Lei Du

Prof. Lei Du

orcid

Research Focus & Metrics

Academic Career Length

19Years (2008-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
23 April 2025 in Journal of Medical Internet Research
Verified
Journal Article
Open access
ZZZhuo Zheng
Zhuo Zheng
JLJiawei Luo
Jiawei Luo
YZYingchao Zhu
Yingchao Zhu
LDLei Du
Lei Du
LLLan Lan
Lan Lan
SHShixin Huang
Shixin Huang
Background: Timely and accurate prediction of short-term mortality is critical in intensive care units (ICUs), where patients’ conditions change rapidly. Traditional scoring systems, such as the Simplified Acute Physiology Score and Acute Physiology and Chronic Health Evaluation, rely on static variables collected within the first 24 hours of admission and do not account for continuously evolving clinical states. These systems lack real-time adaptability, interpretability, and generalizability. With the increasing availability of high-frequency electronic medical record (EMR) data, machine learning (ML) approaches have emerged as powerful tools to model complex temporal patterns and support dynamic clinical decision-making. However, existing models are often limited by their inability to handle irregular sampling and missing values, and many lack rigorous external validation across institutions. Objective: We aimed to develop a real-time, interpretable risk prediction model that continuously assesses ICU patient mortality using irregular, longitudinal EMR data, with improved performance and generalizability over traditional static scoring systems. Methods: A time-aware bidirectional attention-based long short-term memory (TBAL) model was developed using EMR data from the MIMIC-IV (Medical Information Mart for Intensive Care) and eICU Collaborative Research Database (eICU-CRD) databases, comprising 176,344 ICU stays. The model incorporated dynamic variables, including vital signs, laboratory results, and medication data, updated hourly, to perform static and continuous mortality risk assessments. External cross-validation and subgroup sensitivity analyses were conducted to evaluate robustness and fairness. Model performance was assessed using the area under the receiver operating characteristic curve (AUROC), area under the precision-recall curve (AUPRC), accuracy, and F1-score. Interpretability was enhanced using integrated gradients to identify key predictors. Results: For the static 12-hour to 1-day mortality prediction task, the TBAL model achieved AUROCs of 95.9 (95% CI 94.2-97.5) and 93.3 (95% CI 91.5-95.3) and AUPRCs of 48.5 and 21.6 in MIMIC-IV and eICU-CRD, respectively. Accuracy and F1-scores reached 94.1 and 46.7 in MIMIC-IV and 92.2 and 28.1 in eICU-CRD. In dynamic prediction tasks, AUROCs reached 93.6 (95% CI 93.2-93.9) and 91.9 (95% CI 91.6-92.1), with AUPRCs of 41.3 and 50, respectively. The model maintained high recall for positive cases (82.6% and 79.1% in MIMIC-IV and eICU-CRD). Cross-database validation yielded AUROCs of 81.3 and 76.1, confirming generalizability. Subgroup analysis showed stable performance across age, sex, and severity strata, with top predictors including lactate, vasopressor use, and Glasgow Coma Scale score. Conclusions: The TBAL model offers a robust, interpretable, and generalizable solution for dynamic real-time mortality risk prediction in ICU patients. Its ability to adapt to irregular temporal patterns and to provide hourly updated predictions positions it as a promising decision-support tool. Future work should validate its utility in prospective clinical trials and investigate its integration into real-world ICU workflows to enhance patient outcomes.

13 September 2022 in Frontiers in Cardiovascular Medicine
Verified
Journal Article
Open access
JTJia Tan
Jia Tan
SBSiwei Bi
Siwei Bi
JLJingyi Li
Jingyi Li
JGJun Gu
Jun Gu
YWYishun Wang
Yishun Wang
LDLei Du
Lei Du
Objective: To compare the outcomes of four types of cardioplegia during cardiac surgery: del Nido (DN), blood cardioplegia (BC), histidine-tryptophan-ketoglutarate (HTK) and St. Thomas.Methods: Randomized controlled trials (RCTs) and observational cohort studies from 2005 to 2021 were identified in PubMed, Embase, and Cochrane databases. Data were extracted for the primary endpoint of perioperative mortality as well as the following secondary endpoints: atrial fibrillation, renal failure, stroke, use of an intra-aortic balloon pump, re-exploration, intensive care unit stay and hospital stay. A network meta-analysis comparing all four types of cardioplegia was performed, as well as direct meta-analysis comparing pairs of cardioplegia types.Results: Data were extracted from 18 RCTs and 49 observational cohort studies involving 18,191 adult patients (55 studies) and 1,634 children (12 studies). Among adult patients, risk of mortality was significantly higher for HTK (1.89, 95% CI 1.10, 3.52) and BC (RR 1.73, 95% CI 1.22, 2.79) than for DN. Risk of atrial fibrillation was significantly higher for BC (RR 1.41, 95% CI 1.09, 1.86) and DN (RR 1.51, 95% CI 1.15, 2.03) than for HTK. Among pediatric patients, no significant differences in endpoints were observed among the four types of cardioplegia.Conclusions: This network meta-analysis suggests that among adult patients undergoing cardiac surgery, DN may be associated with lower perioperative mortality than HTK or BC, while risk of atrial fibrillation may be lower with HTK than with BC or DN.

