Development and Application of an Early Prediction Model for Risk of Bloodstream Infection based on Real-world Study

Development and Application of an Early Prediction Model for Risk of Bloodstream Infection based on Real-world 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 Development and Application of an Early Prediction Model for Risk of Bloodstream Infection based on Real-world Study Xiefei Hu, Shenshen Zhi, Yang Li, Yuming Cheng, Haiping Fan, Haorong Li, and 4 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5859635/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 14 May, 2025 Read the published version in BMC Medical Informatics and Decision Making → Version 1 posted 4 You are reading this latest preprint version Abstract Background Bloodstream Infection (BSI) is a severe systemic infectious disease that can lead to sepsis and Multiple Organ Dysfunction Syndrome (MODS), resulting in high mortality rates and posing a major public health burden globally. Early identification of BSI is crucial for effective intervention, reducing mortality, and improving patient outcomes. However, existing diagnostic methods are flawed by low specificity, long detection times and high demands on testing platforms. The development of artificial intelligence provides a new approach for early disease identification. This study aims to explore the optimal combination of routine laboratory data and clinical monitoring indicators, and to utilize machine learning algorithms to construct an early, rapid, and universally applicable BSI risk prediction model, to assist in the early diagnosis of BSI in clinical practice. Methods Clinical data of 2582 suspected BSI patients admitted to the Chongqing University Central Hospital, from January 1, 2021 to December 31, 2023 were collected for this study. The data were divided into a modeling dataset and an external validation dataset based on chronological order, while the modeling dataset was further divided into a training set and an internal validation set. The occurrence rate of BSI, distribution of pathogens, and microbial primary reporting time were analyzed within the training set. During the feature selection stage, univariate regression and ML algorithms were applied. First, Univariate logistic regression was used to screen for predictive factors of BSI. Then, the Boruta algorithm, Lasso regression, and Recursive Feature Elimination with Cross-validation (RFE-CV) were employed to determine the optimal combination of predictors for predicting BSI. Based on the optimal combination, six machine learning algorithms were used to construct an early BSI risk prediction model. The best model was selected by models’ performance, and the Shapley Additive Explanations (SHAP) method was used to explain the model. The external validation set was used to evaluate the predictive performance and generalizability of the selected model, and the research findings were ultimately applied in clinical practice. Results The incidence of BSI among inpatients at the Chongqing University Central Hospital was 12.91%. Following further feature selection, a set of 5 variables was determined, including white blood cell count, standard bicarbonate, base excess of extracellular fluid, interleukin-6, and body temperature. BSI early risk prediction models were constructed using six machine learning algorithms, with the XGBoost model demonstrating the best performance, achieving an AUC value of 0.782 in the internal validation set and an AUC value of 0.776 in the external validation set. This model is made publicly available as an online webpage tool for clinical use. Conclusions This study successfully identified a set of 5 features by analyzing routine laboratory data clinical monitoring indicators among hospitalized patients. Based on this set, a machine learning-based early risk prediction model for BSI was constructed. The model is capable of early and rapid differentiation between BSI and non-BSI patients. The inclusion of minimal risk prediction factors enhances its applicability in clinical settings, particularly at the primary care level. To further improve the model’s real-world applicability and more convenient for clinical use, the online application of the model could greatly improve the efficiency of BSI diagnosis and reducing patients’ mortality. Bloodstream Infection Risk Prediction Real-world Model Construction Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction Bloodstream Infection (BSI) refers to the invasion of microorganisms into the bloodstream, leading to a systemic infection, which can cause damage to all organs of the body. It is prone to inducing sepsis and multiple organ dysfunction syndrome (MODS), and associated with a high mortality rate [1, 2] . Annually, approximately 1.2 million patients are diagnosed with BSI in Europe [3] . Every hour of delayed treatment for BSI raises the mortality rate by 8%, reaching 58% after a 6-hour delay [4] . In the ICU, factors like weakened immunity, frequent risky procedures, multiple complications, and extended hospital stays heighten the risk of BSI, making it a common issue in these settings. Untreated BSI can rapidly lead to sepsis, progressing to MODS, causing poor outcomes and life-threatening conditions [5–7] . Early detection, appropriate antibiotics, and addressing the source of BSI greatly reduce morbidity and mortality rates [8] . Blood culture is the benchmark for diagnosing BSI but has limitations such as low positivity rates, long turnaround times, contamination risks, and challenges in detecting certain pathogens with standard culturing [9] . Machine learning (ML), a vital part of artificial intelligence, has advanced analytical powers that can independently detect disease patterns in data and forecast unknown results [10,,11] . Compared to traditional diagnostic and therapeutic approaches, ML offers a deeper insight into complex relationships. In recent years, ML has shown significant promise in disease screening, diagnosis, prognosis prediction, and risk analysis [12] . Developing early prediction models for BSI using ML is crucial for enhancing early diagnosis, treatment, and personalized healthcare. However, current prediction models often require a large number of features [13–15] . While including more features can improve the predictive ability of the models, it can also lead to increased complexity, requiring more data for training, and reducing interpretability and generalizability of the model. This poses a challenge in practical clinical settings, especially in primary care facilities where extensive testing and comprehensive patient data