Xin Y, He B, Wei XH, Yan YL, Huang C, Gao CY, Wang S, Zhang GM, Li R, Wu Y. Construction and validation of a predictive model for the risk of emergence delirium in older adult patients. World J Psychiatry 2026; 16(6): 115839 [DOI: 10.5498/wjp.v16.i6.115839]
Corresponding Author of This Article
Ying Wu, Chief Nurse, Department of Nursing, Tongren Hospital Shanghai Jiao Tong University School of Medicine, No. 1111 Xianxia Road, Changning District, Shanghai 200335, China. 1285042230@qq.com
Research Domain of This Article
Psychiatry
Article-Type of This Article
research-article
Open-Access Policy of This Article
This article is an open-access article which was selected by an in-house editor and fully peer-reviewed by external reviewers. It is distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/
Baishideng Publishing Group Inc, 7041 Koll Center Parkway, Suite 160, Pleasanton, CA 94566, USA
Share the Article
Xin Y, He B, Wei XH, Yan YL, Huang C, Gao CY, Wang S, Zhang GM, Li R, Wu Y. Construction and validation of a predictive model for the risk of emergence delirium in older adult patients. World J Psychiatry 2026; 16(6): 115839 [DOI: 10.5498/wjp.v16.i6.115839]
Yi Xin, Department of Anesthesiology, Fudan University Shanghai Cancer Center, Shanghai 200032, China
Bin He, Department of Cardiovascular Surgery, Peking University Shenzhen Hospital, Shenzhen 518036, Guangdong Province, China
Xiao-Hui Wei, Rui Li, Department of Nursing, Shanghai Pulmonary Hospital, Shanghai 200082, China
Ya-Ling Yan, Department of Vascular Surgery, West China Hospital Sichuan University, Chengdu 610041, Sichuan Province, China
Chen Huang, Chun-Yan Gao, Shuo Wang, Guang-Ming Zhang, Department of Anesthesiology, Tongren Hospital Shanghai Jiao Tong University School of Medicine, Shanghai 200335, China
Ying Wu, Department of Nursing, Tongren Hospital Shanghai Jiao Tong University School of Medicine, Shanghai 200335, China
Author contributions: Xin Y, He B, Wu Y, and Li R contributed to the research design and manuscript writing; Xin Y and He B jointly contributed to experimental design, data collection, and analysis, playing a crucial role in ensuring the reliability and validity of the research, and collaborated in writing and revising the manuscript, thereby enhancing the overall quality of the manuscript as co-first authors; Xin Y, He B, Wei XH, Yan YL, Huang C, Gao CY, Wang S, and Zhang GM contributed to the data collection; Wu Y and Li R possess expertise in the fields of anesthesiology and nursing, providing crucial professional support and advice for the research, and they serve as leaders and mentors within the research team, playing significant organizational and guidance roles throughout the entire research project as co-corresponding authors. All authors contributed to the article and approved the submitted version.
Supported by the 2024 Shanghai Jiao Tong University School of Medicine Nursing Research Top Priority Project, No. Jyhz2410; and Advanced Anesthesia Specialty Nursing Training Base, No. 2022zkh1jd.
Institutional review board statement: The study was reviewed and approved by the Shanghai Tongren Hospital’s Ethics Committee (approval No. Tong Ren Lun Audit 2024-002-02).
Clinical trial registration statement: The research did not involve any clinical trials or interventional procedures and is an observational cohort study. As such, it does not meet the criteria for clinical trial registration.
Informed consent statement: All study participants, or their legal guardian, provided informed written consent prior to study enrollment.
Conflict-of-interest statement: All the authors report no relevant conflicts of interest for this article.
CONSORT 2010 statement: The authors have read the CONSORT 2010 Statement, and the manuscript was prepared and revised according to the CONSORT 2010 Statement.
Data sharing statement:sharing statement: No additional data are available.
Corresponding author: Ying Wu, Chief Nurse, Department of Nursing, Tongren Hospital Shanghai Jiao Tong University School of Medicine, No. 1111 Xianxia Road, Changning District, Shanghai 200335, China. 1285042230@qq.com
Received: October 27, 2025 Revised: December 24, 2025 Accepted: February 12, 2026 Published online: June 19, 2026 Processing time: 213 Days and 21.2 Hours
Abstract
BACKGROUND
Emergence delirium (ED) is a common postoperative complication in older adult patients, posing a significant burden on both patients and medical staff. Despite its prevalence, there is a notable lack of research focused on identifying predictive factors and constructing models for ED in the post-anesthesia care unit. Therefore, developing a risk prediction model for ED in older adult patients is imperative. We anticipate that such a model would demonstrate strong predictive efficacy and be applicable in clinical settings.
AIM
To develop and validate an ED risk-prediction model for early intervention in older adults.
METHODS
This study enrolled 705 older surgical patients (January 2024 to October 2024) for modeling and 115 (November 2024 to December 2024) for validation. Using least absolute shrinkage and selection operator and multivariable logistic regression, we developed a predictive model with an online dynamic nomogram. Internal (10-fold cross-validation) and external validation demonstrated strong discrimination, calibration, and clinical utility.
RESULTS
The incidence of ED in older adult patients postoperatively was found to be 17.16%. Independent risk factors for postoperative ED included preoperative Mini-Mental State Examination score, preoperative albumin level, surgical duration, surgical risk score, number of indwelling catheters, and extubation time (all P < 0.05). The model demonstrated the area under curve (AUC) of 0.924 [95% confidence interval (CI): 0.897-0.951], with the calibration curve closely aligning with the ideal curve. The Hosmer-Lemeshow test yielded χ2 = 7.934, P = 0.541, indicating good clinical utility. Internal validation resulted in an AUC of 0.920 (95%CI: 0.571-0.959), while external validation showed an AUC of 0.931 (95%CI: 0.866-0.997). The calibration curve for the validation cohort closely matched the ideal curve, with the Hosmer-Lemeshow test showing χ2 = 5.772, P = 0.763, further supporting its clinical applicability.
CONCLUSION
The dynamic nomogram accurately predicts ED risk in older adults, aiding early identification and clinical intervention.
Core Tip: In this study, 705 older adult surgical patients were selected as the modeling cohort, while 115 older adult surgical patients from different time periods were chosen for external validation. The results indicated that the incidence of postoperative emergence delirium in older adult patients was 17.16%. Preoperative Mini-Mental State Examination score, preoperative albumin level, surgical risk score, the number of indwelling catheters, and extubation time were identified as independent factors influencing the risk of emergence delirium. The risk nomogram model developed based on these factors demonstrated robust predictive efficacy.
