Xuechao Hao

Global surgical volumes continue to rise, yet postoperative morbidity and mortality remain substantial, particularly among geriatric patients. The scarcity of multi-center prospective perioperative cohorts with active follow-up and large-scale, high-quality data limits the understanding of risk profiles and hinders individualized perioperative management in this population. To address this gap, the PeRiOperative sTress risk assEssment and Clinical decision cohorT (PROTECT) was established by creating a dedicated perioperative data platform for geriatric patients. This cohort profile specifically describes the study design, recruitment strategy, data structure, and current status of PROTECT. PROTECT is an ongoing, ambispective, real-world observational cohort across three tertiary medical centers. The study continuously enrolls inpatients aged ≥ 65 years undergoing surgery under anesthesia. Preoperatively, standardized pre-anesthesia assessments are conducted to collect comprehensive medical information. Intraoperatively, anesthetic and surgical data, along with high-frequency biosignals are recorded. Postoperative outcomes are evaluated at 48 hours, 7 days, and 30 days. Descriptive analyses were performed to summarize baseline characteristics and postoperative outcomes. The first participant was enrolled in August 2019. As of May 2025, 61,289 participants aged 65-100 years have been included. The median age is 71 years, and 44.4% are female. 52.2% are classified as American Society of Anesthesiologists physical status III or higher. The most common surgeries are abdominal (42.8%) and orthopedic (19.9%). Since October 2023, postoperative follow-up response rates have reached 96.5% at 48 hours, 77.7% at 7 days, and 72.1% at 30 days. The 30-day all-cause mortality rate is 0.7%. The most frequent postoperative adverse outcomes are pulmonary infection (13.0%), nausea (11.9%), hepatic dysfunction (9.9%), and intensive care unit admission (8.6%). Leveraging the PROTECT dataset, several prediction models for major postoperative complications have already been developed and implemented in clinical practice. PROTECT provides one of the largest and most comprehensive perioperative datasets for geriatric population. It enables detailed characterization of risk profiles, morbidity, and mortality in geriatric patients, supports identification of contributors to adverse outcomes, and offers a robust platform for developing perioperative risk assessment tools and multivariable models to support clinical decision-making. Future expansions will include extending follow-up to 90, 180, and 365 days and recruiting additional participating centers. Registered at chictr.org.cn on 15/10/2025 (ChiCTR2500110517).

BACKGROUND High-performing perioperative prediction models have not consistently translated into clinical benefit, in part because model outputs must be delivered through clinical decision support systems (CDSS) that align with anesthesia workflows and end-user needs. OBJECTIVE To identify anesthesia professionals’ requirements for perioperative CDSS and use these findings to inform the design specification of a user-centered perioperative CDSS. METHODS This user-centered study was conducted in four sequential phases: translation of a previously validated explainable machine-learning model into candidate CDSS functions; three rounds of focus group–based iterative prototyping; a nationwide cross-sectional questionnaire survey; and CDSS finalization based on iterative prototyping and survey findings. The survey assessed requirements for information display, alerting, explainability, intervention support, and workflow integration among anesthesia-related professionals in China. RESULTS Three rounds of focus group discussion and iterative prototyping generated a preliminary prototype comprising candidate modules for information display, alerting, explainability, intervention support, and workflow integration. A total of 2401 valid questionnaires were analyzed. Respondents generally preferred direct risk presentation, probability-based alerting, interpretable displays of modifiable risk factors, actionable intervention support, and integration within existing clinical platforms. These findings informed the final specification of an integrated CDSS within the anesthesia information system, including dynamic risk prediction, threshold-based alerting, explainable risk attribution, and evidence-informed intervention recommendations. CONCLUSIONS In this user-centered design study, anesthesia professionals identified key requirements for perioperative CDSS, including direct information display, clinically meaningful alerts, explainable risk-factor presentation, actionable recommendations, and workflow integration. These findings may inform the translation of perioperative prediction models into decision support tools that are more usable and acceptable in routine anesthesia practice.

Digital health interventions (DHIs), delivered via digital platforms such as internet-based programs, mobile applications or short messages, may improve patient-reported outcomes (PROs), but comparative effectiveness is unclear. We conducted a network meta-analysis of randomized controlled trials in adults undergoing elective surgery under general anesthesia, identified in PubMed, Embase, CENTRAL, and Web of Science to March 1, 2025. Standardized mean differences (SMDs), mean differences (MDs) with minimal important differences (MIDs), and 95% CIs were estimated. Risk of bias was assessed with RoB 2 and certainty of evidence with GRADE. Fifty-six trials (6,154 patients) were included. Extended reality (XR) most effectively reduced perioperative anxiety (SMD 0.60; 95% CI 0.37–0.84; MD 8.05; MID 6.71; moderate-certainty). For postoperative pain, mobile applications (SMD 0.64; 95% CI 0.32–0.95; MD 1.36; MID 1.0; moderate-certainty) and XR (SMD 0.51; 95% CI 0.26–0.76; MD 1.09; MID 1.0; moderate-certainty) were probably effective. For quality of life, 2D video yielded the greatest gain (SMD 0.99; 95% CI 0.11–1.88; MD 0.11; MID 0.05; high-certainty). XR also improved satisfaction (SMD 1.27; 95% CI 0.63–1.91; MD 1.91; MID 0.75; moderate-certainty). These findings suggest that DHIs may improve perioperative PROs.

