Predicting Mortality in Intensive Care Unit Patients with Allergic Bronchopulmonary Aspergillosis (ABPA) Using an Interpretable Machine Learning Model: A Retrospective Cohort Study

Predicting Mortality in Intensive Care Unit Patients with Allergic Bronchopulmonary Aspergillosis (ABPA) Using an Interpretable Machine Learning Model: A Retrospective Cohort Study | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Predicting Mortality in Intensive Care Unit Patients with Allergic Bronchopulmonary Aspergillosis (ABPA) Using an Interpretable Machine Learning Model: A Retrospective Cohort Study Jing Zhang, Juntao Tang, Jinjuan Li, Xinghua Liu, Ye Sun, peng wang This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7822617/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background Allergic bronchopulmonary aspergillosis (ABPA) is a hypersensitivity lung disease caused by Aspergillus infection, with severe cases often requiring admission to the intensive care unit (ICU). Early prediction of in-hospital mortality in ICU ABPA patients is crucial for optimizing clinical decision-making and resource allocation. Methods This retrospective study collected clinical data from ICU patients diagnosed with ABPA at Yuebei People's Hospital between January 2020 and July 2024. An in-hospital mortality prediction model was developed using an explainable XGBoost machine learning algorithm. SHapley Additive Explanations (SHAP) was employed to interpret key predictive factors, and internal validation was conducted to assess model performance. Results A total of 82 ICU ABPA patients were included, with mortality rates of 46.3% (26/57) in the training set and 48% (12/25) in the validation set. The XGBoost model demonstrated excellent predictive performance, achieving areas under the receiver operating characteristic (ROC) curve (AUC) of 0.995 (95% CI: 0.903–1.000) in the training set and 0.881 (95% CI: 0.846–0.909) in the validation set. SHAP analysis identified key predictors of mortality, including BMI, peak procalcitonin level, peak eosinophil count, age, asthma history, peak leukocyte count, and lowest platelet count. Conclusion The XGBoost model effectively predicts in-hospital mortality in ICU ABPA patients and provides interpretable results using SHAP analysis. Although the model performed well in internal validation, external validation is needed to enhance its generalizability. Future multicenter studies and integration of dynamic biomarkers are recommended to optimize predictive accuracy and support individualized clinical decision-making. Allergic bronchopulmonary aspergillosis (ABPA) XGBoost machine learning predictive model Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Introduction Allergic bronchopulmonary aspergillosis (ABPA) is a hypersensitivity lung disease caused by a complex immune reaction triggered by Aspergillus colonization in the airways [ 1 ] . ABPA typically occurs in patients with preexisting respiratory conditions such as asthma and cystic fibrosis (CF), and it is characterized by difficult-to-control asthma-like symptoms and recurrent pulmonary infiltrates, with or without bronchiectasis. With increasing awareness among clinicians, ABPA has also been identified as a secondary condition in patients with bronchiectasis, chronic obstructive pulmonary disease (COPD), and other pulmonary disorders [ 2 , 3 ] . The prevalence of ABPA in the general population remains unclear; however, it has been reported to range from 2% to 32% in asthma patients [ 4 ] and from 3% to 25% in CF patients [ 5 ] . The prognosis of ABPA largely depends on early diagnosis and timely treatment. The natural course of ABPA is variable, with recurrent exacerbations and remissions being its hallmark. Long-term recurrence can lead to irreversible lung damage, including bronchiectasis, pulmonary fibrosis, and even progression to chronic pulmonary heart disease or respiratory failure. Serum total IgE levels are commonly used as a biomarker for monitoring ABPA disease activity [ 6 ] . High-attenuation mucus (HAM) impaction and the severity of central bronchiectasis have been identified as independent risk factors for ABPA relapse [ 7 , 8 ] . Critically ill patients in the intensive care unit (ICU) require specialized care and multidisciplinary support [ 9 ] . Although the ICU plays a crucial role in sustaining patients’ lives, it also poses challenges such as workforce shortages, limited medical resources, and significant financial burdens [ 10 ] . Therefore, early identification of hospital mortality risk in ICU patients is essential, as it may facilitate appropriate treatment strategies and support clinical decision-making [ 11 ] . In recent years, artificial intelligence has been widely employed to identify early warning predictors for various diseases. Given the inherent strength of machine learning algorithms in capturing nonlinear relationships, an increasing number of researchers advocate for the use of machine learning-based predictive models to support appropriate patient treatment, rather than relying solely on traditional disease severity scoring systems such as the Sequential Organ Failure Assessment (SOFA), Acute Physiology and Chronic Health Evaluation II (APACHE II), or Simplified Acute Physiology Score II (SAPS II) [ 12 – 14 ] . Although numerous predictive models have demonstrated promising performance in research settings, their clinical application remains limited, and there is still a lack of evidence supporting interpretable risk prediction models that contribute to disease prognosis [ 15 – 18 ] . The use of an explainable machine learning model based on Extreme Gradient Boosting (XGBoost) to predict mortality in intensive care unit (ICU) patients with allergic bronchopulmonary aspergillosis (ABPA) and to perform internal validation represents an emerging research direction. By analyzing clinical data, the XGBoost model can effectively predict mortality in ABPA patients while providing interpretable predictions. Internal validation, achieved through methods such as cross-validation, ensures the stability and reliability of the model. This approach not only assists clinicians in better assessing patient prognosis but also provides valuable insights for developing personalized treatment strategies. The objective of this study is to utilize existing clinical data and patient characteristics from ICU admission to construct and interpret an XGBoost model for ABPA prognosis. The SHapley Additive Explanations (SHAP) method is employed to explain the model's predictions and explore prognostic factors associated with ABPA. Methods Data Source This retrospective study collected clinical data from patients diagnosed with allergic bronchopulmonary aspergillosis (ABPA) at Yuebei People's Hospital between January 2020 and July 2024. The diagnosis was confirmed based on metagenomic sequencing of sputum, which identified Aspergillus infection. The dataset comprehensively recorded demographic information, vital signs, diagnostic details, and treatment information for all patients. The inclusion criteria were based on the 2017 Expert Consensus on the Diagnosis and Treatment of Allergic Bronchopulmonary Aspergillosis established by the Asthma Group of the Chinese Thoracic Society. Patients were included if they met the following criteria: Associated diseases : Presence of asthma or other underlying respiratory conditions, such as bronchiectasis, chronic obstructive pulmonary disease (COPD), or pulmonary cystic fibrosis. Essential criteria : Aspergillus fumigatus -specific IgE > 0.35 kUA/L or a positive immediate skin test reaction to Aspergillus fumigatus . Elevated serum total IgE levels (> 1000 IU/mL). Additional criteria : Blood eosinophil count > 0.5 × 10⁹/L. Pulmonary imaging findings consistent with ABPA, including but not limited to pulmonary consolidation, nodules, toothpaste sign, finger-in-glove sign, transient migratory opacities, or persistent changes such as bronchiectasis and pleuroparenchymal fibrosis. Positive serum Aspergillus -specific IgG antibodies or precipitin test. A diagnosis of ABPA required meeting criterion 1 , criterion 2 , and at least two conditions from criterion 3 . Patients with serum total IgE levels < 1000 IU/mL could still be diagnosed with ABPA if all other criteria were met. Additionally, only patients aged 18 years or older were included in the study. The exclusion criteria were as follows: Presence of malignancy or severe organ dysfunction (e.g., cardiac, cerebral, or renal failure). Pregnant or lactating women. Incomplete medical records or insufficient diagnostic data. Ethical Considerations This study was approved by the Ethics Committee of Yuebei People's Hospital. As all protected health information was anonymized and the study was retrospective in nature, the requirement for informed consent from patients and their families was waived. Predictor Variables and Explainable Machine Learning Tools The outcome variable in this study was the probability of in-hospital mortality, defined based on the patient’s status at discharge.Model interpretability was enhanced using SHapley Additive Explanations (SHAP), which quantify the individual contribution of each feature to the prediction outcome [ 19 ] . SHAP values indicate the extent to which each predictor variable influences the target variable, either positively or negatively. Additionally, each individual observation in the dataset can be explained by a specific set of SHAP values. Statistical Analysis All statistical analyses and computations were performed using R software (version 3.8.0). Categorical variables were expressed as counts and percentages, and differences between groups were compared using the chi-square (χ²) test or Fisher’s exact test (when the expected frequency was < 10). Continuous variables were reported as medians with interquartile ranges (IQR) and were compared between groups using the Wilcoxon rank-sum test. The predictive model was developed using the Extreme Gradient Boosting (XGBoost) machine learning algorithm. Model performance was evaluated by calculating the area under the receiver operating characteristic (ROC) curve (AUC). Additionally, accuracy, sensitivity, positive predictive value (PPV), and negative predictive value (NPV) were computed. To assess the clinical utility of the model, decision curve analysis (DCA) was conducted, which quantifies net benefit across different probability thresholds for decision-making [ 20 ] . SHAP values were calculated using the R language implementation, enabling interpretation of individual feature contributions to the model predictions.”, which is specifically optimized for tree-based models such as XGBoost. This approach enables calculation of consistent, locally accurate feature attribution values for each patient’s mortality prediction. Results Patient Characteristics A total of 82 adult patients diagnosed with allergic bronchopulmonary aspergillosis (ABPA) in the ICU were included in the final cohort of this study. The patient screening process is illustrated in Fig. 1 . The dataset was split into a training set (70%) and a validation set (30%) using stratified random sampling based on the outcome variable (in-hospital mortality) to ensure balanced distribution of the event across the subsets.The dataset was randomly divided into two subsets: 70% (n = 57) of the data was used for model training, while 30% (n = 25) was used for model validation. A total of 24 potential predictive factors were identified and used for model development. Patients in the non-survivor group were older than those in the survivor group ( P = 0.04). The in-hospital mortality rate was 46.3% (26/57) in the training dataset and 48% (12/25) in the testing dataset. Table 1 presents a comparison of predictor variables between survivors and non-survivors during hospitalization. Model Development and Evaluation An XGBoost model was developed using the training dataset, and its receiver operating characteristic (ROC) curve for the test dataset is shown in Fig. 2 , with an area under the curve (AUC) of 0.995 (95% CI: 0.903–1.000). Decision curve analysis (DCA) was performed on the test dataset to compare the net benefit of the optimal model against alternative clinical decision-making strategies. Clinical net benefit was defined as the minimum probability of disease occurrence at which further intervention would be warranted [ 21 ] . The DCA curve quantifies net benefit across different probability thresholds, as shown in Fig. 3 . A calibration curve illustrating the discrepancy between predicted and actual mortality is presented in Fig. 4 . Due to the heterogeneity of the study population, machine learning-based treatment strategies outperformed the default strategies of treating all or treating none of the patients. For the validation dataset, the ROC curve is shown in Fig. 5 , with an AUC of 0.881 (95% CI: 0.846–0.909). Additionally, DCA (Fig. 6 ) and calibration curve analysis (Fig. 7 ) were conducted on the validation dataset, demonstrating excellent predictive performance of the model. Interpretation of the XGBoost Model Using SHAP Method The SHAP algorithm was employed to assess the importance of each predictor variable in the XGBoost model’s predictions. The variable importance plot ranks the most significant features in descending order (Fig. 8 ). Among all predictors, BMI had the strongest predictive value, followed by peak procalcitonin level, peak eosinophil count, age, asthma history, peak leukocyte count , and lowest platelet count . Additionally, to determine the direction of the relationship between predictors and the outcome, SHAP values were used to identify mortality risk factors. As shown in Fig. 9 , the horizontal position indicates whether a given feature's effect is associated with a higher or lower mortality prediction, while the color represents whether the variable value is high (red) or low (blue) for a specific observation. The results indicate that an increase in BMI positively influences mortality prediction , shifting it toward a higher probability of death, whereas an increase in age negatively influences mortality prediction , shifting it toward survival. SHAP Force Plot Figure 10 presents the individual force plot for a specific patient. The SHAP values illustrate the predictive features relevant to that patient and the contribution of each feature to the mortality prediction. The bold number represents