Citation: Xin Y, He B, Wei XH, Yan YL, Huang C, Gao CY, Wang S, Zhang GM, Li R, Wu Y. Construction and validation of a predictive model for the risk of emergence delirium in older adult patients. World J Psychiatry 2026; 16(6): 115839
Delirium, a neuropsychiatric syndrome characterized by acute onset and fluctuating course, manifests as alterations in consciousness, inattention, and disturbances in the sleep-wake cycle[1]. As an early stage of postoperative delirium, emergence delirium (ED) typically occurs during the transition from general anesthesia to full recovery, spanning the period from the end of surgery to discharge from the post-anesthesia care unit (PACU)[2]. Due to age-related physiological decline, older adults face a significantly elevated risk of perioperative complications, with ED incidence reaching as high as 37.6%[3]. Clinically, ED predominantly presents as the hypoactive or mixed subtype, while the hyperactive form - though more readily recognized - is relatively less common[4]. This phenotypic variability contributes to considerable underdiagnosis, often leading to an underestimation of its severity and clinical impact. ED can precipitate adverse events such as wound dehiscence, self-extubation, falls, and violent behavior, thereby increasing the workload of anesthesia and nursing staff. In the longer term, it is associated with persistent delirium and cognitive decline[5-7]. Beyond compromising patient outcomes, ED escalates healthcare costs, imposes substantial economic burdens on families, and challenges healthcare systems[8]. Notably, prolonged or inadequately treated delirium is linked to a higher incidence of lasting cognitive impairment[9]. Despite the recognized importance of early detection, systematic research and validated predictive models for ED in older surgical patients remain scarce.
Studies suggest that 30%-40% of delirium cases may be preventable through proactive management[10]. A scientifically robust and accurate predictive model for ED is therefore crucial for early identification of high-risk patients, enabling targeted interventions and optimizing perioperative resource allocation[11]. To address this need, we conducted a prospective cohort study integrating multidimensional preoperative, intraoperative, and postoperative variables. Using least absolute shrinkage and selection operator (LASSO) regression for feature selection, we developed a parsimonious yet high-performance predictive model. The model underwent rigorous internal and external validation to ensure its feasibility and reliability in clinical practice. Furthermore, we implemented it as an interactive web-based dynamic nomogram, providing clinicians with an intuitive and efficient tool for real-time risk assessment and decision support. This approach not only holds promise for reducing the incidence of ED but also offers a strategic pathway to enhance perioperative care quality and resource utilization.
MATERIALS AND METHODS
Study participants
This cohort study involved selecting geriatric surgical patients at a tertiary hospital in China through convenience sampling from January 2024 to December 2024. Patients enrolled from January 2024 to October 2024 constituted the training group for model fitting and internal validation. In contrast, those enrolled from November 2024 to December 2024 formed the validation group for time-series external validation of the model developed using the training group. Inclusion criteria: (1) Patients aged 60 years or older who underwent general anesthesia; (2) Patients who underwent elective surgery; (3) Patients transferred to the PACU after surgery; and (4) Patients who provided informed consent and voluntarily participated in this study. Exclusion criteria: (1) History of dementia or psychosis; (2) Patients undergoing neurosurgery or cardiac surgery; (3) Patients unable to cooperate in completing the assessment; and (4) Persistent coma (Glasgow Coma Scale ≤ 8 points) or deep sedation [Richmond Agitation-Sedation Scale (RASS) < -3 points] during PACU resuscitation.
The pilot study indicated that the incidence of postoperative ED in older adult patients was approximately 17%. The sample size was determined using a binary outcome prediction modeling formula[12]. A tolerance error of δ = 0.03 was set, and considering a 10% attrition rate, the final sample size required was calculated to be at least 670 cases. The study was approved by the hospital’s Ethics Committee (approval No. Tong Ren Lun Audit 2024-002-02), and all participants provided informed consent and voluntarily participated.
Design
Tools: Utilizing literature analysis, expert consultations, and group discussions, we identified candidate variables related to ED in older adult patients. We developed a “Questionnaire on risk factors for ED in older adult patients during the awakening period”. This questionnaire comprises five sections: (1) Demographic information: Sex, age, body mass index, history of diabetes, history of stroke, and smoking history; (2) Preoperative factors: Preoperative Mini-Mental State Examination (MMSE) score (0-30, with higher scores indicating better cognitive function), American Society of Anesthesiology (ASA) classification, preoperative albumin level, and preoperative low hemoglobin level; (3) Perioperative medication factors: Use of dexmedetomidine, benzodiazepines, and anticholinergic agents; (4) Intraoperative factors: Surgical duration, nerve blocks, type of surgery, surgical risk score, intraoperative blood loss, intraoperative blood pressure variability (the extent of mean arterial pressure fluctuations, calculated via variance[13]) and number of indwelling catheters; and (5) Postoperative factors: PACU admission temperature and extubation time (the duration from the end of surgery to the removal of the endotracheal tube or laryngeal mask).
Diagnostic methods: This study employed a two-step approach for screening postoperative ED in older adult patients. The first step involved assessing the patient’s level of consciousness using the RASS, which ranges from +4 to -5, encompassing 10 sedation levels, each corresponding to a specific state of consciousness. A RASS score of less than -3 indicates the patient is unresponsive, necessitating reassessment only after achieving a state of light sedation. Conversely, a RASS score of -3 or higher allows for the implementation of the Confusion Assessment Method for the intensive care unit (ICU) to evaluate the patient. The second step utilized the Confusion Assessment Method for the ICU, which includes: (1) An acute change or fluctuation in consciousness; (2) Attention deficits; (3) Disorganized thinking; and (4) An altered level of consciousness. The presence of ED is indicated when both criteria (1) and (2) are positive, along with either (3) or (4) being positive. Following standardized training, two assessors, including the researcher and a PACU anesthesiology nurse, evaluated the occurrence of ED every 10 minutes after the patient entered the PACU and extubation, recording the time and outcome of ED. A positive assessment at any time point was considered indicative of ED. In cases of disagreement between the two assessors, the on-duty physician made the final determination.