3 February 2017 in Scientific Reports
Verified
Journal Article
Open access
PDF file
SLShuhua Luo
Shuhua Luo
YWYabo Wang
Yabo Wang
QAQi An
Qi An
HCHao Chen
Hao Chen
JZJunfei Zhao
Junfei Zhao
LDLei Du
Lei Du
Systemic inflammatory responses can severely injure lungs, prompting efforts to explore how to attenuate such injury. Here we explored whether platelets can help attenuate lung injury in mice resulting from extracorporeal circulation (ECC)-induced systemic inflammatory responses. Mice were subjected to ECC for 30 min, then treated with phosphate-buffered saline, platelets, the GPIIb/IIIa inhibitor Tirofiban, or the combination of platelets and Tirofiban. Blood and lung tissues were harvested 60 min later, and lung injury and inflammatory status were assessed. As expected, ECC caused systemic inflammation and pulmonary dysfunction, and platelet transfusion resulted in significantly milder lung injury and higher lung function. It also led to greater numbers of circulating platelet-leukocyte aggregates and greater platelet accumulation in the lung. Platelet transfusion was associated with higher production of transforming growth factor-β and as well as lower levels of tumour necrosis factor-α and neutrophil elastase in plasma and lung. None of these platelet effects was observed in the presence of Tirofiban. Our results suggest that, at least under certain conditions, platelets can protect lung from injury induced by systemic inflammatory responses.

1 February 2019 in Medicine
Verified
Journal Article
Open access
DHDou Huang
Dou Huang
CCChangwei Chen
Changwei Chen
YMYue Ming
Yue Ming
JLJing Liu
Jing Liu
LZLi Zhou
Li Zhou
LDLei Du
Lei Du
Cardiac surgery under cardiopulmonary bypass (CPB) accounts for most consumption of red blood cells (RBCs). Identifying risk factors for massive red blood cell transfusion (MRT) in cardiac surgery may help to reduce this consumption. We retrospectively analyzed 8238 patients who underwent valve surgery and/or coronary artery bypass grafting (CABG) under CPB at 2 major heart centers in China. Uni- and multivariate logistic regression was carried out to assess whether risk factors for MRT (defined as receiving at least 4 units RBCs) varied with type of cardiac surgery. A total of 1691 patients (21%) received at least 4 units RBCs (6.77 ± 4.78 units per person). This MRT group consumed 70% of the total units of allogeneic RBCs in the study. MRT incidence was 2-fold higher among patients undergoing CABG with or without valve surgery than among patients undergoing valve surgery alone. Multivariate logistic analysis identified the following MRT risk factors common to valve surgery alone, CABG alone, and their combination: female sex, older age, renal dysfunction, lower body mass index, lower preoperative hemoglobin, and longer CPB. Several independent MRT risk factors were also identified specific to valve surgery: active endocarditis, nonatrial fibrillation, smaller left atrium diameter, abnormal international normalized ratio, and repeat surgery. Different types of cardiac surgery share several, but not all, MRT risk factors. This study may help guide the prediction and management of patients at higher MRT risk.

24 January 2022 in Journal of Clinical Anesthesia
Verified
Journal Article
Closed access
YWYi Shun Wang
Yi Shun Wang
DCDong Xu Chen
Dong Xu Chen
MYMin Yan
Min Yan
ZWZhong Wu
Zhong Wu
YGYing Qiang Guo
Ying Qiang Guo
LDLei Du
Lei Du

28 August 2020 in PLOS ONE
Verified
Journal Article
Open access
PDF file
TPTailai Peng
Tailai Peng
XLXinhao Liu
Xinhao Liu
HNHefeng Ni
Hefeng Ni
ZCZhe Cui
Zhe Cui
LDLei Du
Lei Du
In December 2019, an outbreak of COVID-19 epidemic occurred in Wuhan, China and infection spread rapidly around the world. To limit the rapid spread locally and nationwide, the Chinese government locked down Wuhan city on January 23 and began implementing nationwide intensive community screening on February 16. To assess the effectiveness of city lockdown and intensive community screening, we built a modified SIR model by introducing an α value into the classic SIR model. The α value represents the proportion of infected individuals who are not effectively isolated from susceptible individuals at a given time point. The accuracy of the modified SIR model was validated using data from Guangdong and Zhejiang provinces. The lockdown of Wuhan city substantially reduced the α value for the rest of China excluding Hubei province, while only slightly reducing the α value for the city itself. Intensive community screening rapidly reduced the α value for Wuhan. City lockdown was efficient in controlling the spread of the epidemic from Wuhan to the rest of the country. Nationwide intensive community screening was extremely effective in suppressing the spread locally within Wuhan city. These experiences should urgently be shared with other countries to help halt the progressing worldwide pandemic.