the probability prediction value ( f(x) ), while the baseline value corresponds to the model’s prediction without any input data. f(x) represents the log-odds ratio for each observation. In the force plot, In the force plot, red features (on the left) represent variables that increase the model’s predicted risk of mortality for that individual patient, while blue features (on the right) represent variables that decrease the prediction.. The length of the arrows visually represents the magnitude of each feature’s influence on the prediction—the longer the arrow, the greater the impact. Discussion In this study, we developed an XGBoost machine learning model based on clinical data from the intensive care unit (ICU) to predict in-hospital mortality in patients with allergic bronchopulmonary aspergillosis (ABPA). The model’s performance was evaluated through internal validation. Our results demonstrated that the XGBoost model exhibited high predictive accuracy in distinguishing mortality risk among ABPA patients. Decision curve analysis (DCA) showed significant clinical net benefit in the probability threshold range of 10%-20%, suggesting that the model can provide valuable reference information for clinical intervention within this risk range. Using the SHAP method to interpret the model, we identified key clinical features closely associated with ABPA mortality, including BMI, peak procalcitonin level, peak eosinophil count, age, asthma history, peak leukocyte count , and lowest platelet count . These findings not only enhance the interpretability of the model but also provide new insights into the mechanisms underlying ABPA severity. Clinical Significance of Key Variables Through an in-depth analysis of the model’s feature contributions using the SHAP method, we identified seven key clinical predictors associated with mortality in ABPA patients. Among them, BMI was the most significant predictor of mortality. Other important variables, such as peak procalcitonin level, peak eosinophil count, age, asthma history, peak leukocyte count , and lowest platelet count , also played critical roles in patient prognosis [ 22 ] . BMI , as an indicator of metabolic status and systemic inflammation, may influence patient outcomes by exacerbating systemic inflammatory responses. Previous studies have also linked BMI to the severity of pulmonary diseases [ 23 – 26 ] . An elevated peak procalcitonin level is typically indicative of severe infections, and its abnormal levels in ABPA may signal a poor prognosis [ 27 ] . Additionally, an increase in peak eosinophil count is associated with more severe airway inflammation, which is closely related to the immunopathological mechanisms of ABPA [ 28 ] . Age is a key factor affecting immune function and may contribute to disease progression in ABPA. Studies have shown that older patients are more likely to experience adverse clinical outcomes [ 29 ] . Interestingly, in our cohort, SHAP analysis suggested a negative correlation between age and mortality, which may be due to selection bias where younger ICU patients had more severe disease presentations, warranting further validation in larger cohorts.A history of asthma is closely linked to the pathogenesis of ABPA, as persistent airway inflammation and structural changes can increase disease complexity. Furthermore, an elevated peak leukocyte count often indicates systemic inflammation or infection, which may be associated with a higher risk of mortality in severe cases [ 30 ] . A decrease in lowest platelet count may reflect endothelial damage or coagulation dysfunction, which is also a marker of poor prognosis in critically ill ABPA patients [ 31 ] . In this study, we developed a mortality risk prediction model for patients with allergic bronchopulmonary aspergillosis (ABPA) in the intensive care unit (ICU) setting, utilizing XGBoost combined with SHAP for model interpretability. This work helps to address a gap in the current literature by focusing on a rare disease (ABPA) within the unique context of critical care. Over the past five years, relevant studies have shown that while machine learning has been widely applied for outcome prediction among critically ill ICU patients (such as those with pneumonia or invasive fungal infections), research specifically targeting ABPA patients—particularly those in the ICU—remains extremely limited. For example, Wang et al. developed a machine learning model for identifying invasive pulmonary aspergillosis in non-neutropenic patients, which included nine ABPA cases, suggesting that machine learning can aid diagnosis and risk stratification in fungal-related diseases; however, its application was limited to specific subgroups and did not focus on mortality prediction in ICU patients with ABPA [ 32 ] . Similarly, Du et al. proposed a prognostic model combining lasso regression and machine learning for non-neutropenic pulmonary aspergillosis but did not explicitly include ABPA nor target ICU patients [ 33 ] . In contrast, Jeon et al. and Wang et al. have conducted machine learning-based mortality prediction studies in ICU patients with pneumonia, using methods such as random forests and XGBoost, demonstrating significant improvements over traditional scoring systems [ 34 , 35 ] . Additionally, Chia et al. explored the application of interpretable machine learning for mortality prediction in the ICU, emphasizing the importance of model transparency for clinical acceptance, which aligns closely with our use of SHAP in this study [ 36 ] . Taken together, compared to existing literature, our study expands and contributes to the field in several key aspects: Unique study population : This is the first study to develop a mortality risk prediction model specifically for ICU patients with ABPA, helping to fill an existing research gap. Methodological innovation : By applying XGBoost in combination with SHAP, we aim to enhance predictive performance while ensuring model interpretability, providing a practical tool for clinical application. Value for small-sample, rare disease modeling : Our approach offers experience and insight into developing predictive models for rare diseases under low-sample conditions, which may help advance research in similar clinical areas. We hope these contributions will provide a useful reference for future studies in the field and support the development of effective, interpretable machine learning applications in critical care for rare disease populations. In summary, these key clinical features provide a solid basis for model interpretation and offer valuable reference points for risk assessment in clinical interventions. The ability of the XGBoost model to integrate these variables through non-linear relationships highlights the advantages of machine learning in capturing complex clinical interactions. It is important to note that the SHAP force plot illustrates the contribution of each feature to an individual prediction and does not necessarily reflect general clinical associations. Model Limitations and Challenges in Clinical Translation Although the model demonstrated robustness through internal validation, several limitations should be considered. First, the mortality rate among ABPA patients in the final cohort was relatively low (46.34%). At a threshold corresponding to 80% specificity, the model’s positive predictive value was 0.20, likely reflecting sample heterogeneity and moderate event prevalence. This result may be influenced by sample heterogeneity within the single-center dataset and the relative rarity of severe ABPA cases. Additionally, the lack of an external validation cohort may limit the generalizability of the model, particularly in patients with immunosuppression or other pulmonary comorbidities. Currently, this model is better suited for risk stratification rather than direct clinical decision-making. Future prospective studies should evaluate the actual benefit and cost-effectiveness of early interventions (e.g., intensified antifungal therapy or immunomodulation) for high-risk patients. Conclusion This study demonstrated that the XGBoost model can effectively predict mortality risk in ABPA patients and, through interpretability analysis, reveal key pathophysiological mechanisms. However, clinical translation of the model requires addressing limitations such as low positive predictive value and the absence of external validation . Future research should incorporate multicenter cohorts and dynamic biomarkers (e.g., serial IgE monitoring or radiomic features of pulmonary imaging) to optimize predictive accuracy and support personalized treatment strategies . Declarations Acknowledgements Not applicable. Authors’ contributions Peng Wang,Ye Sun,and Jing Zhang designed the work. Juntao Tang, Jinjuan Li and Xinghua Liu record and summarized the patient of features. Peng Wang and Jing Zhang analyzed datasets. Jing Zhang wrote this paper. All authors read and approved the fnal manuscript. Funding Not applicable. Availability of data and materials The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Ethics approval and consent to participate This study was reviewed and approved by the Ethics Committee of Yuebei People's Hospital (Ethics Approval Number: YBSKY-2025-029-001). The committee granted a waiver of informed consent for this research in accordance with national regulations and institutional ethical guidelines. The waiver was justified based on the retrospective and anonymized nature of the data analyzed, which posed no foreseeable risks to participants' privacy or rights. All personal identifiers were removed from the dataset prior to analysis to ensure confidentiality. The study strictly adhered to the principles outlined in the Declaration of Helsinki and relevant Chinese ethical standards for biomedical research involving human subjects. Clinical trial number Not applicable. Consent for publication Not applicable. Competing interests The authors declare that they have no competing interests. References Agarwal R, Sehgal IS, Dhooria S, Muthu V, Prasad KT, Bal A, Aggarwal AN, Chakrabarti A. Allergic bronchopulmonary aspergillosis. Indian J Med Res. 2020;151(6):529–49. 10.4103/ijmr.IJMR_1187_19 . PMID: 32719226; PMCID: PMC7602921. Tiew PY, Lim AYH, Keir HR, Dicker AJ, Mac Aogáin M, Pang SL, Low TB, Hassan TM, Poh ME, Xu H, Ong TH, Koh MS, Abisheganaden JA, Tee A, Chew FT, Chalmers JD, Chotirmall SH. High Frequency of Allergic Bronchopulmonary Aspergillosis in Bronchiectasis-COPD Overlap. Chest. 2022;161(1):40–53. Epub 2021 Aug 5. PMID: 34364870. Zhang P, Ma Y, Chen X, Ma Y, Yang L, Zhang M, Gao Z. The Difference in All-Cause Mortality Between Allergic Bronchopulmonary Aspergillosis with and without Chronic Obstructive Pulmonary Disease. J Asthma Allergy. 2022;15:1861–75. PMID: 36601290; PMCID: PMC9807121. Agarwal R, Aggarwal AN, Gupta D, Jindal SK. Aspergillus hypersensitivity and allergic bronchopulmonary aspergillosis in patients with bronchial asthma: systematic review and meta-analysis. Int J Tuberc Lung Dis. 2009;13(8):936–44. PMID: 19723372. Maturu VN, Agarwal R. Prevalence of Aspergillus sensitization and allergic bronchopulmonary aspergillosis in cystic fibrosis: systematic review and meta-analysis. Clin Exp Allergy. 2015;45(12):1765-78. 10.1111/cea.12595 . PMID: 26177981. Agarwal R, Chakrabarti A, Shah A, Gupta D, Meis JF, Guleria R, Moss R, Denning DW. ABPA complicating asthma ISHAM working group. Allergic bronchopulmonary aspergillosis: review of literature and proposal of new diagnostic and classification criteria. Clin Exp Allergy. 2013;43(8):850 – 73. 10.1111/cea.12141 . PMID: 23889240. Agarwal R, Khan A, Gupta D, Aggarwal AN, Saxena AK, Chakrabarti A. An alternate method of classifying allergic bronchopulmonary aspergillosis based on high-attenuation mucus. PLoS ONE. 2010;5(12):e15346. 10.1371/journal.pone.0015346 . PMID: 21179536; PMCID: PMC3002283. Agarwal R, Gupta D, Aggarwal AN, Saxena AK, Chakrabarti A, Jindal SK. Clinical significance of hyperattenuating mucoid impaction in allergic bronchopulmonary aspergillosis: an analysis of 155 patients. Chest. 2007;132(4):1183-90. 10.1378/chest.07-0808 . Epub 2007 Jul 23. PMID: 17646221. Monteiro F, Meloni F, Baranauskas JA, Macedo AA. Prediction of mortality in intensive care units: a multivariate feature selection. J Biomed Inf. 2020;107:103456. 10.1016/j.jbi.2020.103456 . https://linkinghub.elsevier.com/retrieve/pii/S1532-0464(20)30084-8 . .S1532-0464(20)30084-8 [PubMed] [CrossRef] [Google Scholar]. Kramer AA, Dasta JF, Kane-Gill SL. The impact of mortality on total costs within the ICU. Crit Care Med. 2017;45(9):1457–63. [PubMed] [CrossRef] [Google Scholar]. Khurrum M, Asmar S, Joseph B. Telemedicine in the ICU: innovation in the critical care process. J Intensive Care Med. 2021;36(12):1377–84. [PubMed] [CrossRef] [Google Scholar]. Kang MW, Kim J, Kim DK, Oh K, Joo KW, Kim YS, Han SS. Machine learning algorithm to predict mortality in patients undergoing continuous renal replacement therapy. Crit Care. 2020;24(1):42. 10.1186/s13054-020-2752-7 . https://ccforum.biomedcentral.com/articles/10.1186/s13054-020-2752-7 . [PMC free article] [PubMed] [CrossRef] [Google Scholar] p. Liu C, Liu X, Mao Z, Hu P, Li X, Hu J, Hong Q, Geng X, Chi K, Zhou F, Cai G, Chen X, Sun X. Interpretable machine learning model for early prediction of mortality in ICU patients with rhabdomyolysis. Med Sci Sports Exerc. 2021;53(9):1826–34. 10.1249/MSS.0000000000002674.00005768-202109000-00004 . [PubMed] [CrossRef] [Google Scholar]. Pirracchio R, Petersen ML, Carone M, Rigon MR, van Chevret S. Mortality prediction in intensive care units with the Super ICU Learner Algorithm (SICULA): a population-based study. Lancet Respir Med. 2015;3(1):42–52. http://europepmc.org/abstract/MED/25466337 . .S2213-2600(14)70239-5 [PMC free article] [PubMed] [CrossRef] [Google Scholar]. doi: 10.1016/S2213-2600(14)70239-5. Pan P, Li Y, Xiao Y, Han B, Su L, Su M, Li Y, Zhang S, Jiang D, Chen X, Zhou F, Ma L, Bao P, Xie L. Prognostic Assessment of COVID-19 in the intensive care unit by machine learning methods: model development and validation. J Med Internet Res. 2020;22(11):e23128. https://www.jmir.org/2020/11/e23128/ v22i11e . 23128 [PMC free article] [PubMed] [CrossRef] [Google Scholar]. doi: 10.2196/23128. Stenwig E, Salvi G, Rossi PS, Skjærvold NK. Comparative analysis of explainable machine learning prediction models for hospital mortality. BMC Med Res Methodol. 2022;22(1):53. 