Data collection: Patients were assessed by an investigator trained in administering the MMSE one day before surgery. For patients with a preoperative MMSE score below 23, anesthesiologists conducted a thorough review of documented dementia history combined with structured clinical interviews to exclude individuals with chronic progressive dementia. Postoperative information was gathered from patients in the PACU following surgery. Additional general information was sourced from the electronic medical records or nursing notes. Two researchers verified the collected data, which were then organized into Excel spreadsheets to create a database.
Statistical analysis: Data analysis was conducted using IBM SPSS 26.0 and R 4.4.0. Dummy variables were created for unordered categorical variables (surgical type, PACU admission temperature). Baseline characteristics of patients were expressed as n (%) for categorical data, and as mean ± SD or medians with interquartile ranges for continuous data. Univariate analyses for categorical data were performed using the χ2 test or Fisher’s exact test, while continuous data were analyzed using independent samples t-tests or the non-parametric Mann-Whitney U test, with a significance level set at P < 0.05.
To address multicollinearity among variables, statistically significant predictors were incorporated into the LASSO regression framework. LASSO regression is a statistical method designed for variable selection and regularization, aiming to enhance both predictive accuracy and model interpretability[14]. By applying an L1-norm penalty, LASSO shrinks a subset of regression coefficients to zero, thereby effectively identifying the subset of covariates most strongly associated with the response variable. This approach is particularly advantageous for high-dimensional data analysis, as it mitigates multicollinearity, guards against overfitting, and improves model robustness[15].
The optimal penalty parameter λ was determined via tenfold cross-validation, which selected the most relevant variables. These retained predictors were subsequently entered into a logistic regression model. The final model was chosen as the subset yielding the minimum Akaike Information criterion, and it served as the basis for constructing a nomogram-based prediction tool. Furthermore, to enhance clinical usability, an interactive web-based dynamic nomogram was developed to facilitate real-time risk assessment. Internal validation was performed using ten-fold cross-validation, while external validation was conducted with an independent validation cohort dataset. Model performance was evaluated using the area under the receiver operating characteristic area under curve (AUC) to assess discrimination ability, and calibration was assessed using calibration curves and the Hosmer-Lemeshow test. Additionally, decision curve analysis was employed to evaluate the clinical utility of the model.
RESULTS
Research design process
After excluding 25 patients lost to follow-up, 19 patients who remained comatose during the PACU stay, 17 patients admitted to the ICU postoperatively, and 6 patients who voluntarily withdrew, a total of 820 older adult surgical patients meeting the inclusion and exclusion criteria were enrolled in this study. Among these, 705 patients from January 2024 to October 2024 constituted the modeling cohort, while 115 patients from November 2024 to December 2024 served as the validation cohort. The study design flowchart is illustrated in Figure 1.
Figure 1 The study design flowchart.
PACU: Post-anesthesia care unit; ICU: Intensive care unit; ROC: Receiver operating characteristic; DCA: Decision curve analysis; LASSO: Least absolute shrinkage and selection operator.
Univariate analysis of demographic characteristics in older adult postoperative patients and the incidence of ED
A total of 820 older adult surgical patients meeting the inclusion criteria were included in this study. The training group comprised 705 patients, with 121 (17.16%) experiencing ED and 584 (82.84%) not experiencing it. The validation group included 115 patients, with 19 (16.52%) experiencing ED and 96 (83.48%) not.
Univariate analysis of postoperative ED in older adult patients revealed that 16 variables demonstrated statistically significant differences, including age, preoperative MMSE score, preoperative albumin level, surgical duration, intraoperative blood loss, intraoperative blood pressure variability, number of indwelling catheters, extubation time, ASA classification, surgical risk score, history of diabetes, history of stroke, preoperative low hemoglobin level, use of benzodiazepines, type of surgery, and PACU admission temperature (all P < 0.05). Detailed results are presented in Table 1.
Table 1 General information about the patients in both the training and validation groups, n (%)/median (interquartile range.
LASSO regression analysis of the ED in older adult patients
The analysis incorporated 16 significant univariate variables identified through univariate analysis into the LASSO regression to determine the factors with the greatest clinical relevance. To enhance the simplicity and practicality of the model for clinical application, we selected λ as lambda.1se as the optimal value. At this point, the model reduced the number of the most meaningful variables to six, which include the preoperative MMSE score, preoperative albumin level, surgical duration, surgical risk score, number of indwelling catheters, and extubation time. Further details can be found in Figure 2.
Figure 2 Least absolute shrinkage and selection operator regression analysis of the emergence delirium in older adult patients.
A: Least absolute shrinkage and selection operator (LASSO) regression coefficient curve for variables (a illustrates the coefficients of various variables as a function of λ. As the value of λ increases, many variable coefficients are progressively driven toward zero, highlighting the effectiveness of LASSO in variable selection); B: Results of LASSO ten-fold cross-validation (in B, the two vertical lines represent critical points in the model selection process. The left black vertical line, labeled “lambda.min”, corresponds to the value of λ that achieves the lowest mean squared error during tenfold cross-validation, indicating the model’s optimal predictive performance. The right black vertical line, labeled “lambda.1se”, represents the λ value obtained by adding one standard error to “lambda.min”. This choice aims to produce a more streamlined model that typically balances strong predictive capability with a reduced number of variables, thereby enhancing the model's practicality in clinical applications).
Multivariate logistic regression analysis of the ED in older adult patients
Using ED as the dependent variable, the six optimal variables identified through LASSO regression were incorporated as independent variables into four multifactorial logistic regression models: Forward stepwise, backward stepwise, bidirectional, and overall logistic regression. The results indicated that the Akaike information criterion values for all four models were identical at 353.364, allowing any of these models to be selected as the optimal subset model. Ultimately, the following six variables were determined to be independent predictors of postoperative ED in older adult patients: Preoperative MMSE score [odds ratio (OR) = 0.519, 95% confidence interval (CI): 0.450-0.599, P < 0.001], preoperative albumin level (OR = 0.908, 95%CI: 0.852-0.969, P = 0.004), surgical duration (OR = 1.006, 95%CI: 1.002-1.010, P = 0.017), surgical risk score (OR = 1.594, 95%CI: 1.048-2.424, P = 0.029), number of indwelling catheters (OR = 1.354, 95%CI: 1.011-1.813, P = 0.042), and extubation time (OR = 1.021, 95%CI: 1.003-1.039, P = 0.029). Detailed results are presented in Table 2.