collection may not be feasible. Therefore, researchers and clinicians need to find a balance—ensuring predictive accuracy while minimizing the number and complexity of required features—to make these predictive models effective in resource-limited environments, such as grassroots healthcare institutions. Consequently, this study aims to analyze routine laboratory/clinical data to identify key predictive factors that play a significant role in the early diagnosis of BSI. The goal is to find the optimal combination of these factors and use machine learning algorithm to develop a broadly applicable early risk prediction model for BSI. This model aims to facilitate early and rapid prediction of BSI in a variety of clinical settings and will be validated and implemented in real-world scenarios. Methods Study Population This study was a secondary analysis of a retrospective observational study conducted from 2021 to 2023 among inpatients at the Chongqing University Central Hospital. The inclusion criteria were (1)age ≥ 18 years; (2)inpatients; (3)had at least one blood culture examination performed during hospital stay. The exclusion criteria were (1)The blood culture results indicated a probable contaminant; (2)Data missing rate ≥ 30%. Clinical or laboratory parameters related to BSI were collected for each patient. For patients with multiple positive BC samples, only the first episode was included. For those with multiple negative BC samples, a single episode was randomly selected. Outcome The outcome assessed was BSI, defined as the growth of a clinically significant pathogen in at least one BC bottle. Potential contaminants were defined by the Center for Disease Control and Prevention (CDC)/National Health Safety Network (NHSN) guidelines for Laboratory Confirmed Bloodstream Infection (LCBI) and were not classified as BSIs. These potential contaminants include coagulase-negative Staphylococci, Corynebacterium species, Bacillus species, Diphtheroids, Aerococcus, and Propionibacterium species [13] . Dataset At the target medical centers, we constructed datasets that included demographics, clinical and laboratory parameters, including microbiology, available within 3 hours before and after BC sampling time.The dataset included as follows: (i) blood cells; (ii) liver function; (iii) renal function; (iv) hemagglutination; (v) blood gas analysis; (vi) electrolytes; (vii) inflammatory markers; (viii) blood culture; (ix) clinical features. We collected datasets from two time periods: the dataset from January 2021 to April 2023 was randomly split into training and validation sets comprising 70% and 30% respectively. The training sets were used for modelling, while the validation sets for internal validation. The dataset from May 2023 to December 2023 was used for external validation of the best model. Data preprocessing Data cleaning and preprocessing are critical steps in the data analysis process, aimed at transforming raw data into a format suitable for statistical analysis or ML modeling [16,17] . In this study, data cleaning and preprocessing primarily involved the removal of duplicate data, analysis and treatment of outliers, imputation of missing values, data standardization, and balancing of data categories. Feature selection and modeling Feature selection: (i) In this study, the initial method for selecting predictive factors involved univariable logistic regression. Univariable logistic regression allowed for the assessment of whether each biomarker was independently associated with BSIs, thus enabling the preliminary selection of predictive factors for model development. (ii) The study also incorporated the Boruta algorithm [18–20] , Lasso regression [21–22] , and Recursive Feature Elimination with Cross-validation (RFE-CV) [23–25] to optimize the results obtained from the univariable logistic regression analysis. Modeling: In this study, we used the Light Gradient Boosting Machine(LightGBM)、eXtreme Gradient Boosting (XGBoost)、Gradient Boosting Decision Tree (GBDT)、Random Forest (RF)、Support Vector Machine (SVM) and Gaussian Naive Bayes (GNB) algorithms to predict the risk of BSI in inpatients by analyzing clinic/laboratory data [26–31] . Throughout the model development phase, we implemented a grid search technique to refine the hyperparameters. Validation and Explanation We evaluated the performance of the model by applying several different indices, namely (i) AUC, (ii) accuracy, (iii) sensitivity, and (iv) specificity. The performance assessment for selecting the best model will primarily be based on the AUC value. First, we conducted an assessment on the internal validation set, which comprised 30% of the original data that was initially set aside for validation purposes only. After model selection, we used the Shapley Additive Explanations (SHAP) algorithm from model-agnostic approaches to explain the best-performing model [32–34] . Finally, the dataset from May 2023 to December 2023 was utilized for external validation of the optimal model. Statistical Analysis Binary variables were presented as counts and percentages, and their significance was assessed using the Chi-square test or Fisher's exact test. Continuous variables that were normally distributed were compared with a t-test and reported as means ± SEM. For variables with a non-normal distribution, the Mann–Whitney U test was applied. A P-value of less than 0.05 was deemed statistically significant. All statistical analyses were conducted in the Beckman Coulter DxAI platform ( https://www.xsmartanalysis.com/beckman/login/ ). Results Patient characteristics Our model construction database initially contained 5,057 inpatients. Following a series of exclusions, 2,323 adult inpatients were included in this study, of which 300 patients developed BSI, representing 12.9% of the study population. The patient selection process is illustrated in Fig. 1 . The baseline characteristics of the patients are presented in Supplementary Table 1.The training and internal validation datasets comprised of 1,626 and 697 patients, respectively. A total of 74 variables, including age, sex, Temperature, White Blood Cell Count (WBC), D-dimer, and other laboratory or clinical parameters related to BSI, were collected for each patient. A comparison of basic information between the two sets were shown in Supplementary Table 2. For external validation of the model, 259 patients were included, of whom 34 developed BSI (13.13%). The baseline characteristics of the patients are presented in Supplementary Table 3. Variables of importance The model's accuracy increased as more variables were incorporated. However, increasing