10.1186 . https://bmcmedresmethodol.biomedcentral.com/articles/10.1186/s12874-022-01540-w . [PMC free article] [PubMed] [CrossRef] [Google Scholar]. /s12874-022-01540-w. Wang K, Tian J, Zheng C, Yang H, Ren J, Liu Y, Han Q, Zhang Y. Interpretable prediction of 3-year all-cause mortality in patients with heart failure caused by coronary heart disease based on machine learning and SHAP. Comput Biol Med. 2021;137:104813. doi: 10.1016/j.compbiomed.2021.104813. https://linkinghub.elsevier.com/retrieve/pii/S0010-4825(21)00607-7 . S0010- [PubMed] [CrossRef] [Google Scholar]. Zihni E, Madai VI, Livne M, Galinovic I, Khalil AA, Fiebach JB, Frey D. Opening the black box of artificial intelligence for clinical decision support: a study predicting stroke outcome. PLoS ONE. 2020;15(4):e0231166. 10.1371 . https://dx.plos.org/10.1371/journal.pone.0231166 . .PONE-D-19-31372 [PMC free article] [PubMed] [CrossRef] [Google Scholar]. /journal.pone.0231166. Lundberg SM, Lee SI. A unified approach to interpreting model predictions. NIPS'17: Proceedings of the 31st International Conference on Neural Information Processing Systems; NIPS'17; Dec 4–9; Long Beach, CA. 2017. pp. 4768–4777. [Google Scholar]. Van Ben C, Laure W, Verbeek Jan FM, Verbakel Jan Y, Christodoulou Evangelia, Vickers Andrew J, Roobol Monique J, Steyerberg Ewout W. Reporting and Interpreting Decision Curve Analysis: A Guide for Investigators. Eur Urol. 2018;74(6):796–804. https://europepmc.org/abstract/MED/30241973 . .S0302-2838(18)30640-7 [PMC free article] [PubMed] [CrossRef] [Google Scholar]. doi: 10.1016/j.eururo.2018.08.038. Lee C, Light A, Alaa A, Thurtle D, van der Schaar M, Gnanapragasam VJ. Application of a novel machine learning framework for predicting non-metastatic prostate cancer-specific mortality in men using the Surveillance, Epidemiology, and End Results (SEER) database. Lancet Digit Health. 2021;3(3):e158-e165. 10.1016/S2589-7500(20)30314-9 . Epub 2021 Feb 3. PMID: 33549512. Zhang C, Jiang Z, Shao C. Clinical characteristics of allergic bronchopulmonary aspergillosis. Clin Respir J. 2020. 10.1111/crj.13147 . Goto T, Hirayama A, Faridi MK. Obesity and severity of acute exacerbation of chronic obstructive pulmonary disease. Annals Am Thorac Soc. 2018;15(2):184–91. https://doi.org/10.1513/AnnalsATS.201706-485OC . Sanchez FF, Faganello MM, Tanni SE. Relationship between disease severity and quality of life in patients with chronic obstructive pulmonary disease. Braz J Med Biol Res. 2008;41(10):860–5. https://doi.org/10.1590/S0100-879X2008005000043 . Murugan AT, Sharma G. Obesity and respiratory diseases. Int J Obes. 2008;32(3):564–72. https://doi.org/10.1177/1479972308096978 . Comes A, Wong AW, Fisher JH, Morisset J, Johannson KA, Farrand E, Fell CD, Kolb M, Manganas H, Cox G, Gershon AS, Halayko AJ, Hambly N, Khalil N, Sadatsafavi M, Shapera S, To T, Wilcox PG, Collard HR, Ryerson CJ. Association of BMI and Change in Weight With Mortality in Patients With Fibrotic Interstitial Lung Disease. Chest. 2022;161(5):1320–9. Epub 2021 Nov 14. PMID: 34788669. Lu HW, Mao B, Wei P, Jiang S, Wang H. The clinical characteristics and prognosis of ABPA are closely related to the mucus plugs in central bronchiectasis. Clin Respir J. 2020. 10.1111/crj.13111 . Okada N, Yamamoto Y, Oguma T, Tanaka J. Allergic bronchopulmonary aspergillosis with atopic, nonatopic, and sans asthma—Factor analysis. Allergy. 2023. 10.1111/all.15820 . Zeng Y, Xue X, Cai H, Zhu G, Zhu M. Clinical characteristics and prognosis of allergic bronchopulmonary aspergillosis: a retrospective cohort study. J Asthma Allergy. 2022. 10.2147/JAA.S345427 . De Baets F, De Keyzer L, Van Daele S. Risk factors and impact of allergic bronchopulmonary aspergillosis in Pseudomonas aeruginosa–negative CF patients. Pediatr Allergy Immunol. 2018. 10.1111/pai.12953 . Wang S, Zhang J, Zhang C, Shao C. Clinical characteristics of allergic bronchopulmonary aspergillosis in patients with and without bronchiectasis. J Asthma Allergy. 2022. 10.1080/02770903.2021.1904979 . Wang X, Lu Y, Sun C, Zhong H, Cai Y. Development and validation of a machine learning-based diagnostic model for identifying nonneutropenic invasive pulmonary aspergillosis in suspected patients. Microbiol Spectr. 2025;e00607–25. https://doi.org/10.1128/spectrum.00607-25 . Du W, Ji W, Luo T, Zhang Y, Guo W. Development of a prognostic nomogram for nonneutropenic invasive pulmonary aspergillosis based on machine learning. J Inflamm Res. 2024;17:365–77. https://doi.org/10.2147/JIR.S499008 . Jeon ET, Lee HJ, Park TY, Jin KN, Ryu B, Lee HW. (2023). Machine learning-based prediction of in-ICU mortality in pneumonia patients. Sci Rep, 13, Article 11023. https://www.nature.com/articles/s41598-023-38765-8 Wang B, Li Y, Tian Y, Ju C, Xu X, Pei S. Novel pneumonia score based on a machine learning model for predicting mortality in pneumonia patients on admission to the intensive care unit. Comput Biol Med. 2023;162:107238. https://www.sciencedirect.com/science/article/pii/S0954611123002512 . Chia AHT, Khoo MS, Lim AZ, Ong KE, Sun Y. (2021). Explainable machine learning. prediction of ICU mortality. Heliyon, 7(8), e07714. https://www.sciencedirect.com/science/article/pii/S235291482100159 Tables Table 1 ‌All predictor variables for patients with ABPA Characteristic All-cause mortality Survival p value Total N 38 44 Baseline variables and in-hospital factors Age(years) 68.82 ± 11.69 74.07 ± 11.12 0.040 BMI(kg/m 2 ), 23.89 ± 2.80 25.05 ± 3.04 0.078 shock 1.00 (0.00–1.00) 0.00 (0.00–1.00) 0.102 peak leukocyte count(10 9 /L) 17.91 (14.75–32.52) 15.54 (11.02–22.93) 0.045 peak eosinophil count(10 9 /L) 0.19 (0.01–0.54) 0.33 (0.15–0.56) 0.129 lowest hemoglobin concentration(g/L) 64.00 (52.25-77.00) 71.00 (58.00-82.50) 0.154 lowest platelet count(10 9 /L) 49.50 (32.00-102.50) 93.50 (26.00-147.75) 0.076 peak procalcitonin level(ng/ml) 11.11 (1.92–28.92) 1.97 (0.88–6.90) < 0.001 peak C-reactive protein (CRP) level(mg/dl) 29.43 (10.78–33.06) 15.29 (10.13–27.54) 0.102 peak blood glucose level(mmol/L) 13.20 (12.03–17.88) 12.89 (9.90–18.50) 0.802 gender 0.429 Male 32 (84.21%) 34 (77.27%) Female 6 (15.79%) 10 (22.73%) Comorbidities asthma history 0.054 N 25 (65.79%) 37 (84.09%) Y 13 (34.21%) 7 (15.91%) history of corticosteroid use 0.005 N 28 (73.68%) 42 (95.45%) Y 10 (26.32%) 2 (4.55%) smoking history 0.852 N 35 (92.11%) 41 (93.18%) Y 3 (7.89%) 3 (6.82%) allergic rhinitis 0.057 N 38 (100.00%) 40 (90.91%) Y 0 (0.00%) 4 (9.09%) food and drug allergies 0.203 N 36 (94.74%) 38 (86.36%) Y 2 (5.26%) 6 (13.64%) diabetes mellitus 0.356 N 27 (71.05%) 27 (61.36%) Y 11 (28.95%) 17 (38.64%) concurrent bacterial infection 0.576 N 17 (44.74%) 17 (38.64%) Y 21 (55.26%) 27 (61.36%) concurrent viral infection 0.375 N 28 (73.68%) 36 (81.82%) Y 10 (26.32%) 8 (18.18%) renal insufficiency or failure < 0.001 N 6 (15.79%) 24 (54.55%) Y 32 (84.21%) 20 (45.45%) hepatic insufficiency or failure 0.165 N 28 (73.68%) 26 (59.09%) Y 10 (26.32%) 18 (40.91%) pulmonary fibrosis 0.010 N 26 (68.42%) 40 (90.91%) Y 12 (31.58%) 4 (9.09%) pulmonary cavities or bronchiectasis 0.017 N 15 (39.47%) 29 (65.91%) Y 23 (60.53%) 15 (34.09%) pulmonary infiltration 0.231 N 14 (36.84%) 22 (50.00%) Y 24 (63.16%) 22 (50.00%) Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7822617","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":547088998,"identity":"57952577-dd6e-4df2-a9bc-fb4c6bac877f","order_by":0,"name":"Jing Zhang","email":"","orcid":"","institution":"Yuebei People's Hospital Affiliated to Shantou University School of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Jing","middleName":"","lastName":"Zhang","suffix":""},{"id":547088999,"identity":"48b20ddd-5355-42cd-863f-ff2c8e12f11b","order_by":1,"name":"Juntao Tang","email":"","orcid":"","institution":"Yuebei People's Hospital Affiliated to Shantou University School of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Juntao","middleName":"","lastName":"Tang","suffix":""},{"id":547089000,"identity":"86f9a997-ab2a-469e-807e-ed0bf20f4b28","order_by":2,"name":"Jinjuan Li","email":"","orcid":"","institution":"Yuebei People's Hospital Affiliated to Shantou University School of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Jinjuan","middleName":"","lastName":"Li","suffix":""},{"id":547089001,"identity":"64e10131-b862-48c2-9b36-466061828362","order_by":3,"name":"Xinghua Liu","email":"","orcid":"","institution":"Yuebei People's Hospital Affiliated to Shantou University School of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Xinghua","middleName":"","lastName":"Liu","suffix":""},{"id":547089002,"identity":"016907c1-5354-4891-a0be-3a8a5ee3351c","order_by":4,"name":"Ye Sun","email":"","orcid":"","institution":"Yuebei People's Hospital Affiliated to Shantou University School of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Ye","middleName":"","lastName":"Sun","suffix":""},{"id":547089003,"identity":"665b0ae1-427c-4603-96ac-0608f9fa9fef","order_by":5,"name":"peng wang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA9UlEQVRIiWNgGAWjYFADZuaDDxIqJOTkidfCzpZs8OGMhbFhA9Fa+HnMJGe2VSQyHCCg0OD42cOvCyru2G04zGMmzTtPIoGxgfnhoxv4tJzJS7OeceZZ8sxmtmJr3m0SeewMbMbGOXi0mB3IMTPmbTuczM/MvPE2UEsxYwMPmzReLeffALX8O5zMxsxgIM07RyKx4QAhLTdyjB/zNhy242dmMZKc2UCEFvsbb8yYZxw7nCDZDArkYxLGhs0E/CLZn2P8uaDmsL3B+cPAqKypk5Nnb374GJ8WIGCTBhKJDXA+M37lYCWfQQ4krG4UjIJRMApGLAAAuHxMi1HkYlkAAAAASUVORK5CYII=","orcid":"","institution":"The Second Affiliated Hospital of Soochow University","correspondingAuthor":true,"prefix":"","firstName":"peng","middleName":"","lastName":"wang","suffix":""}],"badges":[],"createdAt":"2025-10-10 04:53:16","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7822617/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7822617/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":96286029,"identity":"ad92d591-ea9d-4f94-95c1-76de7668999e","added_by":"auto","created_at":"2025-11-19 12:00:37","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":56554,"visible":true,"origin":"","legend":"","description":"","filename":"script.docx","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/2d8e7bc30bf7e0e21ce47630.docx"},{"id":96286024,"identity":"c5f54473-fc76-4db1-9be5-27755235b39a","added_by":"auto","created_at":"2025-11-19 12:00:36","extension":"png","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":15078,"visible":true,"origin":"","legend":"","description":"","filename":"figure10.png","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/b288aa959597edff84d8c881.png"},{"id":96365163,"identity":"9fceda0c-b64a-4007-a4d0-d83f3cc55ecb","added_by":"auto","created_at":"2025-11-20 10:10:03","extension":"docx","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":36084,"visible":true,"origin":"","legend":"","description":"","filename":"table1.docx","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/d782f099cbbde96e67bd38ea.docx"},{"id":96364832,"identity":"701715bc-5674-4dc8-9e45-bb04c4c50c1f","added_by":"auto","created_at":"2025-11-20 10:09:42","extension":"jpeg","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":55253,"visible":true,"origin":"","legend":"","description":"","filename":"figure2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/92185c5facda3a598204acd8.jpeg"},{"id":96286032,"identity":"3bc9a9c5-d38d-4394-8e8b-f7aecd7f5fd6","added_by":"auto","created_at":"2025-11-19 12:00:37","extension":"jpeg","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":72968,"visible":true,"origin":"","legend":"","description":"","filename":"figure3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/9e8bc94be7f479d79f03f937.jpeg"},{"id":96364351,"identity":"698f6842-d8f6-4cc1-b69d-fe82aa6ea27b","added_by":"auto","created_at":"2025-11-20 10:09:13","extension":"png","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":11139,"visible":true,"origin":"","legend":"","description":"","filename":"figure4.png","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/2bd2379d2a92b94966bc6ccc.png"},{"id":96364111,"identity":"54d84925-5954-4cbf-830a-5d845c5789c0","added_by":"auto","created_at":"2025-11-20 10:08:54","extension":"jpeg","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":37321,"visible":true,"origin":"","legend":"","description":"","filename":"figure5.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/52358daa576d22c6d0ff5dbf.jpeg"},{"id":96286041,"identity":"62146c9f-fa1d-49b6-8822-0bd1f05b9c05","added_by":"auto","created_at":"2025-11-19 12:00:37","extension":"jpeg","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":79316,"visible":true,"origin":"","legend":"","description":"","filename":"figure6.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/e004940271ca9430fff9e8b2.jpeg"},{"id":96363726,"identity":"0ac9a3dc-0bbb-4088-8fc0-892bd247be3b","added_by":"auto","created_at":"2025-11-20 10:07:51","extension":"jpeg","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":60121,"visible":true,"origin":"","legend":"","description":"","filename":"figure7.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/bc5dc070ae708dc0de0ca508.jpeg"},{"id":96286037,"identity":"c282318b-6e4d-4fec-812f-55658092e3d2","added_by":"auto","created_at":"2025-11-19 12:00:37","extension":"png","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":17037,"visible":true,"origin":"","legend":"","description":"","filename":"figure8.png","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/2a7ba09906966c91e883422a.png"},{"id":96364181,"identity":"d77e7254-0d7c-459d-a428-4230efbddead","added_by":"auto","created_at":"2025-11-20 10:09:01","extension":"jpeg","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":185838,"visible":true,"origin":"","legend":"","description":"","filename":"figure9.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/09320690ca99c214449ca37e.jpeg"},{"id":96286044,"identity":"0a634ea2-774f-4b7b-ab9d-f6dbbcdd267d","added_by":"auto","created_at":"2025-11-19 12:00:37","extension":"json","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":7635,"visible":true,"origin":"","legend":"","description":"","filename":"5d86b00dc3ab4e5085e6f042c14ced62.json","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/85c75cbeae4f39d3e3ebb798.json"},{"id":96365487,"identity":"8812bcb9-4090-4102-afa9-01656e1bda56","added_by":"auto","created_at":"2025-11-20 10:10:24","extension":"xml","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":128781,"visible":true,"origin":"","legend":"","description":"","filename":"5d86b00dc3ab4e5085e6f042c14ced621enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/8f5ed4e03f140cbf77d6baf4.xml"},{"id":96286046,"identity":"5d5eef3d-1dc4-4c93-a7bb-fc3aca975263","added_by":"auto","created_at":"2025-11-19 12:00:37","extension":"pdf","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":18091,"visible":true,"origin":"","legend":"","description":"","filename":"figure1.