Table 2 Multivariate analysis of emergence delirium in postoperative older adult patients (n = 705).
Construction of a nomogram to predict the risk of the ED in older adult patients
Based on the results of the multifactorial logistic analysis, a nomogram risk prediction model for ED in older adult patients was developed using the six independent influencing factors identified. Each variable is assigned a score, and the total score is calculated by summing all variable scores. This cumulative score corresponds to the risk probability of postoperative ED, which can be read from the scale at the bottom of the ED risk probability chart. For example, an older adult surgical patient with a preoperative MMSE score of 21, preoperative albumin level of 32 g/L, a surgery duration of 100 minutes, a surgical risk score of 0, one indwelling catheter, and an extubation time of 45 minutes corresponds to an approximate postoperative ED probability of 89.9%. Additionally, this study developed an online dynamic nomogram using the Shinyapps platform to facilitate effective risk prediction of postoperative ED in older adult patients for clinical healthcare providers (URL: https://gtxy.shinyapps.io/DynNomapp/). Refer to Figure 3 for more details.
Figure 3 Different presentation forms of nomogram.
A: Nomogram for predicting the risk of emergence delirium (ED) in postoperative older adult patients; B: Example application of the nomogram for predicting the risk of ED in postoperative older adult patients; C: Example application of the web-based dynamic nomogram for predicting the risk of ED in postoperative older adult patients. MMSE: Mini-Mental State Examination; ED: Emergence delirium.
Internal and external validation of the risk prediction model of the ED in older patients
The AUC for the modeling cohort in this study was 0.924 (95%CI: 0.897-0.951), with a cutoff diagnostic threshold of 0.182, sensitivity of 0.860, and specificity of 0.877. Additionally, the results of ten-fold cross-validation indicated an AUC of 0.920 (95%CI: 0.571-0.959) for the model. External validation using a validation cohort demonstrated an AUC of 0.931 (95%CI: 0.866-0.997), with a cutoff diagnostic threshold of 0.144, sensitivity of 0.895, and specificity of 0.854, as illustrated in Figure 4. All three AUC values exceeded 0.75, indicating good discriminative ability of the model. The slight differences in AUC values suggest that the model performs similarly in both the modeling and validation cohorts, demonstrating robust generalizability. Calibration curves for both cohorts closely approached the ideal curve, with the Hosmer-Lemeshow test yielding χ2 = 7.934, P = 0.541 for the modeling cohort and χ2 = 5.772, P = 0.763 for the validation cohort, indicating good calibration of the model, as detailed in Figure 5. In the clinical decision curves for both cohorts, the horizontal solid line labeled “None” indicates that all samples are negative and no intervention is performed, resulting in a clinical benefit of zero. The line labeled “All” indicates that all samples are positive and receive intervention, with no additional benefit when the probability threshold exceeds 18%. The figures illustrate that this predictive model expands the benefit threshold, as shown in Figure 6.
Figure 4 Receiver operating characteristic curve.
A: Receiver operating characteristic curve for the prediction model of emergence delirium in postoperative older adult patients; B: Receiver operating characteristic curve for the validation cohort of the prediction model for emergence delirium in postoperative older adult patients. ROC: Receiver operating characteristic; AUC: Area under curve.
Figure 5 Calibration curve.
A: Calibration curve of the prediction model for emergence delirium in postoperative older adult patients; B: Calibration curve of the validation cohort for the prediction model of emergence delirium in postoperative older adult patients. ROC: Receiver operating characteristic.
Figure 6 Decision curve.
A: Decision curve of the prediction model for emergence delirium in postoperative older adult patients; B: Decision curve of the validation cohort for the prediction model of emergence delirium in postoperative older adult patients.
DISCUSSION
The current situation of ED in older adult patients after surgery
This study found that the incidence of ED in older adult patients postoperatively was 17.16%, which is lower than previously reported levels[3,5,16,17]. This discrepancy may be attributed to the inclusion and exclusion criteria of our study. We excluded patients undergoing cardiac and neurosurgery due to the risk of ED associated with cardiopulmonary bypass or direct manipulation of the central nervous system, as well as those with visual or auditory impairments, which could hinder assessment and introduce bias. To minimize confounding factors and enhance the clinical applicability of the nomogram, we focused on more common surgical procedures. Notably, existing research indicates that patients with sensory impairments are at a higher risk for delirium[18], potentially contributing to the lower incidence of ED observed in our study. Furthermore, our findings exceed those reported by Ramroop et al[19] and Fields et al[20], likely due to our focus on surgical patients aged 60 and above. Compared to younger adults, older adult surgical patients exhibit poorer tolerance to incision pain, diminished sleep quality, and reduced metabolic capacity for anesthetic agents[16,21], making them more susceptible to ED. Therefore, it is crucial to analyze the influencing factors, proactively prevent ED, and promptly identify and correct its underlying causes to facilitate rapid resolution and optimize long-term outcomes.
The occurrence of ED in older adult patients postoperatively is influenced by multiple factors
MMSE: The preoperative MMSE reflects the cognitive function status of patients prior to surgery. This study found that preoperative cognitive impairment is a significant risk factor for ED in older adult patients (OR = 0.519). For each one-point increase in the MMSE score, the risk of developing ED postoperatively decreases by 51.9%, consistent with previous findings[5,22,23]. Research indicates that patients with cognitive impairment often exhibit neuronal damage and neurotransmitter dysregulation, which affect neural network connectivity in the brain[24]. Similarly, delirium results from the collapse of functional neural networks[25]. Both conditions share analogous underlying pathological mechanisms, and additional factors such as exogenous inflammatory stimuli, neuronal metabolic abnormalities, and pharmacological influences can further disrupt neural connectivity, promoting the onset and progression of neurodegenerative diseases[26]. Consequently, patients with preoperative cognitive impairment are more susceptible to ED. Moreover, cognitive impairment is typically associated with lower cognitive reserve, with higher cognitive reserve reducing the risk of mild cognitive impairment by 47%[27]. The brain reserve theory posits that greater cognitive reserve, characterized by a higher number of neurons and synapses, can more effectively delay cognitive decline[28]. Therefore, enhancing patients’ preoperative cognitive reserve is a crucial preventive measure to improve cognitive function and reduce the risk of postoperative ED. This underscores the importance of anesthetic nurses conducting thorough cognitive assessments prior to surgery and collaborating with the clinical team. For patients with poor cognitive function, interventions such as cognitive training, mindfulness exercises[29], physical training[30], virtual reality rehabilitation[31], and game-based cognitive training software[32] may be employed to enhance cognitive function, increase cognitive reserve, and subsequently lower the incidence of postoperative ED.