the number of variables did not correspond with the practicality needed for clinical application. In order to indentify the most significant features, we employed univariate logistic regression to preliminarily screen the variables associated with BSI within the training set.We identified 27 variables that are crucial for predicting BSI, which were shown in Supplementary Table 4 Based on the results of the univariate logistic regression analysis, the individual indicators that were screened (WBC, EOS, EOS%, Neu%, Mon, Mon%, RDW, Hct, PLT, A/G, Alb, CHE, PA, Cr, UA, Urea, Fib, SB, AB, BEf, Lac, TCO2, Cl, Mg, IL-6, hs-CRP, and T) were used separately to predict whether patients had BSI. As shown in Fig. 2, the AUC values for Neu%, Cr, Urea, and T exceeded 0.600, while the AUC values for the remaining indicators were all below 0.600. The efficacy of single indicators for predicting BSI was poor. FIGURE 2. The ROC Curves of Predictive Factors Identified by Univariate Logistic Regression Analysis We utilized the Boruta algorithm, Lasso regression, and RFE-CV to further reduce the number of variables. As shown in Fig. 3, the Boruta algorithm indentified 19 variables such as Hct, Fib, UA, Cl, Alb, hs-CRP, WBC, TCO2, Urea, AB, Cr, Mg, Mon, IL-6, Mon%, BEf, SB, Neu%, and T. The Lasso regression analysis highlighted 15 features that help minimize the model's prediction error: WBC, EOS, PLT, PA, Lac, UA, TCO2, AB, SB, BEf, Na, Cl, hs-CRP, IL-6, and T. Meanwhile, the RFE-CV method selected the top five feature indicators based on their contribution rankings, which are WBC, SB, BEf, IL-6, and T. Ultimately, by taking the intersection of the results from these three algorithms, we identified the 5 key features that contribute the most to the model's predictive capability: WBC, SB, BEf, IL-6, and T. FIGURE 3 Selection of key features for BSI. a) Variable Selection Plot of Boruta; b) Variable Selection Plot of Lasso; c) Variable Selection Plot of RFE-CV; d)Venn graph displaying 5 features shared by Boruta, Lasso and RFE-CV. Classification Results Based on the selected 5 key features (WBC, SB, BEf, IL-6, and T), we constructed six early prediction models for BSI risk using machine learning algorithms: the LightGBM model, the XGBoost model, the GBDT model, the RF model, the SVM model, and the GNB model. The model construction process involved hyperparameter optimization using grid search techniques. As shown in Fig. 4, the average AUC values for the XGBoost, LightGBM, RF, GBDT, GNB, and SVM models on the internal validation set were 0.782 (95% CI: 0.715–0.849), 0.700 (95% CI: 0.627–0.773), 0.772 (95% CI: 0.704–0.841), 0.723 (95% CI: 0.650–0.797), 0.562 (95% CI: 0.483–0.642), and 0.528 (95% CI: 0.446–0.611), respectively. The XGBoost model had the highest AUC value of 0.782, while the SVM model had the lowest AUC value of 0.528. , FIGURE 4 ROC Curves of Six Models in the Internal Validation Set As shown in Table 1 , the RF model had the highest accuracy rate at 0.882; the GNB model had the highest sensitivity at 0.747; and the XGBoost model had the highest specificity at 0.824. Considering the AUC values and the evaluation metrics, the XGBoost model emerged as the best model. Table 1 Evaluation Metrics Results of Six Models Models Accuracy Sensitivity Specificity XGBoost 0.763 0.633 0.824 LightGBM 0.830 0.527 0.788 RF 0.882 0.653 0.799 GBDT 0.729 0.643 0.738 GNB 0.470 0.747 0.416 SVM 0.595 0.510 0.642 In the external validation, the AUROC of the XGBoost model decreased to 0.776, with an accuracy of 0.685, sensitivity of 0.647, and specificity of 0.800. The calibration curve was close to the 45° line, indicating a good fit between the model's predictions and the actual values. The results are shown in Fig. 5. FIGURE 5 Performance Evaluation of the XGBoost Model. a) ROC Curve of External Validation Set in the XGBoost Model; b) calibration curve of XGBoost Model Model Interpretation and Online Application To better understand the prediction results of the XGBoost model and the basis for decision-making, the SHAP algorithm was used to quantify the contribution of each feature to the model's predictive outcomes. Figure 6a displays the ranking of feature contributions in the XGBoost model, with the indicators ranked from highest to lowest contribution being SB, BEf, IL-6, T, and WBC. For individual patients, as shown in Fig. 6b and c ,the figure uses color coding to represent the impact of features on the prediction. Blue indicates features that negatively influence the prediction (leftward arrows, which correspond to a decrease in SHAP values), and red signifies features that positively affect the prediction (rightward arrows, indicating an increase in SHAP values). The base value represents the average model output for the training set, and the SHAP values for an individual patient's model output are indicated by f(x). In Fig. 6b, the f(x) value is below the base value (0.03 compared to 0.20), which suggests the model predicts a low risk of BSI for this patient. In contrast, in Fig. 6c, the f(x) value exceeds the base value (0.62 compared to 0.20), leading the model to predict a high risk of BSI for the patient. FIGURE 6 Model Interpretation of XGBoost. a) Importance Ranking of Features; b) Example of Low-risk Patient; c) Example of Hight-risk Patient To enhance the practicality and broad applicability of the constructed model in clinical practice, early risk prediction for patients can be conducted via an online link. The URL for the online prediction tool is: [ http://www.xsmartanalysis.com/model/list/predict/model/html?mid=13885&symbol=11im71SWNC211Qj91806](http://www.xsmartanalysis.com/model/list/predict/model/html?mid=13885&symbol=11im71SWNC211Qj91806) . Discussion The study findings revealed a 12.91% incidence rate of BSI among hospitalized patients from 2021 to 2023. This rate exceeded that of a 6-year U.S. retrospective study (12.91% vs. 5.90%) [35] , likely due to the study hospital's role as a national critical care center, which saw a higher volume of critically ill patients prone to BSI. The period of data collection had coincided with the COVID-19 pandemic, potentially contributing to the elevated BSI rates [36] . Finnish research had indicated higher BSI incidence and mortality in the elderly, particularly those over 80 [37] . Our study population had an older median age (68.0 years for the cohort, 72.0 years for BSI cases), which may have explained the higher incidence. With an aging population, addressing BSI in elderly patients was crucial. It was important to note that pre-admission BSI cases had not been excluded, possibly inflating the incidence rate