Early identification and prediction of postoperative pulmonary complications (PPCs) are vital for patient management in lung cancer (LC) surgery. However, existing predictive models often lack comprehensive validation and interpretability. This study aimed to develop and validate an explainable machine learning (ML) model to predict PPCs in patients with LC undergoing surgery. A risk factor variable pool was determined by meta-analysis and Delphi surveys. Patients undergoing LC surgery who were admitted to the Thoracic Surgery Department at the Affiliated Hospital of Zunyi Medical University from 1st January 2022 to 31st October 2023 (retrospective) and from 1st November 2023 to 31st July 2024 (prospective) were used for model development and prospective validation, respectively. The retrospective cohort was randomly split into a training set and an internal validation set at an 8:2 ratio. Feature selection involved univariate analysis, collinearity analysis, nine ML algorithms, and expert consensus. Twelve independent ML models and 26 stacking ensemble models were developed. Predictive performance was evaluated using the area under the receiver-operating-characteristic curve (AUROC), accuracy, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and F1 score. Prospective validation was analysed using AUC, Hosmer–Lemeshow test, calibration curves, and decision curve analysis (DCA). The Shapley Additive Explanation (SHAP) method was utilised to interpret the predictive model. A total of 883 patients were included in the retrospective cohort with an incidence of PPCs of 35.4% (313/883), and a total of 308 patients were included in the prospective cohort with PPCs of 29.5% (91/308). Nine key characteristics were selected for model development: age, duration of surgery, Charson comorbidity index (CCI), tumour stage, measured carbon monoxide diffusion (DLCO, mmol/min/kPa), intra-operative infusion volume (IFIV, mL), red blood cell volume distribution width–coefficient of variation (RDW-CV, %), body mass index (BMI), and number of years of smoking. Amongst the independent models, the Gradient Boosting Decision Tree (GBDT) showed best performance, achieving an AUROC of 0.829 (95% CI: 0.774–0.885). The stacking ensemble combining Support Vector Machine (SVM) and Decision Tree (DT) showed the highest overall performance, with an AUROC of 0.860 (95% CI: 0.809–0.911), and DCA showed higher clinical utility compared to other models. In the prospective validation, the AUROC was 0.790 (95% CI: 0.744–0.835). The stacking ensemble model combining SVM and DT demonstrated robust predictive performance and favourable clinical utility for prediction PPCs in patients undergoing LC surgery. However, the model has not been applied in clinical practice and requires future validation in large, multi-centre cohorts. Further work should aim to identify high-risk patients early through clinical data analysis, enabling timely interventions and more efficient allocation of limited healthcare resources. The Science and Technology Foundation of Guizhou Provincial Health Commission; the Key Talent Team of Guizhou Provincial Science and Technology Innovation; and Guizhou Science and Technology Cooperation Basic Research Project.

Postoperative pulmonary complications (PPCs) are associated with high mortality rates after lung cancer surgery. Although some studies have discussed the different risk factors for PPCs, the relationship between these factors and their impact on PPCs remains unclear. Hence, this study aimed to systematically summarize the incidence and determine the risk factors for PPCs.

Background: Pulmonary rehabilitation (PR) has been shown to effectively support postsurgical recovery in patients with lung cancer (LC) at various stages. While digital PR programs offer a potential solution to traditional challenges, such as time and space constraints, their efficacy and feasibility for patients undergoing LC surgery remain unclear. Objective: This systematic review aims to assess the feasibility and effectiveness of digital PR programs for individuals undergoing LC surgery. Methods: A systematic review was conducted, retrieving data from 6 English and 4 Chinese databases from their inception to January 1, 2024. References in related studies were also manually reviewed. The primary outcomes assessed were physical capacity, lung function, and the incidence of postoperative pulmonary complications (PPCs). The secondary outcomes were compliance, hospital stay, chest tube duration, anxiety, depression, and quality of life. Where applicable, recruitment and withdrawal rates were also evaluated. Meta-analysis and descriptive analysis were used to assess the outcomes. Results: A total of 5 randomized controlled trials and 6 quasi-experimental studies (n=1063) were included, with 4 studies being included in the meta-analyses. Our meta-analyses revealed that digital PR reduced the decline in 6-minute walk distance (6-MWD) by an average of 15 m compared with routine PR programs from admission to discharge, demonstrating a clinically significant improvement in physical capacity (mean difference –15.00, 95% CI –25.65 to –4.34, P=.006). Additionally, digital PR was associated with a reduction (26/58, 45%) in the likelihood of PPCs (risk ratio 0.45, 95% CI 0.30-0.66, P<.001) and a reduction of 1.53 days in chest tube duration (mean difference –1.53, 95% CI –2.95 to –0.12, P=.03), without a statistically significant effect on postoperative hospital stay (mean difference –1.42, 95% CI –3.45 to 0.62, P=.17). Descriptive analyses suggested that digital PR has the potential to improve knowledge, lung function, quality of life, and self-efficacy, while reducing depression and anxiety. Notably, digital PR was found to be a safe, feasible, and acceptable supplementary intervention. Despite challenges with low recruitment, digital PR enhanced exercise compliance, increased patient satisfaction, and lowered dropout rates. Conclusions: This systematic review is the first comprehensive analysis to suggest that digital PR is a safe, feasible, acceptable, and effective intervention for promoting recovery in patients with LC after surgery. Digital PR has the potential to be a valuable supplement, expanding access to traditional PR programs. Future research should prioritize the development of interactive and inclusive digital solutions tailored to diverse age groups and educational backgrounds. Rigorous studies, including large-scale, high-quality randomized controlled trials with detailed protocols and robust methodologies, are needed to assess the short-, medium-, and long-term efficacy of digital PR, ensuring reproducibility in future research. Trial Registration: PROSPERO CRD42023430271; https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=430271