Preoperative albumin level: This study found that a lower preoperative serum albumin level is a significant risk factor for ED in older adult patients (OR = 0.908), consistent with previous research[16]. For each 1 g/L increase in preoperative albumin, the risk of developing postoperative ED decreases by 9.2%. Preoperative serum albumin levels not only reflect nutritional status but also pathological and physiological changes associated with inflammatory responses. Firstly, patients with low albumin levels often have insufficient nutritional reserves, leading to decreased resilience against postoperative stress[33]. Low albumin levels are associated with postoperative fluid imbalance and tissue edema, which may further impair cerebral perfusion and oxygen delivery, thereby increasing the risk of ED[34]. Furthermore, research indicates that hypoalbuminemia is linked to inflammatory responses and blood-brain barrier dysfunction. Under physiological stress, compensatory mechanisms in the body can lead to decreased serum albumin levels. Low serum albumin increases vascular permeability, triggering central nervous system inflammation, which can contribute to the onset of ED[20,35]. Therefore, optimizing preoperative albumin levels may be crucial for reducing the incidence of postoperative ED.
For patients with preoperative hypoalbuminemia, healthcare professionals should start by assessing the underlying causes and implement comprehensive, individualized interventions that include nutritional support, inflammation management, and fluid balance. Evaluating nutritional status through albumin levels, body mass index, and nutritional tools such as the Patient-Generated Subjective Global Assessment, while screening for inflammation, edema, and comorbidities, is essential for actively treating underlying conditions. Malnourished patients should increase protein and caloric intake, with nutritional support provided as necessary. Those with inflammation should have inflammatory markers monitored and receive enhanced anti-infective and anti-inflammatory treatment[36]. Nursing care must strictly manage fluid balance and administer albumin as needed. Through multidisciplinary collaboration in implementing individualized interventions, we can improve preoperative nutritional status and reduce the risk of postoperative ED.
Surgical duration: This study found that prolonged surgical duration is a significant risk factor for ED in older adult patients (OR = 1.006), consistent with previous research[20,37]. For each additional minute of surgical time, the risk of developing postoperative ED increases by 0.6%. Surgical duration is directly proportional to anesthesia time; as surgical time extends, the consumption of anesthetic agents rises, leading to prolonged mechanical ventilation, which can adversely affect respiratory and circulatory functions[38], thereby triggering the onset of ED. Moreover, prolonged surgery disrupts the brain’s autoregulatory capacity, increasing the likelihood of conditions such as hypercapnia, blood loss, and hypothermia, which can further impair self-regulation and elevate the risk of ED[39].
To mitigate these risks, hospitals should prioritize the development of specialized surgical teams within the operating room[40]. Preoperatively, surgeons, nurses, and anesthesiologists should enhance their expertise and skills. For complex surgeries, conducting rehearsals and simulations can ensure effective communication and collaboration within the team, thereby improving surgical efficiency. It is essential to ensure that all equipment, instruments, and medications are adequately prepared to prevent delays during surgery. For patients at high risk for ED, selecting more precise techniques or minimally invasive procedures based on their overall condition can help reduce surgical time. Postoperatively, feedback sessions should review the surgical process to identify prolonged stages for targeted optimization and improvement[41]. By enhancing surgical processes and techniques while maintaining quality and safety, we can reduce the risk of ED associated with extended surgical durations.
Surgical risk score: This study found that a higher surgical risk score is a significant risk factor for ED in older adult patients (OR = 1.594), with relatively few related studies in the literature. For each additional point in the surgical risk score, the likelihood of developing postoperative ED increases by 59.4%. The surgical risk score reflects the patient’s overall preoperative condition, the complexity of the surgery, the risk of surgical infection, and the potential for complications[42]. A higher score indicates more severe underlying diseases, greater intraoperative complexity, and an elevated risk of infection, which may lead to an excessive stress response in the body, significantly increasing the risk of ED. A higher surgical risk score is often associated with factors such as significant intraoperative trauma, prolonged surgical duration, increased blood loss, and deeper anesthesia. These factors can trigger systemic inflammatory responses, inadequate cerebral perfusion, and metabolic disturbances, thereby impairing central nervous system function and precipitating postoperative delirium. Additionally, high-risk surgeries may require more complex intraoperative management, including control of hemodynamic fluctuations, the use of multiple medications, and fluid management, all of which can negatively impact the patient’s nervous system and further increase the risk of ED[43].
For patients identified as high surgical risk, a multidisciplinary team should assess their ASA classification, nutritional status, and comorbidities preoperatively to optimize management of underlying conditions and develop individualized surgical and anesthetic plans. During surgery, it is crucial to control the duration of the procedure, accurately regulate the depth of anesthesia, maintain hemodynamic stability, implement strict aseptic techniques, minimize tissue trauma and inflammation, and prioritize minimally invasive techniques. Postoperatively, enhanced monitoring and early intervention are essential for promptly addressing complications such as infections to reduce the risk of ED.
Number of indwelling catheters: This study found that a greater number of indwelling catheters is a significant risk factor for ED in older adult patients (OR = 1.354), consistent with previous research[20]. For each additional indwelling catheter, the likelihood of developing postoperative ED increases by 35.4%. Indwelling catheters serve as crucial interventions for maintaining physiological function postoperatively, but the discomfort and constraints they impose may trigger adverse physiological and psychological stress responses in patients, becoming a key mechanism for the onset of postoperative ED. On one hand, catheter-related pain and discomfort can lead to physical resistance from patients, particularly during the awakening phase from anesthesia, when patients may attempt to remove the catheters, resulting in ED[44]. On the other hand, the presence of multiple catheters can restrict patient mobility, increasing feelings of constraint and psychological distress, which can exacerbate anxiety and further impact central nervous system function[45]. Additionally, prolonged indwelling of catheters may elevate local tissue inflammation and systemic inflammatory responses, potentially triggering neuroinflammation and exacerbating cerebral dysfunction. Specifically, catheter-associated bladder irritation can induce neuroreflective responses that may further disrupt the patient’s nervous system, serving as a potential trigger for ED.