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/0cea4769f73f2391ab6625bd.pdf"},{"id":96286047,"identity":"93c5e90a-32a6-4f98-8015-e53cc2be74cf","added_by":"auto","created_at":"2025-11-19 12:00:37","extension":"png","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":15078,"visible":true,"origin":"","legend":"","description":"","filename":"figure10.png","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/4433b27a2cec68d0bd8be472.png"},{"id":96364347,"identity":"c379e509-6c58-4b53-9a65-41c73a706a3c","added_by":"auto","created_at":"2025-11-20 10:09:13","extension":"jpeg","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":55253,"visible":true,"origin":"","legend":"","description":"","filename":"figure2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/6e44062500b1eaad7fcb2f49.jpeg"},{"id":96286050,"identity":"b505ffe4-d41e-4f8a-bfa2-74ef0a71299c","added_by":"auto","created_at":"2025-11-19 12:00:37","extension":"jpeg","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":72968,"visible":true,"origin":"","legend":"","description":"","filename":"figure3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/ff0168b6bc4c564967146a80.jpeg"},{"id":96286063,"identity":"1f47dcdc-f312-4fa8-be67-86dc6e4579df","added_by":"auto","created_at":"2025-11-19 12:00:37","extension":"png","order_by":18,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":11139,"visible":true,"origin":"","legend":"","description":"","filename":"figure4.png","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/2e7d9dd8190b7541dd876e32.png"},{"id":96364512,"identity":"b4350906-5654-4304-980c-cb3e119872e0","added_by":"auto","created_at":"2025-11-20 10:09:23","extension":"jpeg","order_by":19,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":37321,"visible":true,"origin":"","legend":"","description":"","filename":"figure5.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/069fbbcf75ceffef5a5b1cce.jpeg"},{"id":96364851,"identity":"0f44979a-94f4-494c-b995-d0dbca2268d5","added_by":"auto","created_at":"2025-11-20 10:09:43","extension":"jpeg","order_by":20,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":79316,"visible":true,"origin":"","legend":"","description":"","filename":"figure6.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/77f0567864cc7cf15346fe32.jpeg"},{"id":96364611,"identity":"acb8fb15-5209-4e9d-b897-8aa17f80e528","added_by":"auto","created_at":"2025-11-20 10:09:29","extension":"jpeg","order_by":21,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":60121,"visible":true,"origin":"","legend":"","description":"","filename":"figure7.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/dcac987f763d0ff5aeb284f5.jpeg"},{"id":96364097,"identity":"0d1d4f9d-8baa-4788-b84e-97498c2a569d","added_by":"auto","created_at":"2025-11-20 10:08:53","extension":"png","order_by":22,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":17037,"visible":true,"origin":"","legend":"","description":"","filename":"figure8.png","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/cbd2c43daae8bc7ab8c40df3.png"},{"id":96286059,"identity":"5c78bdeb-654b-4f99-a9e0-1e658750c550","added_by":"auto","created_at":"2025-11-19 12:00:37","extension":"jpeg","order_by":23,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":185838,"visible":true,"origin":"","legend":"","description":"","filename":"figure9.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/5bd34152fbd33e6586d66436.jpeg"},{"id":96365384,"identity":"bb20a91b-f3dc-48b7-b863-708f8290e6a5","added_by":"auto","created_at":"2025-11-20 10:10:19","extension":"png","order_by":24,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":14382,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefigure10.png","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/6e8b08fc566ed77bf3b00747.png"},{"id":96286057,"identity":"83dc29bb-0d33-48fc-9af4-8f9a2bd1e512","added_by":"auto","created_at":"2025-11-19 12:00:37","extension":"png","order_by":25,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":15387,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefigure2.png","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/24220fd6d471804ac0b9c910.png"},{"id":96286053,"identity":"7f232ee0-6ec2-4809-b02b-9c035fa96e96","added_by":"auto","created_at":"2025-11-19 12:00:37","extension":"png","order_by":26,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":9371,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefigure3.png","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/d48fb689a39c5361b940ec05.png"},{"id":96364338,"identity":"9a7bc0dc-a48d-406a-9cd9-961285bfb45f","added_by":"auto","created_at":"2025-11-20 10:09:13","extension":"png","order_by":27,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":9833,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefigure4.png","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/c8a8c9c8bc8cd490550bbe7b.png"},{"id":96286055,"identity":"5e6d268a-19d6-4bea-961a-52832efa92dc","added_by":"auto","created_at":"2025-11-19 12:00:37","extension":"png","order_by":28,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":13295,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefigure5.png","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/3c52dfd93f67fcdb5410d85a.png"},{"id":96286064,"identity":"9ed99d9a-47eb-4310-aa35-6b19cbf770b2","added_by":"auto","created_at":"2025-11-19 12:00:37","extension":"png","order_by":29,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":10060,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefigure6.png","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/a041925327690a7e611ddd7c.png"},{"id":96286060,"identity":"1d568c13-24f4-4a0e-be3d-a6a5ed1beae3","added_by":"auto","created_at":"2025-11-19 12:00:37","extension":"png","order_by":30,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":8943,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefigure7.png","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/f106b45428f10402d97fef34.png"},{"id":96286061,"identity":"582d94a7-b041-4422-904a-69b5ae64c99f","added_by":"auto","created_at":"2025-11-19 12:00:37","extension":"png","order_by":31,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":16167,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefigure8.png","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/e266b2bca02ef79cbbc57ed6.png"},{"id":96365075,"identity":"7b4d06aa-7b54-4911-99c0-2b2a2e0f041d","added_by":"auto","created_at":"2025-11-20 10:09:58","extension":"png","order_by":32,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":20599,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefigure9.png","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/d04bfc5e954030bccfe18830.png"},{"id":96364270,"identity":"64bfae65-9ddb-4383-8fb4-68a7b6d8604f","added_by":"auto","created_at":"2025-11-20 10:09:08","extension":"xml","order_by":33,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":126710,"visible":true,"origin":"","legend":"","description":"","filename":"5d86b00dc3ab4e5085e6f042c14ced621structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/146550c073fd05f027dfdf8d.xml"},{"id":96286068,"identity":"c9d1236f-e2ed-4eec-9a9c-b6d4f907fc2a","added_by":"auto","created_at":"2025-11-19 12:00:38","extension":"html","order_by":34,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":139425,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/e754ff5c68f2ae77971c798e.html"},{"id":96286025,"identity":"394e7b23-87e4-48ef-94de-e5830031c25d","added_by":"auto","created_at":"2025-11-19 12:00:36","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":99833,"visible":true,"origin":"","legend":"\u003cp\u003eFlowchart of patient selection\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/f17780f631d7a3643caddc3b.png"},{"id":96286026,"identity":"bc4a72c8-0275-4513-b826-24a4cafd0180","added_by":"auto","created_at":"2025-11-19 12:00:37","extension":"jpeg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":55253,"visible":true,"origin":"","legend":"\u003cp\u003eReceiver Operating Characteristic (ROC) Curve of the XGBoost Model in the Training Cohort of Patients with ABPA\u003c/p\u003e","description":"","filename":"figure2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/411e6c86d3c9a2e3acfd0e21.jpeg"},{"id":96286030,"identity":"91e1f834-1a07-41e2-b0b0-32ab24eb6d8f","added_by":"auto","created_at":"2025-11-19 12:00:37","extension":"jpeg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":72968,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDecision curve analysis (DCA) of the predictive model in the training cohort for in-hospital mortality among ICU patients with ABPA.\u003c/strong\u003e\u003cbr\u003e\n The y-axis indicates net benefit, and the x-axis shows the high-risk threshold (probability threshold), with the lower axis displaying the corresponding cost:benefit ratio. The blue curve (‘DCA.tr’) represents the model’s net benefit across thresholds, the grey curve (‘All’) represents the net benefit of treating all patients, and the black line (‘None’) indicates the net benefit if no patients are treated. The DCA demonstrates that the model provides a higher net benefit across a wide range of clinically relevant thresholds compared to ‘treat all’ and ‘treat none’ strategies.\u003c/p\u003e","description":"","filename":"figure3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/b310341f50c7ca5b4e4ba8da.jpeg"},{"id":96286027,"identity":"e4b1572e-9ec8-4bfb-88fe-5856ba09089b","added_by":"auto","created_at":"2025-11-19 12:00:37","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":11139,"visible":true,"origin":"","legend":"\u003cp\u003eCalibration Curve Analysis of the XGBoost Model in the Training Cohort of Patients with ABPA\u003c/p\u003e","description":"","filename":"figure4.png","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/65b01aa155ae2b8c864442b9.png"},{"id":96365347,"identity":"bf633db7-89c9-4c8e-a8d8-c0bc386b90e7","added_by":"auto","created_at":"2025-11-20 10:10:17","extension":"jpeg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":37321,"visible":true,"origin":"","legend":"\u003cp\u003eReceiver Operating Characteristic (ROC) Curve of the XGBoost Model in the Validation Cohort of Patients with ABPA\u003c/p\u003e","description":"","filename":"figure5.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/b919a8ed767f3922f6b1088a.jpeg"},{"id":96286034,"identity":"ae30663f-77b8-41ae-9c56-079dabdc61bc","added_by":"auto","created_at":"2025-11-19 12:00:37","extension":"jpeg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":79316,"visible":true,"origin":"","legend":"\u003cp\u003eDecision Curve Analysis (DCA) of the predictive model in the validation cohort for in-hospital mortality among ICU patients with ABPA.\u003cbr\u003e\n The y-axis represents net benefit, while the x-axis indicates the high-risk threshold (probability threshold). The lower axis shows the corresponding cost:benefit ratio for each threshold. The blue curve (‘DCA’) shows the net benefit of the predictive model in the validation cohort across different thresholds. The grey curve (‘All’) represents the net benefit of treating all patients, and the black line (‘None’) represents the net benefit of treating no patients. The DCA demonstrates that the predictive model provides a higher net benefit across a clinically relevant range of thresholds compared to ‘treat all’ and ‘treat none’ strategies.\u003c/p\u003e","description":"","filename":"figure6.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/756714cdf2cac23f06b34462.jpeg"},{"id":96286043,"identity":"442c9009-2dcb-4759-8c0b-2d91254c29de","added_by":"auto","created_at":"2025-11-19 12:00:37","extension":"jpeg","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":60121,"visible":true,"origin":"","legend":"\u003cp\u003eCalibration Curve Analysis of the XGBoost Model in the Validation Cohort of Patients with ABPA\u003c/p\u003e","description":"","filename":"figure7.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/2c19644ea27590f998dd361d.jpeg"},{"id":96364969,"identity":"4cd83fd0-9855-4e2b-951b-f2c7db55be08","added_by":"auto","created_at":"2025-11-20 10:09:51","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":17037,"visible":true,"origin":"","legend":"\u003cp\u003eThe weights of variables \u0026nbsp;importance\u003c/p\u003e","description":"","filename":"figure8.png","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/30e9edfc5080194b5f53f312.png"},{"id":96286039,"identity":"a1ba2fea-ff0b-4045-b893-de57d4295bbe","added_by":"auto","created_at":"2025-11-19 12:00:37","extension":"jpeg","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":185838,"visible":true,"origin":"","legend":"\u003cp\u003eThe SHapley Additive exPlanation(SHAP)values\u003c/p\u003e","description":"","filename":"figure9.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/5ec380b01ca9c5a59f02c13a.jpeg"},{"id":96365757,"identity":"ae7e0032-c1a9-4adf-a23f-86e232d5f144","added_by":"auto","created_at":"2025-11-20 10:10:44","extension":"png","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":15078,"visible":true,"origin":"","legend":"\u003cp\u003eIndividualized SHapley Additive exPlanation (SHAP) Force Plot Interpretation for Risk Stratification in Allergic Bronchopulmonary Aspergillosis (ABPA) Patients\u003c/p\u003e","description":"","filename":"figure10.png","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/a4dfcb13fb48ce5658d5edb9.png"},{"id":104887935,"identity":"12d98366-7455-4eb8-8ad2-a94a0f28749a","added_by":"auto","created_at":"2026-03-18 10:12:58","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2017028,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7822617/v1/c0988bb5-8a67-4c4d-913b-14d8500d4dd8.