with community-acquired BSI. Early diagnosis of BSI is vital for lowering mortality and enhancing patient outcomes. As artificial intelligence evolves, Machine Learning (ML) algorithms are becoming pivotal in medicine, particularly for BSI diagnosis. Studies like Roimi's achieved an AUC of 0.930 with 50 features [13] , Zhang's LSTM model reached 0.892 with over 100 features [15] , and Zoabi et al. reported 0.810 with 25 features [38] . While more features can improve model performance, extensive data collection complicates practical use, especially in primary care where early BSI diagnosis is challenging. This study initially narrowed 74 predictors to 27 via univariate logistic regression, but single-factor prediction was inadequate. Further analysis led to feature selection using ML methods, including Boruta, Lasso, and RFE-CV, pinpointing 5 key ,indicators for early BSI risk: SB, BEf, IL-6, T, and WBC. WBC [40–41] , IL-6 [42–44] , and T [45,46] are standard in infectious disease management and are key in BSI diagnosis. Blood gas analysis, often focusing on TCO2 and pH, has seen less research on SB and BEf for early BSI detection. Some studies have indicated that during the early stages of infectious diseases, more pronounced changes occur in SB and BEF. Research suggests that the systemic inflammatory response induced by infection can impair the normal function of the circulatory system, thereby affecting tissue oxygenation. Even when the blood pH of patients has not shown significant fluctuations, the SB level begins to decline in the context of hypoxia [47] . When BSI patients experience acid-base balance disorders, BEF exhibits marked abnormalities. Song-Mao Ouyang and colleagues have observed statistically significant differences in BEF values between infected and non-infected patients (P < 0.05) [48] . The XGBoost model outperformed others with an AUC of 0.782 and high specificity, aligning closely with the 45° line in calibration curves. It was chosen as the optimal model for further external validation and clinical use.External validation is key to evaluating a model's performance and generalizability. It ensures accuracy and reliability [49] . The XGBoost model, tested on a new dataset, achieved an AUC of 0.776 for BSI prediction, demonstrating robust predictive ability. The calibration curve showed a close match between predictions and actual results. For clinical use, the model is accessible online, allowing clinicians to input WBC, SB, T, BEf, and IL-6 values to receive BSI risk predictions. This study's goal is to create an early BSI risk prediction model using standard, affordable, and easy-to-administer lab tests. We aimed to pinpoint key tests for early BSI prediction to streamline clinical diagnostics. Machine Learning was employed to develop the prediction model, which is designed for easy use in various healthcare settings. Our analysis revealed five key predictors: WBC, SB, T, BEf, and IL-6. While WBC, IL-6, and T are standard infection markers, SB and BEf's role in BSI prediction is underexplored. Our model underscores the importance of these latter two indicators, suggesting they deserve more research attention. The study has two key limitations. First, all patient data was sourced from one institution, potentially leading to selection bias. The model, developed with a focus on critical patients at Chongqing University Affiliated Central Hospital, may not generalize to other patient groups. Future research should use multi-center data to enhance the model's universality and reliability. Second, the model, available online, lacks prospective clinical validation due to BSI's low incidence. Subsequent research should validate the model with real-world clinical data across various populations, times, and settings to confirm its predictive value for BSI and its utility in clinical decisions. Abbreviations BSI Bloodstream Infection MODS Multiple Organ Dysfunction Syndrome RFE-CV Recursive Feature Elimination with Cross-validation SHAP Shapley Additive Explanations ML Machine learning CDC Disease Control and Prevention NHSN National Health Safety Network LCBI Laboratory Confirmed Bloodstream Infection LightGBM Light Gradient Boosting XGBoost Extreme Gradient Boosting GBDT Gradient Boosting Decision Tree RF Random Forest SVM Support Vector Machine GNB Gaussian Naive Bayes MCHC Mean Corpuscular Hemoglobin Concentration RDW Red Cell Distribution Width MCH Mean Corpuscular Hemoglobin MCV Mean Corpuscular Volume PLT Platelet Count WBC White Blood Cell Count Neu Neutrophil Count Neu% Neutrophils Percentage Eos Direct Eosinophil Count Eos% Eosinophils Percentage Mon Monocyte Count Mon% Monocytes Percentage Baso Direct Basophil Count Baso% Basophils Percentage Lym Lymphocyte Count Lym% Lymphocyte Percentage RBC Red Blood Cell Count Hb Hemoglobin Hct Hematocrit Plateletcrit Plateletcrit AFU α-fucosidase ALP Alkaline Phosphatase ALT Alanine Aminotransferase Alb Albumin CHE Cholinesterase GCA Glycocholic Acid LDH Lactate Dehydrogenase PA Prealbumin TBA Total Bile Acids TP Total Protein TBiL Total Bilirubin A/G Alb/Glob Ratio 5-n 5-nucleotidase GGT Gamma-glutamyl Transferase Cr Creatinine Cys-C Cystatin C UA Uric Acid α1-MG α1-microglobulin β2-MG β2-microglobulin PT Prothrombin Time APTT Activated Partial Thromboplastin Time Fib Fibrinogen TT Thrombin Time INR International Normalized Ratio D-D D-dimer PT% Prothrombin Activity AB Actual Bicarbonate AG Anion Gap BEf Base Excess of Extracellular Fluid BOP Blood Osmotic Pressure FiO2 Fraction of Inspired Oxygen PCO2 Partial Pressure of Carbon Dioxide pH Pondus Hydrogeni SB Standard Bicarbonate TCO2 Total Carbon Dioxide Partial Pressure Lac Lactic Acid PO2 Oxygen Partial Pressure Na Sodium K Potassium Cl Chlorine Mg Magnesium P Phosphorus PCT Procalcitonin IL-6 Interleukin-6 hs-CRP High-sensitive C-reactive Protein Declarations Acknowledgements Not applicable. Funding This work was supported by the Science and Technology Research Project of the Chongqing Municipal Education Commission (grant number: KJZD-M202300101), the Emergency Medicine Chongqing Key Laboratory Talent Development Innovation Joint Fund Project (grant number: 2024RCCX06) and the Wu Jieping Medical Foundation (grant number: 320.6750.2024-23-1 1). The statements made herein are solely the responsibility of the authors. Disclosure of potential conflicts of interest The authors declare that they have no confict of interest. Contributions Xiefei Hu: Conceptualization, Data Curation, Methodology, Software, Writing- Original draft preparation, Writing- Reviewing and Editing. Shenshen Zhi: Conceptualization, case data collection and article