Research suggests that early removal of urinary catheters postoperatively can effectively reduce the incidence of bladder irritation, alleviating discomfort and related psychological stress responses, thereby lowering the rate of ED[46,47]. Therefore, it is crucial to minimize the number of indwelling catheters during surgery and enhance postoperative management. Employing flexible catheters and lubricants can reduce discomfort and irritation. Timely assessment of the necessity for catheters postoperatively and shortening the duration of their use is essential. Optimizing nursing care and providing psychological support can mitigate adverse reactions and reduce the risk of ED. Implementing multifaceted strategies to enhance patient experience and improve care quality is vital.
Extubation time: This study found that prolonged extubation time is a significant risk factor for ED in older adult patients (OR = 1.021), with each additional minute of extubation time increasing the risk of postoperative ED by 2.1%. The findings align with those of Wang et al[48], likely due to the prolonged presence of the endotracheal tube, which can irritate the airway, throat, and vocal cords, leading to local inflammation, injury, or discomfort. This continuous physiological stimulation can heighten the body’s stress response, triggering a systemic stress response syndrome that affects central nervous system function. Furthermore, negative emotions are a critical risk factor for the onset of ED symptoms[49]. Prior to extubation, although patients remain sedated, their central nervous systems are gradually returning to normal function[50]. Patients may subconsciously perceive the presence of the endotracheal tube, which can induce psychological anxiety and stress responses. As the duration of tube placement increases, patients’ anxiety and tension may escalate, further increasing the risk of ED[51].
Thus, anesthesiologists and nursing staff must closely monitor the patient’s awakening process, ensuring that they maintain adequate spontaneous breathing and airway protective reflexes, and perform extubation as soon as it is deemed safe[52]. For patients who have not met extubation criteria for an extended period, a thorough assessment of factors contributing to their delayed recovery is essential. This includes evaluating for residual neuromuscular blockade, unstable respiratory or circulatory function, or neurological complications, and implementing targeted treatment measures as needed to develop a personalized extubation strategy. Post-extubation, anesthetic nurses can provide appropriate psychological support and care to improve the patient’s mental state, reduce stress, alleviate discomfort, and facilitate recovery, ultimately lowering the incidence of ED.
Exploration and innovation of a risk prediction model for ED in older adult patients
Currently, various delirium prediction models have been reported. Among them, the classic delirium prediction model for adult ICU patients (PRE-DELRIC) was developed by Dutch scholar van den Boogaard et al[53] includes ten risk factors: Age, diagnosis category, emergency admission, Acute Physiology and Chronic Health Evaluation II score, infection, coma status, maximum blood urea nitrogen concentration, morphine usage, sedative usage, and metabolic acidosis. This model has been applied in ICU patients across several countries, including Australia, demonstrating good predictive efficacy. In earlier delirium-related risk prediction models, “morphine usage” was a significant assessment indicator. However, with the advent of new opioid medications and the promotion of multimodal analgesia strategies[54], morphine use in clinical practice has significantly decreased. The widespread application of combination analgesia and non-opioid medications has further diminished the relevance of this variable, leading to a gradual decline in the applicability of “morphine usage” in existing ED risk prediction models. In response, Wassenaar et al[55] developed an early prediction model for delirium risk (E-PRE-DELIRIC) based on 2914 patients from 13 ICUs across seven countries, providing important evidence for delirium risk prediction. This model ultimately identified nine predictive factors: Age, history of cognitive impairment, history of alcoholism, treatment experiences, emergency admission, mean arterial pressure upon ICU admission, corticosteroid use, respiratory failure, and blood urea nitrogen level upon ICU admission, allowing for delirium risk assessment at the time of ICU admission. Some foreign scholars conducted comparative analyses of the predictive performance of the E-PRE-DELIRIC model and the PRE-DELIRIC model within the same ICU population, and the results indicated that the predictive efficacy of the E-PRE-DELIRIC model was lower than that of the PRE-DELIRIC model[56,57]. Additionally, because the E-PRE-DELIRIC model’s sample primarily came from a general ICU patient population, it lacks sufficient representativeness in predicting the risk of ED in older adult patients who have just undergone anesthesia, which may lead to biased predictions. Wang et al[48] constructed a nomogram prediction model for ED occurrence in patients undergoing radical thyroid cancer surgery under general anesthesia, incorporating age, ASA classification, dosage of anesthetic agents, extubation time, and PACU stay time as predictive factors, yielding an AUC value of 0.878, indicating high predictive efficacy. However, this study did not consider key factors that may influence ED occurrence during variable selection, such as preoperative cognitive function, preoperative low hemoglobin, preoperative albumin levels, nerve blocks, surgical type, surgical risk, temperature, and the number of indwelling catheters, all of which have been proven to be closely related to ED occurrence in previous studies[5,18,20,58,59]. Furthermore, the study did not address the issue of collinearity among variables, which may affect the model’s stability and interpretability. Additionally, the model did not adequately assess calibration and clinical applicability, and the study subjects were limited to patients undergoing specific types of surgery, restricting its applicability and generalizability to other populations. Although several studies have attempted to construct delirium-related risk prediction models[60,61], existing models still have certain limitations due to differences in research design, variable selection, and model validation. Most models fail to address the unique risk factors associated with the postoperative anesthesia recovery period in older adult patients and lack a unified standard for assessing predictive performance and clinical value. Currently, there is no recognized standardized model for assessing the risk of ED in older adult patients postoperatively.
In the 2017 European Society of Anaesthesiology guidelines, there is a strong emphasis on the importance of recognizing ED in older adult patients during clinical practice[62]. As age increases, the susceptibility factors for ED also rise, and the longer the duration of ED in patients, the later the treatment begins, leading to a higher incidence of cognitive impairment[9]. Therefore, accurately identifying high-risk factors for postoperative ED in older adult patients and implementing early perioperative preventive interventions is crucial. However, there are many factors that influence ED, and some variables may exhibit multicollinearity. To address this, our study employed LASSO regression analysis, which effectively eliminates multicollinearity among variables while mitigating the risk of model overfitting. This approach allows for the construction of a concise optimal model with good predictive performance from both variable selection and model evaluation perspectives. Ultimately, our study identified six independent predictive factors: Preoperative MMSE score, preoperative albumin level, surgical duration, surgical risk score, number of indwelling catheters, and extubation time. These factors are easily obtainable clinical indicators, facilitating practical application. The constructed nomogram risk prediction model achieved an AUC value of 0.924, slightly higher than that of existing studies, indicating superior predictive efficacy. Compared to previous models, our study utilized fewer predictive factors, requiring only six indicators for risk assessment, thereby simplifying the clinical workflow and enhancing practicality and generalizability. The dynamic web-based nomogram, developed using the R language, visualizes the logistic regression results, streamlining the calculation process and increasing clinical applicability. Healthcare professionals can use this model for preliminary screening of ED in older adult surgical patients, subsequently providing supportive or restorative treatments to correct risk factors, thus improving prevention, management, and handover processes related to ED, ultimately reducing the incidence of postoperative ED and its long-term consequences in older adult patients.