Background: Post-induction hypotension (PIH) affects 20–40% of surgical patients and is revealed to be associated with adverse outcomes. However, current understanding of PIH and its clinical implications remains limited by a potentially oversimplified description of these longitudinal and complex hemodynamic changes. Methods: We conducted a multicenter, large sample study with the INSPIRE cohort and two Chinese cohorts. Heterogeneous linear mixed effects models were applied to determine the distinct PIH trajectory phenotypes among patients. Relationships among trajectory patterns and postoperative complications were analyzed with the hierarchical logistic regression model by adjusting multiple confounding factors. A trajectory prediction model and a classification model were developed and validated based on the XGBboost model, and the contributing factors were explored with the SHAP (SHapley Additive exPlanations) value approach. Results: With 169 470 patients included in analysis, four distinct PIH trajectory phenotypes were identified and validated: Gradual Decline (66.0%), Sharp Drop (22.9%), Drop and Recovery (8.0%), and Overshoot and Rise (4.1%). The Sharp Drop and Overshoot and Rise phenotypes demonstrated significantly higher risks for adverse outcomes compared to non-PIH patients, including prolonged ICU stay [odds ratio (OR) 2.24 and 2.71], mechanical ventilation (OR 2.47 and 2.94), and postoperative pulmonary complications (OR 1.63 for both). The PIH trajectory prediction models were shown with a macro-average AUROC of 0.8240 [95% confidence interval (CI) 0.7699–0.8927] by the internal validation. ASA (American Society of Anesthesiologists) classification, age, anesthetic agents, body mass index (BMI), coexisting hypertension, and vasopressor were found to be the key determinants of trajectory patterns. We also developed an open-source computerized platform to facilitate the local implementation and validation of PIH trajectory prediction and classification. Conclusions: This study identifies and validates four distinct PIH trajectory phenotypes with different prognostic implications and contributing factors, providing new insights for optimal targets of perioperative hemodynamic management. We propose a methodological framework for blood pressure trajectory analysis that can be adapted to different clinical settings.

Early prediction of the major perioperative adverse events is of great significance for reducing mortality, morbidity and medical costs. Machine learning (ML) leverages the capacity for predicting the probability of perioperative adverse events, revealing the promise to facilitate risk stratification, tailored prevention, and individualized perioperative management. However, significant heterogeneity has been demonstrated in the model’s performance of discrimination, calibration, interpretability, and transparency among studies, which raises concerns over their clinical efficacy and usability. A lack of guidance for non-expert medical professionals and stakeholders hinders rigorously conducting research with standard procedure, appropriate methodology, consistent measures, and complete reports. We established a multidisciplinary team consisting of clinicians, data scientists, computer scientists. Multiple libraries including Medline, PubMed, Web of Science, Embase, and CINAHL were searched. We comprehensively summarized critical issues within the entire workflow of ML-based model study, including scenarios and problems, task definition, data collecting and processing, feature representation, model development and validation, clinical implementation and evaluation, aiming to provide guidance and insights for this topic. This review provides a practical checklist of the ML workflow tailored for perioperative teams, bridging technical innovations with clinical translation.

Postoperative cognitive dysfunction (POCD), a heterogeneous spectrum of surgery/anesthesia-associated neurocognitive impairments, represents a critical clinical challenge due to its associations with prolonged hospitalization, increased mortality, and accelerated long-term cognitive deterioration. The development of targeted perioperative management strategies fundamentally depends on systematic risk factor identification, though current research remains hampered by significant heterogeneity and incomplete evidence synthesis across observational studies. To address this knowledge gap, we constructed the first panoramic knowledge base integrating multidimensional POCD risk factors through systematic analysis of 2,837 PubMed-indexed clinical studies (1991-20 May 2024). Utilizing a rigorous protocol developed by a multidisciplinary expert panel, 860 distinct predictive risk factors were extracted following stringent inclusion criteria and manual curation. This knowledge synthesis framework enables hierarchical visualization of risk elements spanning demographic, clinical, anesthetic, surgical, and biomarker domains. Postoperative cognitive dysfunction knowledge base (POCDKB) is a retrospective database study that serves dual translational purposes: providing clinicians with evidence-based decision support for personalized perioperative treatment plans, and establishing standardized data parameters for future mechanistic investigations. Implemented as a dynamic platform: (http://sysbio.org.cn/pocdkb/), the resource features continuous evidence updates and collaborative data-sharing infrastructure to advance POCD research and therapeutic innovation.