design. Yuming Chen and Yang Li: Conceptualization, Methodology. Haiping Fan, Haorong Li, Zihao Meng and Jiaxin Xie: Data Curation, Methodology, Software. Shu Tang and Wei Li: Overall planning. All authors contributed to the article and approved the submitted version. Corresponding author Correspondence to Shu Tang and Wei Li: Ethics declarations Ethics approval and consent to participate The studies involving humans were approved by the Ethics Committee of Chongqing Emergency Medical Center and Chongqing University Central Hospital (Approval Ethics Review No.RS202410). The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study Consent for the publication Not applicable. Competing interests The authors declare no competing interests. References Lamy B, Sundqvist M, Idelevich E A. Bloodstream infections - Standard and progress in pathogen diagnostics[J]. Clinical microbiology and infection, 2020, 26(2): 142-150. Shanghai Society for Microbiology, Clinical Microbiology Professional Committee, Shanghai Medical Association, Critical Care Medicine Specialty Branch, Shanghai Medical Association, Critical Care Medicine Specialty Branch. Expert Consensus on Clinical Laboratory Testing Pathways for Bloodstream Infections [J]. Chinese Journal of Infectious Diseases, 2022, 40(08): 457-475. Vincent J L, Sakr Y, Singer M, et al. Prevalence and Outcomes of Infection Among Patients in Intensive Care Units in 2017[J]. Jama, 2020, 323(15): 1478-1487. Lin K, Zhang H C, Zhao Y H, et al. The direct application of plasma droplet digital PCR in the ultra-early pathogen detection and warning during sepsis: Case reports[J]. Journal of infection and public health, 2022, 15(4): 450-454. Rudd K E, Johnson S C, Agesa K M, et al. Global, regional, and national sepsis incidence and mortality, 1990-2017: analysis for the Global Burden of Disease Study[J]. Lancet (London, England), 2020, 395(10219): 200-211. Xie J, Wang H, Kang Y, et al. The Epidemiology of Sepsis in Chinese ICUs: A National Cross-Sectional Survey[J]. Critical care medicine, 2020, 48(3): e209-e18. Overbeek R, Leitl C J, Stoll S E, et al. The Value of Next-Generation Sequencing in Diagnosis and Therapy of Critically Ill Patients with Suspected Bloodstream Infections: A Retrospective Cohort Study[J]. Journal of clinical medicine, 2024, 13(2). Schenz J, Weigand M A, Uhle F. Molecular and biomarker-based diagnostics in early sepsis: current challenges and future perspectives[J]. Expert review of molecular diagnostics, 2019, 19(12): 1069-1078. Warren B G, Yarrington M E, Polage C R, et al. Evaluation of hospital blood culture utilization rates to identify opportunities for diagnostic stewardship[J]. Infection control and hospital epidemiology, 2023, 44(2): 200-205. Swanson K, Wu E, Zhang A, et al. From patterns to patients: Advances in clinical machine learning for cancer diagnosis, prognosis, and treatment[J]. Cell, 2023, 186(8): 1772-1791. Peiffer-Smadja N, Rawson T M, Ahmad R, et al. Machine learning for clinical decision support in infectious diseases: a narrative review of current applications[J]. Clinical microbiology and infection, 2020, 26(5): 584-595. Shibue K. Artificial Intelligence and Machine Learning in Clinical Medicine[J]. The New England journal of medicine, 2023, 388(25): 2398. Roimi M, Neuberger A, Shrot A, et al. Early diagnosis of bloodstream infections in the intensive care unit using machine-learning algorithms[J]. Intensive care medicine, 2020, 46(3): 454-462. Li X, Xu X, Xie F, et al. A Time-Phased Machine Learning Model for Real-Time Prediction of Sepsis in Critical Care[J]. Critical care medicine, 2020, 48(10): e884-e888. Zhang D, Yin C, Hunold K M, et al. An interpretable deep-learning model for early prediction of sepsis in the emergency department[J]. Patterns (New York, NY), 2021, 2(2): 100196. Chicco D, Oneto L, Tavazzi E. Eleven quick tips for data cleaning and feature engineering[J]. PLoS computational biology, 2022, 18(12): e1010718. Maletzky A, Böck C, Tschoellitsch T, et al. Lifting Hospital Electronic Health Record Data Treasures: Challenges and Opportunities[J]. JMIR medical informatics, 2022, 10(10): e38557. Yue S, Li S, Huang X, et al. Machine learning for the prediction of acute kidney injury in patients with sepsis[J]. Journal of translational medicine, 2022, 20(1): 215. Zhou H, Xin Y, Li S. A diabetes prediction model based on Boruta feature selection and ensemble learning[J]. BMC bioinformatics, 2023, 24(1): 224. Kong C, Zhu Y, Xie X, et al. Six potential biomarkers in septic shock: a deep bioinformatics and prospective observational study[J]. Frontiers in immunology, 2023, 14: 1184700. Zhou T, Ren Z, Ma Y, et al. Early identification of bloodstream infection in hemodialysis patients by machine learning[J]. Heliyon, 2023, 9(7): e18263. Tang G, Qi L, Sun Z, et al. Evaluation and analysis of incidence and risk factors of lower extremity venous thrombosis after urologic surgeries: A prospective two-center cohort study using LASSO-logistic regression[J]. International journal of surgery (London, England), 2021, 89: 105948. Han Y, Huang L, Zhou F. A dynamic recursive feature elimination framework (dRFE) to further refine a set of OMIC biomarkers[J]. Bioinformatics (Oxford, England), 2021, 37(15): 2183-2189. Al Abir F, Shovan S M, Hasan M A M, et al. Biomarker identification by reversing the learning mechanism of an autoencoder and recursive feature elimination[J]. Molecular omics, 2022, 18(7): 652-661. Zhang Z, Wang S, Zhu Z, et al. Identification of potential feature genes in non-alcoholic fatty liver disease using bioinformatics analysis and machine learning strategies[J]. Computers in biology and medicine, 2023, 157: 106724. Amin M N, Salami B A, Zahid M, et al. Investigating the Bond Strength of FRP Laminates with Concrete Using LIGHT GBM and SHAPASH Analysis[J]. Polymers, 2022, 14(21). Moore A, Bell M. XGBoost, A Novel Explainable AI Technique, in the Prediction of Myocardial Infarction: A UK Biobank Cohort Study[J]. Clinical Medicine Insights Cardiology, 2022, 16: 11795468221133611. Seto H, Oyama A, Kitora S, et al. Gradient boosting decision tree becomes more reliable than logistic regression in predicting probability for diabetes with big data[J]. Scientific reports, 2022, 12(1): 15889. Blanchet L, Vitale R, van Vorstenbosch R, et al. Constructing bi-plots for random forest: Tutorial[J]. Analytica chimica acta, 2020, 1131: 146-155. Hao P Y, Chiang J H, Chen Y D. Possibilistic classification by support vector networks[J]. Neural networks, 2022, 149: 40-56. Liu T, Zhang X, Chen R, et al. Development, comparison, and validation of four intelligent, practical machine