 30-day major adverse cardiac or cerebrovascular events (MACCE), acute kidney injury, perioperative myocardial injury, PACU length of stay, hospital length of stay, and 30-day all-cause mortality.
Interventions: Intervention Group (RL-Guided): The RL-PRAIS system provides real-time dosing recommendations for vasoactive agents (norepinephrine, phenylephrine, and ephedrine) based on continuous invasive arterial blood pressure monitoring and patient state features. Recommendations are displayed on a bedside screen at 1-minute intervals. The attending anesthesiologist retains full clinical autonomy and is responsible for all final dosing decisions. Control Group (Standard Care): The anesthesiologist manages intraoperative hemodynamics according to institutional standard practice and clinical judgment, with continuous invasive arterial blood pressure monitoring but without input from the RL-PRAIS system.
Study Population: Adult patients aged 18 to 85 years, ASA physical status I-IV, scheduled for elective non-cardiac surgery under general anesthesia with continuous invasive arterial blood pressure monitoring. Surgery is expected to last at least 2 hours. Key exclusion criteria include emergency surgery, severe obesity (BMI >= 35), and allergy to study-related agents.
Statistical Considerations: The sample size of 40 participants (20 per arm) is designed for a proof-of-concept evaluation, consistent with the benchmarking study (Wang et al., Nature Medicine 2023, RL-DITR, n = 16 in feasibility RCT). The primary analysis will compare MAE between groups using a two-sided paired or independent t-test (or non-parametric equivalent). Secondary hemodynamic outcomes will be compared using the Wilcoxon rank-sum test. Exploratory outcomes will be reported descriptively without formal hypothesis testing.
Beijing Tsinghua Chang Gung Hospital is the lead sponsor of 96 studies on the registry; 52 are open to participants now.
Counted across the registry records on this site, refreshed daily.
Exclusion Criteria:
Participants in this arm will have their intraoperative hemodynamic management guided by the RL-PRAIS clinical decision support system. The system provides real-time dosing recommendations for vasoactive agents (norepinephrine, phenylephrine, and ephedrine) based on continuous invasive arterial blood pressure monitoring and patient state features. Recommendations are displayed on a bedside screen at 1-minute decision epochs. The attending anesthesiologist retains full clinical autonomy and is responsible for all final dosing decisions. Anesthetic management (propofol, remifentanil, sevoflurane/desflurane) follows institutional standard protocols in both arms.
Device: Device: AI Clinical Decision Support System
Participants in this arm will receive standard-of-care general anesthesia management. The anesthesiologist will make all clinical decisions based on their professional judgment and standard institutional practices, without the aid of the investigational AI system.
Procedure: Procedure: Standard of Care Anesthesia
An AI-based software system that provides real-time, on-screen recommendations to the anesthesiologist regarding drug administration and adjustments to maintain optimal patient physiological parameter
Standard anesthesia management according to institutional guidelines. The attending anesthesiologist makes all clinical decisions based on their expertise and judgment without input from the investigational AI system.
Mean Absolute Error (MAE) of RL-Recommended Vasoactive Agent Dosing Versus Anesthesiologist-Executed Dosing
The primary outcome is the mean absolute error (MAE) between vasoactive agent doses recommended by the RL-PRAIS system and doses actually administered by the attending anesthesiologist. Vasoactive agents (norepinephrine, phenylephrine, ephedrine) are converted to norepinephrine-equivalent units (NEq, mcg/kg/min). MAE is computed at each 1-minute decision epoch during hemodynamic optimization and averaged across all epochs per patient. Concordance rate is also reported as the proportion of epochs where the RL-recommended dose falls within a clinically acceptable range of the executed dose.
Time frame: From the onset of hemodynamic optimization (15 minutes after surgical incision) until surgical wound closure
Time-Weighted Average of Mean Arterial Pressure Below 65 mmHg (TWA-MAP < 65 mmHg)
The time-weighted average of hypotension during the hemodynamic optimization period, calculated as the area under the MAP threshold of 65 mmHg (depth in mmHg multiplied by duration in minutes) divided by the total hemodynamic optimization time (minutes). This composite metric integrates both the severity and duration of intraoperative hypotension. Continuous invasive arterial blood pressure data sampled at 5-second intervals are used for calculation via the trapezoidal rule. A lower value indicates better hemodynamic management.