Intensive care unit-acquired weakness (ICU-AW) is a common and debilitating complication in critically ill patients, significantly affecting both short- and long-term outcomes. The existing ICU-AW diagnostic methods are not widely accepted and have a narrow application window. Biomarkers offer potential for diagnosing, predicting, and prognosticating ICU-AW, but a comprehensive synthesis of the available evidence is still lacking. We conducted a systematic search across PubMed, Cochrane Library, Embase, Web of Science, CNKI, Wanfang Database, China Science and Technology Journal Database (VIP Database), and China Biomedical Literature Database (SinoMed Database) from inception to January 23, 2025. Study quality was assessed using the revised Newcastle-Ottawa scale and the Quality Assessment of Diagnostic Accuracy Studies-2 tool. Data extraction included basic characteristics of the included studies, name of biomarkers, objective, specimen types, sampling time, type of biomarker, ICU-AW diagnostic criteria, and outcomes. Out of 5,769 publications screened, 11 studies of moderate to high quality (scores ≥ 6) involving 1,176 critically ill patients were included. Ten biomarkers were identified and categorized into five mechanisms: muscle injury (myoglobin, N-titin, urinary titin), metabolic pathway (glucose transporter protein type-4), neurological injury (neurofilament light/heavy chain), stress response (growth differentiation factor-15), and inflammatory process (monocyte chemoattractant protein-1, NETs marker cfDNA, and miR-181a). Six biomarkers demonstrated strong predictive and diagnostic accuracy with AUC values exceeding 0.80. Notably, growth differentiation factor-15 exhibited excellent clinical utility across diagnostic, predictive, and prognostic applications (AUC ≥ 0.85). The remaining four biomarkers showed moderate performance, with AUC values ranging from 0.60 to 0.80. While ten biomarkers exhibit potential for ICU-AW assessment, their clinical utility remains inconsistent. This highlights the need for large-scale, prospective validation studies and the incorporation of advanced technologies to refine existing biomarkers and identify novel candidates for ICU-AW prediction, diagnosis and management. Registered 1 August 2024. PROSPERO ID: CRD42024574437.

Background: Animal models focusing on neuromuscular outcomes are crucial for understanding the mechanisms of intensive care unit-acquired weakness (ICU-AW) and exploring potential innovative prevention and treatment strategies.Aim: To analyse and evaluate preclinical ICU-AW models.Methods: We manually searched five English and four Chinese databases from 1 January 2002, to 1 February 2024, and reviewed related study references. Full-text publications describing animal models of muscle weakness and atrophy in critical illness were included. Detailed information about model types, animal species, sex, age, induction methods, outcome measures, drawbacks and strengths was extracted from each included study.Results: A total of 3,451 citations were initially retrieved, with 84 studies included in the final analysis. The most frequently studied animal model included rodents (86.9%), 64.3% of which were male animals. ICU-AW animal models were mostly induced by comprehensive intensive care unit (ICU) interventions (38.1%) and sepsis (51.2%). Most studies focused on limb muscles (66.7%), diaphragm muscles (21.4%) or both (9.5%). Reported outcomes primarily included muscular pathological changes (83.3%), electrophysiological examinations of muscles (57.1%) and animal grip strength (16.6%). However, details such as animal age, mortality data, experimental design, randomisation, blinding, sample size and interventions for the experimental group and/or control group were inadequately reported.Conclusion: Many preclinical models are used to study ICU-AW, but the reporting of methodological details is often incomplete. Although current ICU animal models can mimic the characteristics of human ICU-AW, there is no standard model. Future preclinical studies should develop a standard ICU-AW animal model to enhance reproducibility and improve scientific rigor in exploring the mechanisms and potential treatment of ICU-AW.