learning models for patients with prostate-specific antigen in the gray zone[J]. Frontiers in oncology, 2023, 13: 1157384. Li X, Zhao Y, Zhang D, et al. Development of an interpretable machine learning model associated with heavy metals' exposure to identify coronary heart disease among US adults via SHAP: Findings of the US NHANES from 2003 to 2018[J]. Chemosphere, 2023, 311(Pt 1): 137039. Hu C, Li L, Huang W, et al. Interpretable Machine Learning for Early Prediction of Prognosis in Sepsis: A Discovery and Validation Study[J]. Infectious diseases and therapy, 2022, 11(3): 1117-1132. Ejiyi C J, Qin Z, Ukwuoma C C, et al. Comparative performance analysis of Boruta, SHAP, and Borutashap for disease diagnosis: A study with multiple machine learning algorithms[J]. Network (Bristol, England), 2024: 1-38. Rhee C, Dantes R, Epstein L, et al. Incidence and Trends of Sepsis in US Hospitals Using Clinical vs Claims Data, 2009-2014[J]. Jama, 2017, 318(13): 1241-1249. Valik J K, Hedberg P, Holmberg F, et al. Impact of the COVID-19 pandemic on the incidence and mortality of hospital-onset bloodstream infection: a cohort study[J]. BMJ quality & safety, 2022, 31(5): 379-382. Kontula K S K, Skogberg K, Ollgren J, et al. Population-Based Study of Bloodstream Infection Incidence and Mortality Rates, Finland, 2004-2018[J]. Emerging infectious diseases, 2021, 27(10): 2560-9. Zoabi Y, Kehat O, Lahav D, et al. Predicting bloodstream infection outcome using machine learning[J]. Scientific reports, 2021, 11(1): 20101. An R, Ou Y, Pang L, et al. Epidemiology and Risk Factors of Community-Associated Bloodstream Infections in Zhejiang Province, China, 2017-2020[J]. Infection and drug resistance, 2023, 16: 1579-1590. Florin T A, Ambroggio L, Brokamp C, et al. Biomarkers and Disease Severity in Children With Community-Acquired Pneumonia[J]. Pediatrics, 2020, 145(6). Tang Y H, Jeng M J, Wang H H, et al. Risk factors and predictive markers for early and late-onset neonatal bacteremic sepsis in preterm and term infants[J]. Journal of the Chinese Medical Association : JCMA, 2022, 85(4): 507-513. Yang X, Zeng J, Yu X, et al. PCT, IL-6, and IL-10 facilitate early diagnosis and pathogen classifications in bloodstream infection[J]. Annals of clinical microbiology and antimicrobials, 2023, 22(1): 103. Zhu Q, Li H, Zheng S, et al. IL-6 and IL-10 Are Associated With Gram-Negative and Gram-Positive Bacteria Infection in Lymphoma[J]. Frontiers in immunology, 2022, 13: 856039. Niu D, Huang Q, Yang F, et al. Serum biomarkers to differentiate Gram-negative, Gram-positive and fungal infection in febrile patients[J]. Journal of medical microbiology, 2021, 70(7). O'Grady N P, Alexander E, Alhazzani W, et al. Society of Critical Care Medicine and the Infectious Diseases Society of America Guidelines for Evaluating New Fever in Adult Patients in the ICU[J]. Critical care medicine, 2023, 51(11): 1570-1586. Doman M, Thy M, Dessajan J, et al. Temperature control in sepsis[J]. Frontiers in medicine, 2023, 10: 1292468. Xie Y, Yang Y, Han Y, et al. Association Between Arterial Blood Gas Variation and Intraocular Pressure in Healthy Subjects Exposed to Acute Short-Term Hypobaric Hypoxia[J]. Translational vision science & technology, 2019, 8(6): 22. Ouyang S M, Zhu H Q, Xie Y N, et al. Temporal changes in laboratory markers of survivors and non-survivors of adult inpatients with COVID-19[J]. BMC infectious diseases, 2020, 20(1): 952. Choi M H, Kim D, Park Y, et al. Development and validation of artificial intelligence models to predict urinary tract infections and secondary bloodstream infections in adult patients[J]. Journal of infection and public health, 2024, 17(1): 10-17. Additional Declarations No competing interests reported. Supplementary Files supplementarymaterials.doc Cite Share Download PDF Status: Published Journal Publication published 14 May, 2025 Read the published version in BMC Medical Informatics and Decision Making → Version 1 posted Editorial decision: Revision requested 03 Feb, 2025 Editor assigned by journal 03 Feb, 2025 Submission checks completed at journal 27 Jan, 2025 First submitted to journal 19 Jan, 2025 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-5859635","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":407680859,"identity":"f4332d9b-087e-43a6-a04e-f304a8f2be4a","order_by":0,"name":"Xiefei Hu","email":"","orcid":"","institution":"Department of Clinical Laboratory, Chongqing Emergency Medical Center, Chongqing University Central Hospital, School of Medicine, Chongqing University, Chongqing, China","correspondingAuthor":false,"prefix":"","firstName":"Xiefei","middleName":"","lastName":"Hu","suffix":""},{"id":407680860,"identity":"5c19454e-88db-4515-a86c-2b370973623a","order_by":1,"name":"Shenshen Zhi","email":"","orcid":"","institution":"Department of Clinical Laboratory, Chongqing Emergency Medical Center, Chongqing University Central Hospital, School of Medicine, Chongqing University, Chongqing, China","correspondingAuthor":false,"prefix":"","firstName":"Shenshen","middleName":"","lastName":"Zhi","suffix":""},{"id":407680861,"identity":"ea7caf67-e113-42d7-a4e7-287861f34e91","order_by":2,"name":"Yang Li","email":"","orcid":"","institution":"Peking University Chongqing Big Data Research Institute","correspondingAuthor":false,"prefix":"","firstName":"Yang","middleName":"","lastName":"Li","suffix":""},{"id":407680862,"identity":"68475c87-1aac-46c6-82b7-df07036657b6","order_by":3,"name":"Yuming Cheng","email":"","orcid":"","institution":"Beckman Coulter Commercial Enterprise (China) Co., Ltd","correspondingAuthor":false,"prefix":"","firstName":"Yuming","middleName":"","lastName":"Cheng","suffix":""},{"id":407680863,"identity":"11f17418-ac02-46d2-9460-07eb9f3c1673","order_by":4,"name":"Haiping Fan","email":"","orcid":"","institution":"School of Medicine ChongQing University","correspondingAuthor":false,"prefix":"","firstName":"Haiping","middleName":"","lastName":"Fan","suffix":""},{"id":407680864,"identity":"b264f032-c733-4968-a389-d0e88b5e433e","order_by":5,"name":"Haorong Li","email":"","orcid":"","institution":"Chongqing University of Posts and Telecommunications","correspondingAuthor":false,"prefix":"","firstName":"Haorong","middleName":"","lastName":"Li","suffix":""},{"id":407680865,"identity":"846cf890-269e-4a6f-9dd9-63ebb2340252","order_by":6,"name":"Zihao Meng","email":"","orcid":"","institution":"Chongqing University of Posts and Telecommunications","correspondingAuthor":false,"prefix":"","firstName":"Zihao","middleName":"","lastName":"Meng","suffix":""},{"id":407680866,"identity":"6611e708-0420-4c97-9a3f-116045f15914","order_by":7,"name":"Jiaxin