Clinical application value of the risk prediction model for postoperative ED in older adult
The nomogram prediction model developed in this study integrates six key perioperative variables - preoperative MMSE score, preoperative albumin level, surgical duration, surgical risk score, number of indwelling catheters, and extubation time - enabling a quantitative assessment of an individual patient’s risk for ED. This provides an evidence-based foundation for clinical nursing decisions. Implemented as a web-based dynamic nomogram, the tool allows nurses to input relevant parameters at preoperative, intraoperative, and postoperative stages, generating real-time, individualized risk predictions. The visual, interactive nature of the model intuitively demonstrates the relative contribution of each predictor, facilitating the design of targeted nursing interventions. For seamless integration into clinical workflows, the nomogram can be embedded within the hospital information system, enabling automatic data extraction to minimize manual entry errors and enhance assessment efficiency.
Risk stratification based on the model guides rational allocation of nursing resources. Preoperative assessment and management: Evaluation of the preoperative MMSE score and albumin level enables the early identification of high-risk patients with baseline cognitive impairment or malnutrition. Preoperative cognitive assessment should be implemented to identify patients with underlying cognitive impairment, enabling the early initiation of targeted cognitive training[63]. Concurrently, nutritional interventions - such as protein and micronutrient supplementation - are recommended for patients with hypoalbuminemia to optimize physiological reserve and enhance surgical preparedness[64]. The surgical duration and surgical risk score reflect procedural complexity and physiological stress. Intraoperative management should focus on optimizing surgical planning to minimize delays and avoid unnecessarily invasive techniques, thereby reducing the cumulative stress burden linked to ED[42]. Postoperative management: The number of indwelling catheters and extubation time serve as indicators of exogenous stimuli and recovery trajectory, both of which influence the risk of ED. During postoperative care, close monitoring of the awakening process and proactive management of catheters are essential. Minimizing unnecessary catheter use and avoiding delayed extubation can reduce psychological stress and physical discomfort[65]. Targeted nursing interventions - including regular assessment of mental status, optimized analgesia, and psychological support - effectively facilitate recovery and mitigate the incidence of postoperative delirium.
In summary, for patients identified as high-risk by the model, care should be prioritized with experienced nursing staff, and perioperative management should be tailored accordingly. This includes optimized analgesia, close monitoring for neurobehavioral changes, psychological intervention, and supportive therapies. Beyond direct patient care, the model also serves as an effective training and risk-management tool, enhancing nurses’ competency in early ED recognition and promoting precision nursing. By consolidating multiple risk factors into a simple, visual, and actionable tool, this nomogram provides nurses with a practical, evidence-based decision-support instrument. Its future integration into routine clinical workflows holds significant potential to improve the early detection and proactive management of postoperative ED, thereby enhancing patient outcomes and elevating the overall quality of perioperative care.
Limitations and future work
Despite its contributions, this study has several limitations. First, to facilitate data collection, the analysis was restricted to routinely available clinical variables, excluding potentially influential factors such as preadmission individual characteristics and familial or social support. Future research should incorporate variables across multiple dimensions of health determinants to refine the predictive features and improve model accuracy and positive predictive value. Second, to ensure cohort homogeneity and enhance model generalizability, patients with dementia or psychiatric disorders were excluded from this study due to their potential to significantly confound the research outcomes. Similarly, individuals undergoing cardiac or neurosurgical procedures were also excluded, as these patient populations represent distinct clinical subgroups with inherently elevated delirium risk. Consequently, the current model may not be directly generalizable to these excluded cohorts. Third, data on alcohol use history were not systematically collected. Given the ongoing debate regarding the association between alcohol use and ED, future studies should aim to clarify this relationship. Finally, as a single-center study, the findings may be subject to selection bias and limited external validity. To address these constraints, future multicenter, large-scale prospective trials are warranted to systematically validate the clinical utility, cost-effectiveness, and real-world impact of this model on clinical decision-making, timing of preventive interventions, and patient-centered outcomes. Such efforts will be critical to advancing this tool into a broadly applicable, standardized decision-support system for perioperative care.
CONCLUSION
The results of this study indicate that the incidence of ED in older adult patients is 17.16%. The independent influencing factors for the occurrence of ED in older adult patients include preoperative MMSE score, preoperative albumin level, surgical duration, surgical risk score, number of indwelling catheters, and extubation time. It is recommended that healthcare professionals assess the cognitive function and nutritional status of older adult patients preoperatively and provide timely training and nutritional interventions. During surgery, priority should be given to selecting low-risk, minimally invasive surgical methods to avoid unnecessary delays. Postoperatively, attention should be paid to the patient’s awakening process, with efforts made to reduce the number of catheters and to extubate promptly, thereby decreasing discomfort and psychological stress responses, which can enhance postoperative recovery quality and reduce the incidence of complications. The risk prediction model for postoperative ED in older adult patients, based on these six influencing factors, can effectively identify high-risk patients. The web-based dynamic nomogram is user-friendly and contributes to improving the scientific and effective management of older adult patients during the perioperative period.
ACKNOWLEDGEMENTS
We thank all participants, researchers, and collaborators who contributed to this study.
Anesthesia and Perioperative Management of the Elderly; Chinese Medical Association; National Clinical Research Center for Geriatric Diseases, National Geriatric Anesthesia Alliance. [Guidelines on perioperative anesthesia management for elderly patients in China (2020 version)].Zhonghua Yixue Zazhi. 2020;33:2565-2578.
[PubMed] [DOI] [Full Text]
Luo AL, Zhang J. [Interpretation of the 2017 edition of the European Society of Anesthesiology's "Evidence based and Expert Consensus based Guidelines for Postoperative Delirium"].Linchuang Waike Zazhi. 2018;26:29-33.