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Predicting Mortality in Intensive Care Unit Patients with Allergic Bronchopulmonary Aspergillosis (ABPA) Using an Interpretable Machine Learning Model: A Retrospective Cohort Study","fulltext":[{"header":"Introduction","content":"\u003cp\u003eAllergic bronchopulmonary aspergillosis (ABPA) is a hypersensitivity lung disease caused by a complex immune reaction triggered by Aspergillus colonization in the airways \u003csup\u003e[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]\u003c/sup\u003e. ABPA typically occurs in patients with preexisting respiratory conditions such as asthma and cystic fibrosis (CF), and it is characterized by difficult-to-control asthma-like symptoms and recurrent pulmonary infiltrates, with or without bronchiectasis. With increasing awareness among clinicians, ABPA has also been identified as a secondary condition in patients with bronchiectasis, chronic obstructive pulmonary disease (COPD), and other pulmonary disorders \u003csup\u003e[\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]\u003c/sup\u003e. The prevalence of ABPA in the general population remains unclear; however, it has been reported to range from 2% to 32% in asthma patients \u003csup\u003e[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]\u003c/sup\u003e and from 3% to 25% in CF patients \u003csup\u003e[\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eThe prognosis of ABPA largely depends on early diagnosis and timely treatment. The natural course of ABPA is variable, with recurrent exacerbations and remissions being its hallmark. Long-term recurrence can lead to irreversible lung damage, including bronchiectasis, pulmonary fibrosis, and even progression to chronic pulmonary heart disease or respiratory failure. Serum total IgE levels are commonly used as a biomarker for monitoring ABPA disease activity\u003csup\u003e[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]\u003c/sup\u003e. High-attenuation mucus (HAM) impaction and the severity of central bronchiectasis have been identified as independent risk factors for ABPA relapse \u003csup\u003e[\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eCritically ill patients in the intensive care unit (ICU) require specialized care and multidisciplinary support \u003csup\u003e[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]\u003c/sup\u003e. Although the ICU plays a crucial role in sustaining patients\u0026rsquo; lives, it also poses challenges such as workforce shortages, limited medical resources, and significant financial burdens \u003csup\u003e[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]\u003c/sup\u003e. Therefore, early identification of hospital mortality risk in ICU patients is essential, as it may facilitate appropriate treatment strategies and support clinical decision-making \u003csup\u003e[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eIn recent years, artificial intelligence has been widely employed to identify early warning predictors for various diseases. Given the inherent strength of machine learning algorithms in capturing nonlinear relationships, an increasing number of researchers advocate for the use of machine learning-based predictive models to support appropriate patient treatment, rather than relying solely on traditional disease severity scoring systems such as the Sequential Organ Failure Assessment (SOFA), Acute Physiology and Chronic Health Evaluation II (APACHE II), or Simplified Acute Physiology Score II (SAPS II) \u003csup\u003e[\u003cspan additionalcitationids=\"CR13\" citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]\u003c/sup\u003e. Although numerous predictive models have demonstrated promising performance in research settings, their clinical application remains limited, and there is still a lack of evidence supporting interpretable risk prediction models that contribute to disease prognosis \u003csup\u003e[\u003cspan additionalcitationids=\"CR16 CR17\" citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eThe use of an explainable machine learning model based on Extreme Gradient Boosting (XGBoost) to predict mortality in intensive care unit (ICU) patients with allergic bronchopulmonary aspergillosis (ABPA) and to perform internal validation represents an emerging research direction. By analyzing clinical data, the XGBoost model can effectively predict mortality in ABPA patients while providing interpretable predictions. Internal validation, achieved through methods such as cross-validation, ensures the stability and reliability of the model. This approach not only assists clinicians in better assessing patient prognosis but also provides valuable insights for developing personalized treatment strategies.\u003c/p\u003e\u003cp\u003eThe objective of this study is to utilize existing clinical data and patient characteristics from ICU admission to construct and interpret an XGBoost model for ABPA prognosis. The SHapley Additive Explanations (SHAP) method is employed to explain the model's predictions and explore prognostic factors associated with ABPA.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003eData Source\u003c/h2\u003e\u003cp\u003eThis retrospective study collected clinical data from patients diagnosed with allergic bronchopulmonary aspergillosis (ABPA) at Yuebei People's Hospital between January 2020 and July 2024. The diagnosis was confirmed based on metagenomic sequencing of sputum, which identified \u003cem\u003eAspergillus\u003c/em\u003e infection. The dataset comprehensively recorded demographic information, vital signs, diagnostic details, and treatment information for all patients.\u003c/p\u003e\u003cp\u003eThe inclusion criteria were based on the \u003cb\u003e2017 Expert Consensus on the Diagnosis and Treatment of Allergic Bronchopulmonary Aspergillosis\u003c/b\u003e established by the Asthma Group of the Chinese Thoracic Society. Patients were included if they met the following criteria:\u003c/p\u003e\u003cp\u003e\u003col\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eAssociated diseases\u003c/b\u003e: Presence of asthma or other underlying respiratory conditions, such as bronchiectasis, chronic obstructive pulmonary disease (COPD), or pulmonary cystic fibrosis.\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eEssential criteria\u003c/b\u003e:\u003c/p\u003e\u003cp\u003e\u003col\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003e\u003cem\u003eAspergillus fumigatus\u003c/em\u003e-specific IgE\u0026thinsp;\u0026gt;\u0026thinsp;0.35 kUA/L or a positive immediate skin test reaction to \u003cem\u003eAspergillus fumigatus\u003c/em\u003e.\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003eElevated serum total IgE levels (\u0026gt;\u0026thinsp;1000 IU/mL).\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003c/ol\u003e\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eAdditional criteria\u003c/b\u003e:\u003c/p\u003e\u003cp\u003e\u003col\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003eBlood eosinophil count\u0026thinsp;\u0026gt;\u0026thinsp;0.5 \u0026times; 10⁹/L.\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003ePulmonary imaging findings consistent with ABPA, including but not limited to pulmonary consolidation, nodules, toothpaste sign, finger-in-glove sign, transient migratory opacities, or persistent changes such as bronchiectasis and pleuroparenchymal fibrosis.\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003ePositive serum \u003cem\u003eAspergillus\u003c/em\u003e-specific IgG antibodies or precipitin test.\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003c/ol\u003e\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003c/ol\u003e\u003c/p\u003e\u003cp\u003eA diagnosis of ABPA required meeting criterion \u003cb\u003e1\u003c/b\u003e, criterion \u003cb\u003e2\u003c/b\u003e, and at least two conditions from criterion \u003cb\u003e3\u003c/b\u003e. Patients with serum total IgE levels\u0026thinsp;\u0026lt;\u0026thinsp;1000 IU/mL could still be diagnosed with ABPA if all other criteria were met. Additionally, only patients aged 18 years or older were included in the study.\u003c/p\u003e\u003cp\u003eThe exclusion criteria were as follows:\u003c/p\u003e\u003cp\u003e\u003col\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003ePresence of malignancy or severe organ dysfunction (e.g., cardiac, cerebral, or renal failure).\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003ePregnant or lactating women.\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003eIncomplete medical records or insufficient diagnostic data.\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003c/ol\u003e\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eEthical Considerations\u003c/h3\u003e\n\u003cp\u003e This study was approved by the Ethics Committee of Yuebei People's Hospital. As all protected health information was anonymized and the study was retrospective in nature, the requirement for informed consent from patients and their families was waived.\u003c/p\u003e\n\u003ch3\u003ePredictor Variables and Explainable Machine Learning Tools\u003c/h3\u003e\n\u003cp\u003eThe outcome variable in this study was the probability of in-hospital mortality, defined based on the patient\u0026rsquo;s status at discharge.Model interpretability was enhanced using SHapley Additive Explanations (SHAP), which quantify the individual contribution of each feature to the prediction outcome\u003csup\u003e[\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]\u003c/sup\u003e. SHAP values indicate the extent to which each predictor variable influences the target variable, either positively or negatively. Additionally, each individual observation in the dataset can be explained by a specific set of SHAP values.\u003c/p\u003e\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\u003ch2\u003eStatistical Analysis\u003c/h2\u003e\u003cp\u003eAll statistical analyses and computations were performed using R software (version 3.8.0). Categorical variables were expressed as counts and percentages, and differences between groups were compared using the chi-square (χ\u0026sup2;) test or Fisher\u0026rsquo;s exact test (when the expected frequency was \u0026lt;\u0026thinsp;10). Continuous variables were reported as medians with interquartile ranges (IQR) and were compared between groups using the Wilcoxon rank-sum test.\u003c/p\u003e\u003cp\u003eThe predictive model was developed using the Extreme Gradient Boosting (XGBoost) machine learning algorithm. Model performance was evaluated by calculating the area under the receiver operating characteristic (ROC) curve (AUC). Additionally, accuracy, sensitivity, positive predictive value (PPV), and negative predictive value (NPV) were computed. To assess the clinical utility of the model, decision curve analysis (DCA) was conducted, which quantifies net benefit across different probability thresholds for decision-making \u003csup\u003e[\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eSHAP values were calculated using the R language implementation, enabling interpretation of individual feature contributions to the model predictions.\u0026rdquo;, which is specifically optimized for tree-based models such as XGBoost. This approach enables calculation of consistent, locally accurate feature attribution values for each patient\u0026rsquo;s mortality prediction.\u003c/p\u003e\u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003ePatient Characteristics\u003c/h2\u003e\u003cp\u003eA total of 82 adult patients diagnosed with allergic bronchopulmonary aspergillosis (ABPA) in the ICU were included in the final cohort of this study. The patient screening process is illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. The dataset was split into a training set (70%) and a validation set (30%) using stratified random sampling based on the outcome variable (in-hospital mortality) to ensure balanced distribution of the event across the subsets.The dataset was randomly divided into two subsets: 70% (n\u0026thinsp;=\u0026thinsp;57) of the data was used for model training, while 30% (n\u0026thinsp;=\u0026thinsp;25) was used for model validation. A total of 24 potential predictive factors were identified and used for model development.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003ePatients in the non-survivor group were older than those in the survivor group (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.04). The in-hospital mortality rate was 46.3% (26/57) in the training dataset and 48% (12/25) in the testing dataset. Table\u0026nbsp;1 presents a comparison of predictor variables between survivors and non-survivors during hospitalization.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eModel Development and Evaluation\u003c/h3\u003e\n\u003cp\u003eAn XGBoost model was developed using the training dataset, and its receiver operating characteristic (ROC) curve for the test dataset is shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, with an area under the curve (AUC) of 0.995 (95% CI: 0.903\u0026ndash;1.000).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eDecision curve analysis (DCA) was performed on the test dataset to compare the net benefit of the optimal model against alternative clinical decision-making strategies. Clinical net benefit was defined as the minimum probability of disease occurrence at which further intervention would be warranted \u003csup\u003e[\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]\u003c/sup\u003e. The DCA curve quantifies net benefit across different probability thresholds, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eA calibration curve illustrating the discrepancy between predicted and actual mortality is presented in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e. Due to the heterogeneity of the study population, machine learning-based treatment strategies outperformed the default strategies of treating all or treating none of the patients.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eFor the validation dataset, the ROC curve is shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e, with an AUC of 0.881 (95% CI: 0.846\u0026ndash;0.909). Additionally, DCA (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e) and calibration curve analysis (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e) were conducted on the validation dataset, demonstrating excellent predictive performance of the model.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\n\u003ch3\u003eInterpretation of the XGBoost Model Using SHAP Method\u003c/h3\u003e\n\u003cp\u003eThe SHAP algorithm was employed to assess the importance of each predictor variable in the XGBoost model\u0026rsquo;s predictions. The variable importance plot ranks the most significant features in descending order (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003e). Among all predictors, \u003cb\u003eBMI\u003c/b\u003e had the strongest predictive value, followed by \u003cb\u003epeak procalcitonin level, peak eosinophil count, age, asthma history, peak leukocyte count\u003c/b\u003e, and \u003cb\u003elowest platelet count\u003c/b\u003e.