Time frame: From the onset of hemodynamic optimization (15 minutes after surgical incision) until surgical wound closure
Percentage of Time with MAP in Target Range (MAP TIR 65-100 mmHg)
The proportion of total hemodynamic optimization time during which mean arterial pressure remains within the target range of 65 to 100 mmHg, expressed as a percentage. A higher percentage indicates more stable hemodynamic control.
Time frame: From the onset of hemodynamic optimization (15 minutes after surgical incision) until surgical wound closure
Incidence of Intraoperative Hypotensive Events
The number of discrete hypotensive events per patient, defined as episodes where MAP falls below 65 mmHg for at least 1 minute continuously. An event is considered terminated when MAP returns to 65 mmHg or above for at least 1 minute.
Time frame: From the onset of hemodynamic optimization (15 minutes after surgical incision) until surgical wound closure
TWA-MAP Below 60 mmHg and Below 55 mmHg
Time-weighted average of MAP below 60 mmHg and below 55 mmHg, calculated using the same method as TWA-MAP \< 65 mmHg but with lower thresholds. These metrics capture moderate and severe intraoperative hypotension, respectively.
Time frame: From the onset of hemodynamic optimization (15 minutes after surgical incision) until surgical wound closure
Cumulative Vasoactive Agent Consumption
Total cumulative dose of all vasoactive agents administered during the hemodynamic optimization period, converted to norepinephrine-equivalent units (NEq, mcg/kg). This metric serves as a safety indicator to assess whether RL-guided management leads to excessive or reduced pharmacological intervention compared to standard care.
Time frame: From the onset of hemodynamic optimization (15 minutes after surgical incision) until surgical wound closure
Mean Arterial Pressure Variability
Coefficient of variation (CV) of mean arterial pressure during the hemodynamic optimization period, calculated as the standard deviation of MAP divided by the mean MAP, expressed as a percentage. A lower CV indicates more stable hemodynamic control.
Time frame: From the onset of hemodynamic optimization (15 minutes after surgical incision) until surgical wound closure
Clinician Acceptance Rate of RL Recommendations (Experimental Arm Only)
The proportion of RL-PRAIS dosing recommendations that are accepted and executed by the attending anesthesiologist without modification, expressed as a percentage of total recommendation epochs. Partial acceptance (dose modification in the recommended direction) is recorded separately.
Time frame: From the onset of hemodynamic optimization (15 minutes after surgical incision) until surgical wound closure
Major Adverse Cardiac or Cerebrovascular Events (MACCE) Within 30 Days
Composite endpoint including all-cause mortality, non-fatal myocardial infarction, non-fatal stroke, and new-onset heart failure within 30 days of surgery, assessed through medical record review and follow-up telephone contact.
Time frame: 30 days after surgery
Acute Kidney Injury (AKI) Within 7 Days
Incidence of acute kidney injury within 7 postoperative days, defined by KDIGO Stage 1 or higher: increase in serum creatinine by 0.3 mg/dL within 48 hours, or increase to 1.5 times baseline within 7 days, or urine output less than 0.5 mL/kg/h for 6 hours.
Time frame: 7 days after surgery
Perioperative Myocardial Injury (PMI)
Elevation of high-sensitivity cardiac troponin above the 99th percentile upper reference limit within the first 3 postoperative days.
Time frame: 3 days after surgery
Duration of Post-Anesthesia Care Unit (PACU) Stay
Time from PACU admission to PACU discharge, measured in minutes.
Time frame: Day of surgery
Postoperative Length of Hospital Stay
Number of days from the date of surgery to the date of hospital discharge.
Time frame: Up to 30 days after surgery
All-Cause 30-Day Mortality
Death from any cause within 30 days of surgery, ascertained through medical record review and follow-up contact.
Time frame: 30 days after surgery
Plan to share: No
No publications or documents are linked to this record.
This study is not yet recruiting, as verified in Apr 2026. You cannot join it, but the record below documents what was studied.
Get an email when the registry record changes — status, dates, results — or when someone posts here.
Sign in to followQuestions and observations about this study, from anyone following it. Not medical advice, and not a channel to the study team — their contact details are on the registry record.
Sign in to join the discussion. Reading takes no account; posting does. You choose a display name, and a pseudonym is the default.
Nothing here yet. If you are running this trial, taking part in it, or weighing whether to, this is the place to say so.
Beijing Tsinghua Chang Gung Hospital