[PubMed] [DOI] [Full Text]
Mo HF, Chen YP, Han H, Zhang YP, Liu YJ, Zhang M, Lu XD, Hua YT, Zhu YJ, Cai TT, Xia YH, Shen JT. [Methods and procedures of clinical predictive model].Zhongguo Xunzheng Yixue Zazhi. 2024;228-236.
[PubMed] [DOI]
Nie B, Zheng HZ, Xu YP, Zhou Y. [Analysis of the Incidence Rate, Analysis of the Incidence Rate,Risk Factors in Elderly Emergence Delirium Patients after Operation Under General Anesthesia].Yixue Lilun Yu Shijian. 2020;33:4065-4068, 4086.
[PubMed] [DOI] [Full Text]
Morandi A, Inzitari M, Udina C, Gual N, Mota M, Tassistro E, Andreano A, Cherubini A, Gentile S, Mossello E, Marengoni A, Olivé A, Riba F, Ruiz D, de Jaime E, Bellelli G; Italian Study Group of Delirium. Visual and Hearing Impairment Are Associated With Delirium in Hospitalized Patients: Results of a Multisite Prevalence Study.J Am Med Dir Assoc. 2021;22:1162-1167.e3.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 22][Cited by in RCA: 27][Article Influence: 5.4][Reference Citation Analysis (0)]
Yuan T, Deng MG, Zhang AN, Peng M. [Association between post-anesthesia care unit delirium and postoperative delirium after noncardiac surgeries in adults: A meta-analysis based on observational studies].Wuhan Daxue Xuebao (Yixueban). 2023;44:1273-1279.
[PubMed] [DOI] [Full Text]
Au E, Thangathurai G, Saripella A, Yan E, Englesakis M, Nagappa M, Chung F. Postoperative Outcomes in Elderly Patients Undergoing Cardiac Surgery With Preoperative Cognitive Impairment: A Systematic Review and Meta-Analysis.Anesth Analg. 2023;136:1016-1028.
[RCA] [PubMed] [DOI] [Full Text][Cited by in RCA: 17][Reference Citation Analysis (0)]
Wang TP, Cai YS, Guo H, Sun Y, Cao JJ, Yang Y. [Establishment and validation of a risk prediction model for postoperative delirium in elderly patients after individualized hip fracture surgery].Lujun Junyi Daxue Xuebao. 2022;44:563-570.
[PubMed] [DOI] [Full Text]
Tang LY, Chen SL, Chen Q, Zheng SL. [Effects of cognitive reserve on waking delirium in general surgical patients under general anesthesia].Xiandai Yixue Weisheng. 2023;39:6-11.
[PubMed] [DOI] [Full Text]
Xie BL, Nie LZ, Zhong B, Xiong J, Nie M, Ai QX, Yang D. Longitudinal Trends in the Incidence of Hyperactive Delirium and Its Causes of Change After Surgery for Degenerative Lumbar Disease: A Population-Based Study of 7250 Surgical Patients Over 11 Years.Orthop Surg. 2025;17:714-723.
[RCA] [PubMed] [DOI] [Full Text][Cited by in RCA: 4][Reference Citation Analysis (0)]
Luo Y, Su W, Ding WY, Liu LX, Li W. [Study on the removal of urinary catheter at different times after interventional embolization in patients with unruptured intracranial aneurysm].Zhonghua Huli Zazhi. 2022;57:2864-2869.
[PubMed] [DOI] [Full Text]
Yang HW, Yi FQ, Zeng YC, Zhong CY, Xiong ZY, Zhu ZY. [Super Early Removal of Urinary Catheter in Patients after General Anesthesia].Huli Xuebao. 2021;28:66-69.
[PubMed] [DOI] [Full Text]
Wang YB, Lu XB, Zhang QJ, Lin SH, Chen YL. [Establishment of Nomogram predictive model of delirium during postoper-ative recovery period of thyroid cancer patients undergoing radical thy-roidectomy under general anesthesia].Zhengzhou Daxue Xuebao. 2020;55:386-390.
[PubMed] [DOI] [Full Text]
Deng XE, Zhang LF, Zhou CF, Xie XL, Huang Y, Li Y, Huang L. [Effect of Early Tracheal Extubation on the Quality of Resuscitation in Patients with Lung Cancer During Recovery from General Anesthesia].Sichuan Yixue. 2021;42:433-437.
[PubMed] [DOI] [Full Text]
van den Boogaard M, Pickkers P, Slooter AJ, Kuiper MA, Spronk PE, van der Voort PH, van der Hoeven JG, Donders R, van Achterberg T, Schoonhoven L. Development and validation of PRE-DELIRIC (PREdiction of DELIRium in ICu patients) delirium prediction model for intensive care patients: observational multicentre study.BMJ. 2012;344:e420.
[RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)][Cited by in Crossref: 242][Cited by in RCA: 286][Article Influence: 20.4][Reference Citation Analysis (0)]
Cheng ZZ, Ai XW, Jiang N, He CZ. [Research progress on postoperative pain care in cardiothoracic surgery].Shiyong Linchuang Yixue. 2022;23:135-138.
[PubMed] [DOI] [Full Text]
Wassenaar A, van den Boogaard M, van Achterberg T, Slooter AJ, Kuiper MA, Hoogendoorn ME, Simons KS, Maseda E, Pinto N, Jones C, Luetz A, Schandl A, Verbrugghe W, Aitken LM, van Haren FM, Donders AR, Schoonhoven L, Pickkers P. Multinational development and validation of an early prediction model for delirium in ICU patients.Intensive Care Med. 2015;41:1048-1056.
[RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)][Cited by in Crossref: 97][Cited by in RCA: 158][Article Influence: 14.4][Reference Citation Analysis (0)]
Nagata C, Hata M, Miyazaki Y, Masuda H, Wada T, Kimura T, Fujii M, Sakurai Y, Matsubara Y, Yoshida K, Miyagawa S, Ikeda M, Ueno T. Development of postoperative delirium prediction models in patients undergoing cardiovascular surgery using machine learning algorithms.Sci Rep. 2023;13:21090.
[RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)][Cited by in RCA: 17][Reference Citation Analysis (0)]
P-Reviewer: Fedotov IA, MD, PhD, Associate Professor, Russia; Noufi P, MD, Assistant Professor, United States S-Editor: Hu XY L-Editor: A P-Editor: Zhao S