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eAdditionally, to determine the direction of the relationship between predictors and the outcome, SHAP values were used to identify mortality risk factors. As shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003e, the horizontal position indicates whether a given feature's effect is associated with a higher or lower mortality prediction, while the color represents whether the variable value is high (red) or low (blue) for a specific observation. The results indicate that an \u003cb\u003eincrease in BMI positively influences mortality prediction\u003c/b\u003e, shifting it toward a higher probability of death, whereas \u003cb\u003ean increase in age negatively influences mortality prediction\u003c/b\u003e, shifting it toward survival.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\u003ch2\u003eSHAP Force Plot\u003c/h2\u003e\u003cp\u003eFigure \u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003e presents the individual force plot for a specific patient. The SHAP values illustrate the predictive features relevant to that patient and the contribution of each feature to the mortality prediction. The \u003cb\u003ebold number\u003c/b\u003e represents the probability prediction value (\u003cem\u003ef(x)\u003c/em\u003e), while the \u003cb\u003ebaseline value\u003c/b\u003e corresponds to the model\u0026rsquo;s prediction without any input data. \u003cem\u003ef(x)\u003c/em\u003e represents the log-odds ratio for each observation.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eIn the force plot, In the force plot, red features (on the left) represent variables that increase the model\u0026rsquo;s predicted risk of mortality for that individual patient, while blue features (on the right) represent variables that decrease the prediction.. The length of the arrows visually represents the magnitude of each feature\u0026rsquo;s influence on the prediction\u0026mdash;the longer the arrow, the greater the impact.\u003c/p\u003e\u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn this study, we developed an XGBoost machine learning model based on clinical data from the intensive care unit (ICU) to predict in-hospital mortality in patients with allergic bronchopulmonary aspergillosis (ABPA). The model\u0026rsquo;s performance was evaluated through internal validation. Our results demonstrated that the XGBoost model exhibited high predictive accuracy in distinguishing mortality risk among ABPA patients. Decision curve analysis (DCA) showed significant clinical net benefit in the probability threshold range of 10%-20%, suggesting that the model can provide valuable reference information for clinical intervention within this risk range. Using the SHAP method to interpret the model, we identified key clinical features closely associated with ABPA mortality, including \u003cb\u003eBMI, peak procalcitonin level, peak eosinophil count, age, asthma history, peak leukocyte count\u003c/b\u003e, and \u003cb\u003elowest platelet count\u003c/b\u003e. These findings not only enhance the interpretability of the model but also provide new insights into the mechanisms underlying ABPA severity.\u003c/p\u003e\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\u003ch2\u003eClinical Significance of Key Variables\u003c/h2\u003e\u003cp\u003eThrough an in-depth analysis of the model\u0026rsquo;s feature contributions using the SHAP method, we identified \u003cb\u003eseven key clinical predictors\u003c/b\u003e associated with mortality in ABPA patients. Among them, \u003cb\u003eBMI\u003c/b\u003e was the most significant predictor of mortality. Other important variables, such as \u003cb\u003epeak procalcitonin level, peak eosinophil count, age, asthma history, peak leukocyte count\u003c/b\u003e, and \u003cb\u003elowest platelet count\u003c/b\u003e, also played critical roles in patient prognosis \u003csup\u003e[\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003e\u003cb\u003eBMI\u003c/b\u003e, as an indicator of metabolic status and systemic inflammation, may influence patient outcomes by exacerbating systemic inflammatory responses. Previous studies have also linked \u003cb\u003eBMI\u003c/b\u003e to the severity of pulmonary diseases\u003csup\u003e[\u003cspan additionalcitationids=\"CR24 CR25\" citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]\u003c/sup\u003e. An \u003cb\u003eelevated peak procalcitonin level\u003c/b\u003e is typically indicative of severe infections, and its abnormal levels in ABPA may signal a poor prognosis \u003csup\u003e[\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]\u003c/sup\u003e. Additionally, an \u003cb\u003eincrease in peak eosinophil count\u003c/b\u003e is associated with more severe airway inflammation, which is closely related to the immunopathological mechanisms of ABPA \u003csup\u003e[\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003e\u003cb\u003eAge\u003c/b\u003e is a key factor affecting immune function and may contribute to disease progression in ABPA. Studies have shown that older patients are more likely to experience adverse clinical outcomes \u003csup\u003e[\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]\u003c/sup\u003e. Interestingly, in our cohort, SHAP analysis suggested a negative correlation between age and mortality, which may be due to selection bias where younger ICU patients had more severe disease presentations, warranting further validation in larger cohorts.A \u003cb\u003ehistory of asthma\u003c/b\u003e is closely linked to the pathogenesis of ABPA, as persistent airway inflammation and structural changes can increase disease complexity. Furthermore, an \u003cb\u003eelevated peak leukocyte count\u003c/b\u003e often indicates systemic inflammation or infection, which may be associated with a higher risk of mortality in severe cases \u003csup\u003e[\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]\u003c/sup\u003e. A \u003cb\u003edecrease in lowest platelet count\u003c/b\u003e may reflect endothelial damage or coagulation dysfunction, which is also a marker of poor prognosis in critically ill ABPA patients \u003csup\u003e[\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eIn this study, we developed a mortality risk prediction model for patients with allergic bronchopulmonary aspergillosis (ABPA) in the intensive care unit (ICU) setting, utilizing XGBoost combined with SHAP for model interpretability. This work helps to address a gap in the current literature by focusing on a rare disease (ABPA) within the unique context of critical care.\u003c/p\u003e\u003cp\u003eOver the past five years, relevant studies have shown that while machine learning has been widely applied for outcome prediction among critically ill ICU patients (such as those with pneumonia or invasive fungal infections), research specifically targeting ABPA patients\u0026mdash;particularly those in the ICU\u0026mdash;remains extremely limited. For example, Wang et al. developed a machine learning model for identifying invasive pulmonary aspergillosis in non-neutropenic patients, which included nine ABPA cases, suggesting that machine learning can aid diagnosis and risk stratification in fungal-related diseases; however, its application was limited to specific subgroups and did not focus on mortality prediction in ICU patients with ABPA \u003csup\u003e[\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]\u003c/sup\u003e. Similarly, Du et al. proposed a prognostic model combining lasso regression and machine learning for non-neutropenic pulmonary aspergillosis but did not explicitly include ABPA nor target ICU patients \u003csup\u003e[\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]\u003c/sup\u003e. In contrast, Jeon et al. and Wang et al. have conducted machine learning-based mortality prediction studies in ICU patients with pneumonia, using methods such as random forests and XGBoost, demonstrating significant improvements over traditional scoring systems\u003csup\u003e[\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e, \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eAdditionally, Chia et al. explored the application of interpretable machine learning for mortality prediction in the ICU, emphasizing the importance of model transparency for clinical acceptance, which aligns closely with our use of SHAP in this study\u003csup\u003e[\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eTaken together, compared to existing literature, our study expands and contributes to the field in several key aspects:\u003c/p\u003e\u003cp\u003e\u003col\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eUnique study population\u003c/b\u003e: This is the first study to develop a mortality risk prediction model specifically for ICU patients with ABPA, helping to fill an existing research gap.\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eMethodological innovation\u003c/b\u003e: By applying XGBoost in combination with SHAP, we aim to enhance predictive performance while ensuring model interpretability, providing a practical tool for clinical application.\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eValue for small-sample, rare disease modeling\u003c/b\u003e: Our approach offers experience and insight into developing predictive models for rare diseases under low-sample conditions, which may help advance research in similar clinical areas.\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003c/ol\u003e\u003c/p\u003e\u003cp\u003eWe hope these contributions will provide a useful reference for future studies in the field and support the development of effective, interpretable machine learning applications in critical care for rare disease populations.\u003c/p\u003e\u003cp\u003eIn summary, these key clinical features provide a solid basis for model interpretation and offer valuable reference points for risk assessment in clinical interventions. The ability of the XGBoost model to integrate these variables through non-linear relationships highlights the advantages of machine learning in capturing complex clinical interactions. It is important to note that the SHAP force plot illustrates the contribution of each feature to an individual prediction and does not necessarily reflect general clinical associations.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e\u003ch2\u003eModel Limitations and Challenges in Clinical Translation\u003c/h2\u003e\u003cp\u003eAlthough the model demonstrated robustness through internal validation, several limitations should be considered. First, the mortality rate among ABPA patients in the final cohort was relatively low (46.34%). At a threshold corresponding to 80% specificity, the model\u0026rsquo;s positive predictive value was 0.20, likely reflecting sample heterogeneity and moderate event prevalence. This result may be influenced by sample heterogeneity within the single-center dataset and the relative rarity of severe ABPA cases. Additionally, the lack of an external validation cohort may limit the generalizability of the model, particularly in patients with immunosuppression or other pulmonary comorbidities.\u003c/p\u003e\u003cp\u003eCurrently, this model is better suited for risk stratification rather than direct clinical decision-making. Future prospective studies should evaluate the actual benefit and cost-effectiveness of early interventions (e.g., intensified antifungal therapy or immunomodulation) for high-risk patients.\u003c/p\u003e\u003c/div\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThis study demonstrated that the \u003cb\u003eXGBoost\u003c/b\u003e model can effectively predict mortality risk in \u003cb\u003eABPA\u003c/b\u003e patients and, through interpretability analysis, reveal key pathophysiological mechanisms. However, \u003cb\u003eclinical translation of the model requires addressing limitations such as low positive predictive value and the absence of external validation\u003c/b\u003e. Future research should incorporate \u003cb\u003emulticenter cohorts\u003c/b\u003e and \u003cb\u003edynamic biomarkers\u003c/b\u003e (e.g., serial IgE monitoring or radiomic features of pulmonary imaging) to optimize predictive accuracy and support \u003cb\u003epersonalized treatment strategies\u003c/b\u003e.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable. \u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026rsquo; contributions\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePeng Wang,Ye Sun,and Jing Zhang designed the work. Juntao Tang, Jinjuan Li and Xinghua Liu record and summarized the patient of features. Peng Wang and Jing Zhang analyzed datasets. Jing Zhang wrote this paper. All authors read and approved the fnal manuscript.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was reviewed and approved by the Ethics Committee of Yuebei People\u0026apos;s Hospital (Ethics Approval Number: YBSKY-2025-029-001). The committee granted a waiver of informed consent for this research in accordance with national regulations and institutional ethical guidelines. The waiver was justified based on the retrospective and anonymized nature of the data analyzed, which posed no foreseeable risks to participants\u0026apos; privacy or rights. All personal identifiers were removed from the dataset prior to analysis to ensure confidentiality. The study strictly adhered to the principles outlined in the Declaration of Helsinki and relevant Chinese ethical standards for biomedical research involving human subjects.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eClinical trial number\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no competing interests.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eAgarwal R, Sehgal IS, Dhooria S, Muthu V, Prasad KT, Bal A, Aggarwal AN, Chakrabarti A. Allergic bronchopulmonary aspergillosis. Indian J Med Res. 2020;151(6):529\u0026ndash;49. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.4103/ijmr.IJMR_1187_19\u003c/span\u003e\u003cspan address=\"10.4103/ijmr.IJMR_1187_19\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PMID: 32719226; PMCID: PMC7602921.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTiew PY, Lim AYH, Keir HR, Dicker AJ, Mac Aog\u0026aacute;in M, Pang SL, Low TB, Hassan TM, Poh ME, Xu H, Ong TH, Koh MS, Abisheganaden JA, Tee A, Chew FT, Chalmers JD, Chotirmall SH. High Frequency of Allergic Bronchopulmonary Aspergillosis in Bronchiectasis-COPD Overlap. Chest. 