Background: Perioperative adverse events (PAEs) pose a substantial global health burden, contributing to elevated morbidity, mortality, and health care expenditures. The adoption of clinical decision support systems (CDSS), particularly mobile-based solutions, offers a promising avenue to address these challenges. However, successful implementation hinges on understanding anesthesia providers’ knowledge, attitudes, and willingness to embrace such technologies. Objective: This study aimed to evaluate the knowledge, attitudes, and willingness of Chinese anesthesia professionals to adopt a mobile CDSS for PAE management, and to identify key factors influencing its implementation. Methods: A nationwide cross-sectional survey was conducted among anesthesia providers in China from September 5 to December 31, 2023. Participants included anesthesiologists and nurse anesthetists, who play pivotal roles in perioperative care. A 51-item questionnaire, structured around the Knowledge-Attitude-Practice (KAP) framework, was distributed via WeChat through professional anesthesia associations. The questionnaire covered four domains: (1) demographic characteristics, (2) knowledge assessment, (3) attitude evaluation, and (4) practice willingness. Multivariable regression analyses identified predictors of KAP outcomes, with sensitivity analyses focusing on nurse anesthetists. Results: The study included 2440 anesthesia professionals (2226 anesthesiologists and 214 nurse anesthetists). Overall, 87.3% (2130/2440) expressed willingness to adopt the CDSS, with 87.5% (1947/2226) of anesthesiologists and 85.5% (183/214) of nurse anesthetists showing readiness. However, only 39.2% (956/2440) were satisfied with existing incident management systems. Key findings indicated that higher knowledge scores were associated with female gender (coefficient=0.19, P=.003), advanced education, and lack of previous informatics experience (coefficient=0.29, P<.001). Nurse anesthetists scored lower than anesthesiologists (coefficient=–0.76, P<.001). Negative attitudes were more prevalent among older practitioners (coefficient=–0.13, P<.001), females (coefficient=–0.66, P<.001), nurse anesthetists (coefficient=–1.12, P=.003), and those without prior PAE exposure (coefficient=–0.97, P<.001). Higher willingness was observed among practitioners in Southwest China (coefficient=0.10, P=.048), those with positive attitudes (coefficient=0.06, P<.001), and those dissatisfied (coefficient=0.32, P<.001) or neutral (coefficient=0.11, P=.02) towards existing systems. Infrequent departmental incident discussions would reduce practice willingness (coefficient=–0.08, P=.01). Conclusions: This national study highlights a strong readiness among Chinese anesthesia professionals to adopt mobile CDSS for PAE management. However, critical barriers, including role-specific knowledge disparities and ineffective organizational communication, must be addressed to ensure successful implementation. Collaborative efforts among local authorities, health care facilities, anesthesia departments, and technology developers are essential to design and implement tailored strategies. Key recommendations include interdisciplinary training programs to enhance nurse anesthetists’ competencies, institution-level incentives to promote incident reporting, and user-centered CDSS designs that prioritize seamless integration into clinical workflows. These measures are vital for improving perioperative incident reporting systems and ultimately advancing the safety and outcomes of surgical patients.

Major adverse cardiovascular events (MACEs) in geriatric patients are an important cause of increased mortality and morbidity. The results of current studies regarding the predictive value of the NT-proBNP, H-FABP, and AUB-HAS2 scales for cardiovascular complications are inconsistent, and there is no relevant large sample study. Therefore, this study aimed to investigate whether preoperative NT-proBNP, H-FABP, and AUB-HAS2 alone or in combination can effectively predict postoperative cardiovascular complications in geriatric patients. A total of 1736 geriatric patients (aged ≥ 65 years) who were scheduled for elective non-cardiac surgery under general anesthesia were enrolled. AUB-HAS2 risk assessment is required for each patient, and blood was collected 1 h before surgery for the measurement of NT-proBNP and H-FABP. The primary outcomes were MACEs within 30 days after surgery. The secondary outcomes were other complications. Its predictive value was analyzed by receiver operating characteristic (ROC) curves. Of the 1736 patients, 71 (4.1%) had MACEs. NT-proBNP was a predictor of MACEs (AUC = 0.763; 95% CI 0.695–0.832; P < 0.001). When H-FABP was combined with AUB-HAS2, AUB-HAS2 increased the predictive value of H-FABP (AUC = 0.736; 95% CI 0.673–0.799; P < 0.001). Multiple logistic regression analysis revealed increased predictive value of the modified AUB-HAS2 scale for MACEs (AUC = 0.794, 95% CI = 0.737–0.851, P < 0.001). Our study revealed the predictive efficacy and prognostic value of NT-proBNP, H-FABP and the AUB-HAS2 score alone or in combination for postoperative MACE risk assessment in geriatric patients undergoing non-cardiac surgery. This trial was registered at the Chinese Clinical Trial Registry (2019/09/27 ChiCTR1900026223, https://www.chictr.org.cn).