Xie","email":"","orcid":"","institution":"School of Medicine ChongQing University","correspondingAuthor":false,"prefix":"","firstName":"Jiaxin","middleName":"","lastName":"Xie","suffix":""},{"id":407680867,"identity":"40a64d7b-4fd6-4d6f-8edc-ffad824cb753","order_by":8,"name":"Shu Tang","email":"","orcid":"","institution":"Chongqing University of Posts and Telecommunications","correspondingAuthor":false,"prefix":"","firstName":"Shu","middleName":"","lastName":"Tang","suffix":""},{"id":407680868,"identity":"19e85124-cbfb-4bad-9341-d323cd9f568b","order_by":9,"name":"Wei Li","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA1klEQVRIiWNgGAWjYBACPmYgkWAgwcPG3gAWYGwgpIUNrKXCQoaP5wCxWsDkmQobOYkEYrWw8xg+eNgGdJjk44e3eRhsZDccYH72AL/DeIwNEkFapNOMrXkY0ow3HGAzN8CvhXebBERLDps0D8PhxA0HeNgkCGjZ/gOsRfIMSMt/orRsY0g4A9QCsoiH4QAxWvg/SyRUANXzpBlbzjFINp55mM0MrxZ+/mOJH38Y1NnLtx9+eONNhZ1s3/HmZ3i1oAAJBlBQMROtHqxlFIyCUTAKRgEWAABFVTjN/S/DWgAAAABJRU5ErkJggg==","orcid":"","institution":"Department of Clinical Laboratory, Chongqing Emergency Medical Center, Chongqing University Central Hospital, School of Medicine, Chongqing University, Chongqing, China","correspondingAuthor":true,"prefix":"","firstName":"Wei","middleName":"","lastName":"Li","suffix":""}],"badges":[],"createdAt":"2025-01-19 13:23:01","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-5859635/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5859635/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s12911-025-03020-9","type":"published","date":"2025-05-14T15:57:02+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":75313047,"identity":"ba6b6510-5b69-4169-a04b-d53619f4670d","added_by":"auto","created_at":"2025-02-03 09:22:46","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":27674,"visible":true,"origin":"","legend":"\u003cp\u003eFlow chart depicting number of patients who were included in analysis after exclusion criteria. The total included encounters were divided into those with and without BSI.\u003c/p\u003e","description":"","filename":"1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5859635/v1/88ad883a998e76c08db445ea.jpg"},{"id":75312549,"identity":"181270dc-08bc-4e72-ae84-924de72535bf","added_by":"auto","created_at":"2025-02-03 09:14:46","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":45251,"visible":true,"origin":"","legend":"\u003cp\u003eThe ROC Curves of Predictive Factors Identified by Univariate Logistic Regression Analysis\u003c/p\u003e","description":"","filename":"2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5859635/v1/13986ca3b88f89b526aeda65.jpg"},{"id":75312547,"identity":"9e5b4fc6-4cbd-4f92-9adc-9d7016e19282","added_by":"auto","created_at":"2025-02-03 09:14:46","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":50982,"visible":true,"origin":"","legend":"\u003cp\u003eSelection of key features for BSI. a) Variable Selection Plot of Boruta; b) Variable Selection Plot of Lasso; c) Variable Selection Plot of RFE-CV; d)Venn graph displaying 5 features shared by Boruta, Lasso and RFE-CV.\u003c/p\u003e","description":"","filename":"3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5859635/v1/c1fa5152bce7793f59377983.jpg"},{"id":75313050,"identity":"09aa9c80-1842-4cf9-9a6f-e0029af5f396","added_by":"auto","created_at":"2025-02-03 09:22:46","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":41281,"visible":true,"origin":"","legend":"\u003cp\u003eROC Curves of Six Models in the Internal Validation Set\u003c/p\u003e","description":"","filename":"4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5859635/v1/98a7e17912d91929378f1a66.jpg"},{"id":75313049,"identity":"fa4811db-f3cb-4fb3-b8da-d0d4311a3ba4","added_by":"auto","created_at":"2025-02-03 09:22:46","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":33106,"visible":true,"origin":"","legend":"\u003cp\u003ePerformance Evaluation of the XGBoost Model. a) ROC Curve of External Validation Set in the XGBoost Model; b) calibration curve of XGBoost Model\u003c/p\u003e","description":"","filename":"5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5859635/v1/23274d21d6e4727487a81449.jpg"},{"id":75313052,"identity":"ac0371ab-eb45-4ad5-89f8-e6f3d30bc807","added_by":"auto","created_at":"2025-02-03 09:22:47","extension":"jpg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":50567,"visible":true,"origin":"","legend":"\u003cp\u003eModel Interpretation of XGBoost. a) Importance Ranking of Features; b) Example of Low-risk Patient; c) Example of Hight-risk Patient\u003c/p\u003e","description":"","filename":"6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5859635/v1/6454af900a984348cf9d988d.jpg"},{"id":83068791,"identity":"1073362f-b4a7-430f-ac10-16159717db71","added_by":"auto","created_at":"2025-05-19 16:10:56","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":955321,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5859635/v1/5a55c5d7-7d2b-40c2-9343-54d72694f507.pdf"},{"id":75312550,"identity":"36e60bbb-592c-48b5-9813-76daa25ce7e7","added_by":"auto","created_at":"2025-02-03 09:14:46","extension":"doc","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":535248,"visible":true,"origin":"","legend":"","description":"","filename":"supplementarymaterials.doc","url":"https://assets-eu.researchsquare.com/files/rs-5859635/v1/5a178359339acdd6723db902.doc"}],"financialInterests":"No competing interests reported.","formattedTitle":"Development and Application of an Early Prediction Model for Risk of Bloodstream Infection based on Real-world Study","fulltext":[{"header":"Introduction","content":"\u003cp\u003eBloodstream Infection (BSI) refers to the invasion of microorganisms into the bloodstream, leading to a systemic infection, which can cause damage to all organs of the body. It is prone to inducing sepsis and multiple organ dysfunction syndrome (MODS), and associated with a high mortality rate\u003csup\u003e[1, 2]\u003c/sup\u003e. Annually, approximately 1.2\u0026nbsp;million patients are diagnosed with BSI in Europe\u003csup\u003e[3]\u003c/sup\u003e. Every hour of delayed treatment for BSI raises the mortality rate by 8%, reaching 58% after a 6-hour delay\u003csup\u003e[4]\u003c/sup\u003e. In the ICU, factors like weakened immunity, frequent risky procedures, multiple complications, and extended hospital stays heighten the risk of BSI, making it a common issue in these settings. Untreated BSI can rapidly lead to sepsis, progressing to MODS, causing poor outcomes and life-threatening conditions\u003csup\u003e[5\u0026ndash;7]\u003c/sup\u003e. Early detection, appropriate antibiotics, and addressing the source of BSI greatly reduce morbidity and mortality rates\u003csup\u003e[8]\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eBlood