2022;161(1):40\u0026ndash;53. Epub 2021 Aug 5. PMID: 34364870.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhang P, Ma Y, Chen X, Ma Y, Yang L, Zhang M, Gao Z. The Difference in All-Cause Mortality Between Allergic Bronchopulmonary Aspergillosis with and without Chronic Obstructive Pulmonary Disease. J Asthma Allergy. 2022;15:1861\u0026ndash;75. PMID: 36601290; PMCID: PMC9807121.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAgarwal R, Aggarwal AN, Gupta D, Jindal SK. Aspergillus hypersensitivity and allergic bronchopulmonary aspergillosis in patients with bronchial asthma: systematic review and meta-analysis. Int J Tuberc Lung Dis. 2009;13(8):936\u0026ndash;44. PMID: 19723372.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMaturu VN, Agarwal R. Prevalence of Aspergillus sensitization and allergic bronchopulmonary aspergillosis in cystic fibrosis: systematic review and meta-analysis. Clin Exp Allergy. 2015;45(12):1765-78. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1111/cea.12595\u003c/span\u003e\u003cspan address=\"10.1111/cea.12595\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PMID: 26177981.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAgarwal R, Chakrabarti A, Shah A, Gupta D, Meis JF, Guleria R, Moss R, Denning DW. ABPA complicating asthma ISHAM working group. Allergic bronchopulmonary aspergillosis: review of literature and proposal of new diagnostic and classification criteria. Clin Exp Allergy. 2013;43(8):850\u0026thinsp;\u0026ndash;\u0026thinsp;73. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1111/cea.12141\u003c/span\u003e\u003cspan address=\"10.1111/cea.12141\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PMID: 23889240.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAgarwal R, Khan A, Gupta D, Aggarwal AN, Saxena AK, Chakrabarti A. An alternate method of classifying allergic bronchopulmonary aspergillosis based on high-attenuation mucus. PLoS ONE. 2010;5(12):e15346. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1371/journal.pone.0015346\u003c/span\u003e\u003cspan address=\"10.1371/journal.pone.0015346\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PMID: 21179536; PMCID: PMC3002283.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAgarwal R, Gupta D, Aggarwal AN, Saxena AK, Chakrabarti A, Jindal SK. Clinical significance of hyperattenuating mucoid impaction in allergic bronchopulmonary aspergillosis: an analysis of 155 patients. Chest. 2007;132(4):1183-90. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1378/chest.07-0808\u003c/span\u003e\u003cspan address=\"10.1378/chest.07-0808\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. Epub 2007 Jul 23. PMID: 17646221.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMonteiro F, Meloni F, Baranauskas JA, Macedo AA. Prediction of mortality in intensive care units: a multivariate feature selection. J Biomed Inf. 2020;107:103456. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.jbi.2020.103456\u003c/span\u003e\u003cspan address=\"10.1016/j.jbi.2020.103456\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://linkinghub.elsevier.com/retrieve/pii/S1532-0464(20)30084-8\u003c/span\u003e\u003cspan address=\"https://linkinghub.elsevier.com/retrieve/pii/S1532-0464(20)30084-8\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. .S1532-0464(20)30084-8 [PubMed] [CrossRef] [Google Scholar].\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKramer AA, Dasta JF, Kane-Gill SL. The impact of mortality on total costs within the ICU. Crit Care Med. 2017;45(9):1457\u0026ndash;63. [PubMed] [CrossRef] [Google Scholar].\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKhurrum M, Asmar S, Joseph B. Telemedicine in the ICU: innovation in the critical care process. J Intensive Care Med. 2021;36(12):1377\u0026ndash;84. [PubMed] [CrossRef] [Google Scholar].\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKang MW, Kim J, Kim DK, Oh K, Joo KW, Kim YS, Han SS. Machine learning algorithm to predict mortality in patients undergoing continuous renal replacement therapy. Crit Care. 2020;24(1):42. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s13054-020-2752-7\u003c/span\u003e\u003cspan address=\"10.1186/s13054-020-2752-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://ccforum.biomedcentral.com/articles/10.1186/s13054-020-2752-7\u003c/span\u003e\u003cspan address=\"https://ccforum.biomedcentral.com/articles/10.1186/s13054-020-2752-7\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. [PMC free article] [PubMed] [CrossRef] [Google Scholar] p.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLiu C, Liu X, Mao Z, Hu P, Li X, Hu J, Hong Q, Geng X, Chi K, Zhou F, Cai G, Chen X, Sun X. Interpretable machine learning model for early prediction of mortality in ICU patients with rhabdomyolysis. Med Sci Sports Exerc. 2021;53(9):1826\u0026ndash;34. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1249/MSS.0000000000002674.00005768-202109000-00004\u003c/span\u003e\u003cspan address=\"10.1249/MSS.0000000000002674.00005768-202109000-00004\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. [PubMed] [CrossRef] [Google Scholar].\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePirracchio R, Petersen ML, Carone M, Rigon MR, van Chevret S. Mortality prediction in intensive care units with the Super ICU Learner Algorithm (SICULA): a population-based study. Lancet Respir Med. 2015;3(1):42\u0026ndash;52. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://europepmc.org/abstract/MED/25466337\u003c/span\u003e\u003cspan address=\"http://europepmc.org/abstract/MED/25466337\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. .S2213-2600(14)70239-5 [PMC free article] [PubMed] [CrossRef] [Google Scholar]. doi: 10.1016/S2213-2600(14)70239-5.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePan P, Li Y, Xiao Y, Han B, Su L, Su M, Li Y, Zhang S, Jiang D, Chen X, Zhou F, Ma L, Bao P, Xie L. Prognostic Assessment of COVID-19 in the intensive care unit by machine learning methods: model development and validation. J Med Internet Res. 2020;22(11):e23128. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.jmir.org/2020/11/e23128/ v22i11e\u003c/span\u003e\u003cspan address=\"https://www.jmir.org/2020/11/e23128/ v22i11e\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. 23128 [PMC free article] [PubMed] [CrossRef] [Google Scholar]. doi: 10.2196/23128.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eStenwig E, Salvi G, Rossi PS, Skj\u0026aelig;rvold NK. Comparative analysis of explainable machine learning prediction models for hospital mortality. BMC Med Res Methodol. 2022;22(1):53. \u003cdiv class=\"ExternalRefDOI\"\u003e10.1186\u003c/div\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://bmcmedresmethodol.biomedcentral.com/articles/10.1186/s12874-022-01540-w\u003c/span\u003e\u003cspan address=\"https://bmcmedresmethodol.biomedcentral.com/articles/10.1186/s12874-022-01540-w\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. [PMC free article] [PubMed] [CrossRef] [Google Scholar]. /s12874-022-01540-w.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWang K, Tian J, Zheng C, Yang H, Ren J, Liu Y, Han Q, Zhang Y. Interpretable prediction of 3-year all-cause mortality in patients with heart failure caused by coronary heart disease based on machine learning and SHAP. \u003cem\u003eComput Biol Med.\u003c/em\u003e 2021;137:104813. doi: 10.1016/j.compbiomed.2021.104813. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://linkinghub.elsevier.com/retrieve/pii/S0010-4825(21)00607-7\u003c/span\u003e\u003cspan address=\"https://linkinghub.elsevier.com/retrieve/pii/S0010-4825(21)00607-7\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. S0010- [PubMed] [CrossRef] [Google Scholar].\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZihni E, Madai VI, Livne M, Galinovic I, Khalil AA, Fiebach JB, Frey D. Opening the black box of artificial intelligence for clinical decision support: a study predicting stroke outcome. PLoS ONE. 2020;15(4):e0231166. \u003cdiv class=\"ExternalRefDOI\"\u003e10.1371\u003c/div\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://dx.plos.org/10.1371/journal.pone.0231166\u003c/span\u003e\u003cspan address=\"https://dx.plos.org/10.1371/journal.pone.0231166\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. .PONE-D-19-31372 [PMC free article] [PubMed] [CrossRef] [Google Scholar]. /journal.pone.0231166.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLundberg SM, Lee SI. A unified approach to interpreting model predictions. NIPS'17: Proceedings of the 31st International Conference on Neural Information Processing Systems; NIPS'17; Dec 4\u0026ndash;9; Long Beach, CA. 2017. pp. 4768\u0026ndash;4777. [Google Scholar].\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eVan Ben C, Laure W, Verbeek Jan FM, Verbakel Jan Y, Christodoulou Evangelia, Vickers Andrew J, Roobol Monique J, Steyerberg Ewout W. Reporting and Interpreting Decision Curve Analysis: A Guide for Investigators. Eur Urol. 2018;74(6):796\u0026ndash;804. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://europepmc.org/abstract/MED/30241973\u003c/span\u003e\u003cspan address=\"https://europepmc.org/abstract/MED/30241973\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. .S0302-2838(18)30640-7 [PMC free article] [PubMed] [CrossRef] [Google Scholar]. doi: 10.1016/j.eururo.2018.08.038.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLee C, Light A, Alaa A, Thurtle D, van der Schaar M, Gnanapragasam VJ. Application of a novel machine learning framework for predicting non-metastatic prostate cancer-specific mortality in men using the Surveillance, Epidemiology, and End Results (SEER) database. Lancet Digit Health. 2021;3(3):e158-e165. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/S2589-7500(20)30314-9\u003c/span\u003e\u003cspan address=\"10.1016/S2589-7500(20)30314-9\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. Epub 2021 Feb 3. PMID: 33549512.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhang C, Jiang Z, Shao C. Clinical characteristics of allergic bronchopulmonary aspergillosis. Clin Respir J. 2020. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1111/crj.13147\u003c/span\u003e\u003cspan address=\"10.1111/crj.13147\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGoto T, Hirayama A, Faridi MK. Obesity and severity of acute exacerbation of chronic obstructive pulmonary disease. Annals Am Thorac Soc. 2018;15(2):184\u0026ndash;91. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1513/AnnalsATS.201706-485OC\u003c/span\u003e\u003cspan address=\"10.1513/AnnalsATS.201706-485OC\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSanchez FF, Faganello MM, Tanni SE. Relationship between disease severity and quality of life in patients with chronic obstructive pulmonary disease. Braz J Med Biol Res. 2008;41(10):860\u0026ndash;5. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1590/S0100-879X2008005000043\u003c/span\u003e\u003cspan address=\"10.1590/S0100-879X2008005000043\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMurugan AT, Sharma G. Obesity and respiratory diseases. Int J Obes. 2008;32(3):564\u0026ndash;72. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/1479972308096978\u003c/span\u003e\u003cspan address=\"10.1177/1479972308096978\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eComes A, Wong AW, Fisher JH, Morisset J, Johannson KA, Farrand E, Fell CD, Kolb M, Manganas H, Cox G, Gershon AS, Halayko AJ, Hambly N, Khalil N, Sadatsafavi M, Shapera S, To T, Wilcox PG, Collard HR, Ryerson CJ. Association of BMI and Change in Weight With Mortality in Patients With Fibrotic Interstitial Lung Disease. Chest. 2022;161(5):1320\u0026ndash;9. Epub 2021 Nov 14. PMID: 34788669.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLu HW, Mao B, Wei P, Jiang S, Wang H. The clinical characteristics and prognosis of ABPA are closely related to the mucus plugs in central bronchiectasis. Clin Respir J. 2020. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1111/crj.13111\u003c/span\u003e\u003cspan address=\"10.1111/crj.13111\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eOkada N, Yamamoto Y, Oguma T, Tanaka J. Allergic bronchopulmonary aspergillosis with atopic, nonatopic, and sans asthma\u0026mdash;Factor analysis. Allergy. 2023. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1111/all.15820\u003c/span\u003e\u003cspan address=\"10.1111/all.15820\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZeng Y, Xue X, Cai H, Zhu G, Zhu M. Clinical characteristics and prognosis of allergic bronchopulmonary aspergillosis: a retrospective cohort study. J Asthma Allergy. 2022. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.2147/JAA.S345427\u003c/span\u003e\u003cspan address=\"10.2147/JAA.S345427\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDe Baets F, De Keyzer L, Van Daele S. Risk factors and impact of allergic bronchopulmonary aspergillosis in Pseudomonas aeruginosa\u0026ndash;negative CF patients. Pediatr Allergy Immunol. 2018. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1111/pai.12953\u003c/span\u003e\u003cspan address=\"10.1111/pai.12953\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWang S, Zhang J, Zhang C, Shao C. Clinical characteristics of allergic bronchopulmonary aspergillosis in patients with and without bronchiectasis. J Asthma Allergy. 2022. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/02770903.2021.1904979\u003c/span\u003e\u003cspan address=\"10.1080/02770903.2021.1904979\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWang X, Lu Y, Sun C, Zhong H, Cai Y. Development and validation of a machine learning-based diagnostic model for identifying nonneutropenic invasive pulmonary aspergillosis in suspected patients. Microbiol Spectr. 2025;e00607\u0026ndash;25. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1128/spectrum.00607-25\u003c/span\u003e\u003cspan address=\"10.1128/spectrum.00607-25\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDu W, Ji W, Luo T, Zhang Y, Guo W. Development of a prognostic nomogram for nonneutropenic invasive pulmonary aspergillosis based on machine learning. J Inflamm Res. 2024;17:365\u0026ndash;77. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.2147/JIR.S499008\u003c/span\u003e\u003cspan address=\"10.2147/JIR.S499008\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJeon ET, Lee HJ, Park TY, Jin KN, Ryu B, Lee HW. (2023). Machine learning-based prediction of in-ICU mortality in pneumonia patients. Sci Rep, 13, Article 11023. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.nature.com/articles/s41598-023-38765-8\u003c/span\u003e\u003cspan address=\"https://www.nature.com/articles/s41598-023-38765-8\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWang B, Li Y, Tian Y, Ju C, Xu X, Pei S. Novel pneumonia score based on a machine learning model for predicting mortality in pneumonia patients on admission to the intensive care unit. Comput Biol Med. 2023;162:107238. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.sciencedirect.com/science/article/pii/S0954611123002512\u003c/span\u003e\u003cspan address=\"https://www.sciencedirect.com/science/article/pii/S0954611123002512\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChia AHT, Khoo MS, Lim AZ, Ong KE, Sun Y. (2021). Explainable machine learning.\u003c/span\u003e\u003cspan\u003eprediction of ICU mortality. Heliyon, 7(8), e07714. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.sciencedirect.com/science/article/pii/S235291482100159\u003c/span\u003e\u003cspan address=\"https://www.sciencedirect.com/science/article/pii/S235291482100159\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003eTable 1 \u0026zwnj;All predictor variables for patients with ABPA\u003c/p\u003e\n\u003cdiv class=\"gridtable\"\u003e\n \u003ctable id=\"Taba\" border=\"1\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eCharacteristic\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eAll-cause mortality\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eSurvival\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ep value\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTotal N\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e38\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e44\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBaseline variables and in-hospital factors\u003c/p\u003e\n \u003cp\u003eAge(years)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e68.82\u0026thinsp;\u0026plusmn;\u0026thinsp;11.69\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e74.07\u0026thinsp;\u0026plusmn;\u0026thinsp;11.12\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.040\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBMI(kg/m\u003csup\u003e2\u003c/sup\u003e ),\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e23.89\u0026thinsp;\u0026plusmn;\u0026thinsp;2.80\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e25.05\u0026thinsp;\u0026plusmn;\u0026thinsp;3.04\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.078\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eshock\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.00 (0.00\u0026ndash;1.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.00 (0.00\u0026ndash;1.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.102\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003epeak leukocyte count(10\u003csup\u003e9\u003c/sup\u003e/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e17.91 (14.75\u0026ndash;32.52)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e15.54 (11.02\u0026ndash;22.93)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.045\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003epeak eosinophil count(10\u003csup\u003e9\u003c/sup\u003e/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.19 (0.01\u0026ndash;0.54)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.33 (0.15\u0026ndash;0.56)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.129\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003elowest hemoglobin concentration(g/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e64.00 (52.25-77.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e71.00 (58.00-82.50)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.154\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003elowest platelet count(10\u003csup\u003e9\u003c/sup\u003e/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e49.50 (32.00-102.50)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e93.50 (26.00-147.75)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.076\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003epeak procalcitonin level(ng/ml)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e11.11 (1.92\u0026ndash;28.92)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.97 (0.88\u0026ndash;6.90)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003epeak C-reactive protein (CRP) level(mg/dl)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e29.43 (10.78\u0026ndash;33.06)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e15.29 (10.13\u0026ndash;27.54)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.102\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003epeak blood glucose level(mmol/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e13.20 (12.03\u0026ndash;17.88)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e12.89 (9.90\u0026ndash;18.50)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.802\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003egender\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.429\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMale\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e32 (84.21%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e34 (77.27%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eFemale\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e6 (15.79%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e10 (22.73%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eComorbidities\u003c/p\u003e\n \u003cp\u003easthma history\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.054\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e25 (65.79%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e37 (84.09%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eY\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e13 (34.21%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e7 (15.91%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ehistory of corticosteroid use\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.005\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e28 (73.68%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e42 (95.45%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eY\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e10 (26.32%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2 (4.55%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003esmoking history\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.852\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e35 (92.11%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e41 (93.18%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eY\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e3 (7.89%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e3 (6.82%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eallergic rhinitis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.057\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e38 (100.00%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e40 (90.91%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eY\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0 (0.00%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e4 (9.09%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003efood and drug allergies\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.203\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e36 (94.74%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e38 (86.36%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eY\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2 (5.26%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e6 (13.64%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ediabetes mellitus\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.356\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e27 (71.05%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e27 (61.36%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eY\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e11 (28.95%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e17 (38.64%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003econcurrent bacterial infection\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.576\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e17 (44.74%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e17 (38.64%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eY\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e21 (55.26%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e27 (61.36%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003econcurrent viral infection\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.375\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e28 (73.68%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e36 (81.82%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eY\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e10 (26.32%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e8 (18.18%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003erenal insufficiency or failure\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e6 (15.79%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e24 (54.55%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eY\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e32 (84.21%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e20 (45.45%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ehepatic insufficiency or failure\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.165\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e28 (73.68%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e26 (59.09%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eY\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e10 (26.32%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e18 (40.91%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003epulmonary fibrosis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.010\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e26 (68.42%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e40 (90.91%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eY\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e12 (31.58%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e4 (9.09%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003epulmonary cavities or bronchiectasis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.017\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e15 (39.47%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e29 (65.91%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eY\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e23 (60.53%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e15 (34.09%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003epulmonary infiltration\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.231\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e14 (36.84%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e22 (50.00%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eY\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e24 (63.16%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e22 (50.00%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Allergic bronchopulmonary aspergillosis (ABPA), XGBoost, machine learning, predictive model","lastPublishedDoi":"10.21203/rs.3.rs-7822617/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7822617/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e\u003cp\u003eAllergic bronchopulmonary aspergillosis (ABPA) is a hypersensitivity lung disease caused by Aspergillus infection, with severe cases often requiring admission to the intensive care unit (ICU). Early prediction of in-hospital mortality in ICU ABPA patients is crucial for optimizing clinical decision-making and resource allocation.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e\u003cp\u003eThis retrospective study collected clinical data from ICU patients diagnosed with ABPA at Yuebei People's Hospital between January 2020 and July 2024. An in-hospital mortality prediction model was developed using an explainable XGBoost machine learning algorithm. SHapley Additive Explanations (SHAP) was employed to interpret key predictive factors, and internal validation was conducted to assess model performance.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e\u003cp\u003eA total of 82 ICU ABPA patients were included, with mortality rates of 46.3% (26/57) in the training set and 48% (12/25) in the validation set. The XGBoost model demonstrated excellent predictive performance, achieving areas under the receiver operating characteristic (ROC) curve (AUC) of 0.995 (95% CI: 0.903\u0026ndash;1.000) in the training set and 0.881 (95% CI: 0.846\u0026ndash;0.909) in the validation set. SHAP analysis identified key predictors of mortality, including BMI, peak procalcitonin level, peak eosinophil count, age, asthma history, peak leukocyte count, and lowest platelet count.\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e\u003cp\u003eThe XGBoost model effectively predicts in-hospital mortality in ICU ABPA patients and provides interpretable results using SHAP analysis. Although the model performed well in internal validation, external validation is needed to enhance its generalizability. Future multicenter studies and integration of dynamic biomarkers are recommended to optimize predictive accuracy and support individualized clinical decision-making.\u003c/p\u003e","manuscriptTitle":"Predicting Mortality in Intensive Care Unit Patients with Allergic Bronchopulmonary Aspergillosis (ABPA) Using an Interpretable Machine Learning Model: A Retrospective Cohort Study","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-11-19 12:00:32","doi":"10.21203/rs.3.rs-7822617/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"dd0c76c7-65af-44b7-a03f-aaf3842290af","owner":[],"postedDate":"November 19th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-03-18T10:10:36+00:00","versionOfRecord":[],"versionCreatedAt":"2025-11-19 12:00:32","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7822617","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7822617","identity":"rs-7822617","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}