culture is the benchmark for diagnosing BSI but has limitations such as low positivity rates, long turnaround times, contamination risks, and challenges in detecting certain pathogens with standard culturing\u003csup\u003e[9]\u003c/sup\u003e. Machine learning (ML), a vital part of artificial intelligence, has advanced analytical powers that can independently detect disease patterns in data and forecast unknown results\u003csup\u003e[10,,11]\u003c/sup\u003e. Compared to traditional diagnostic and therapeutic approaches, ML offers a deeper insight into complex relationships. In recent years, ML has shown significant promise in disease screening, diagnosis, prognosis prediction, and risk analysis\u003csup\u003e[12]\u003c/sup\u003e. Developing early prediction models for BSI using ML is crucial for enhancing early diagnosis, treatment, and personalized healthcare. However, current prediction models often require a large number of features\u003csup\u003e[13\u0026ndash;15]\u003c/sup\u003e. While including more features can improve the predictive ability of the models, it can also lead to increased complexity, requiring more data for training, and reducing interpretability and generalizability of the model. This poses a challenge in practical clinical settings, especially in primary care facilities where extensive testing and comprehensive patient data collection may not be feasible. Therefore, researchers and clinicians need to find a balance\u0026mdash;ensuring predictive accuracy while minimizing the number and complexity of required features\u0026mdash;to make these predictive models effective in resource-limited environments, such as grassroots healthcare institutions.\u003c/p\u003e \u003cp\u003eConsequently, this study aims to analyze routine laboratory/clinical data to identify key predictive factors that play a significant role in the early diagnosis of BSI. The goal is to find the optimal combination of these factors and use machine learning algorithm to develop a broadly applicable early risk prediction model for BSI. This model aims to facilitate early and rapid prediction of BSI in a variety of clinical settings and will be validated and implemented in real-world scenarios.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eStudy Population\u003c/h2\u003e \u003cp\u003eThis study was a secondary analysis of a retrospective observational study conducted from 2021 to 2023 among inpatients at the Chongqing University Central Hospital. The inclusion criteria were (1)age\u0026thinsp;\u0026ge;\u0026thinsp;18 years; (2)inpatients; (3)had at least one blood culture examination performed during hospital stay. The exclusion criteria were (1)The blood culture results indicated a probable contaminant; (2)Data missing rate\u0026thinsp;\u0026ge;\u0026thinsp;30%. Clinical or laboratory parameters related to BSI were collected for each patient. For patients with multiple positive BC samples, only the first episode was included. For those with multiple negative BC samples, a single episode was randomly selected.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eOutcome\u003c/h3\u003e\n\u003cp\u003eThe outcome assessed was BSI, defined as the growth of a clinically significant pathogen in at least one BC bottle. Potential contaminants were defined by the Center for Disease Control and Prevention (CDC)/National Health Safety Network (NHSN) guidelines for Laboratory Confirmed Bloodstream Infection (LCBI) and were not classified as BSIs. These potential contaminants include coagulase-negative Staphylococci, Corynebacterium species, Bacillus species, Diphtheroids, Aerococcus, and Propionibacterium species\u003csup\u003e[13]\u003c/sup\u003e.\u003c/p\u003e\n\u003ch3\u003eDataset\u003c/h3\u003e\n\u003cp\u003eAt the target medical centers, we constructed datasets that included demographics, clinical and laboratory parameters, including microbiology, available within 3 hours before and after BC sampling time.The dataset included as follows: (i) blood cells; (ii) liver function; (iii) renal function; (iv) hemagglutination; (v) blood gas analysis; (vi) electrolytes; (vii) inflammatory markers; (viii) blood culture; (ix) clinical features.\u003c/p\u003e \u003cp\u003eWe collected datasets from two time periods: the dataset from January 2021 to April 2023 was randomly split into training and validation sets comprising 70% and 30% respectively. The training sets were used for modelling, while the validation sets for internal validation. The dataset from May 2023 to December 2023 was used for external validation of the best model.\u003c/p\u003e\n\u003ch3\u003eData preprocessing\u003c/h3\u003e\n\u003cp\u003eData cleaning and preprocessing are critical steps in the data analysis process, aimed at transforming raw data into a format suitable for statistical analysis or ML modeling\u003csup\u003e[16,17]\u003c/sup\u003e. In this study, data cleaning and preprocessing primarily involved the removal of duplicate data, analysis and treatment of outliers, imputation of missing values, data standardization, and balancing of data categories.\u003c/p\u003e\n\u003ch3\u003eFeature selection and modeling\u003c/h3\u003e\n\u003cp\u003eFeature selection: (i) In this study, the initial method for selecting predictive factors involved univariable logistic regression. Univariable logistic regression allowed for the assessment of whether each biomarker was independently associated with BSIs, thus enabling the preliminary selection of predictive factors for model development. (ii) The study also incorporated the Boruta algorithm\u003csup\u003e[18\u0026ndash;20]\u003c/sup\u003e, Lasso regression\u003csup\u003e[21\u0026ndash;22]\u003c/sup\u003e, and Recursive Feature Elimination with Cross-validation (RFE-CV)\u003csup\u003e[23\u0026ndash;25]\u003c/sup\u003e to optimize the results obtained from the univariable logistic regression analysis.\u003c/p\u003e \u003cp\u003eModeling: In this study, we used the Light Gradient Boosting Machine(LightGBM)、eXtreme Gradient Boosting (XGBoost)、Gradient Boosting Decision Tree (GBDT)、Random Forest (RF)、Support Vector Machine (SVM) and Gaussian Naive Bayes (GNB) algorithms to predict the risk of BSI in inpatients by analyzing clinic/laboratory data\u003csup\u003e[26\u0026ndash;31]\u003c/sup\u003e. Throughout the model development phase, we implemented a grid search technique to refine the hyperparameters.\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eValidation and Explanation\u003c/h2\u003e \u003cp\u003eWe evaluated the performance of the model by applying several different indices, namely (i) AUC, (ii) accuracy, (iii) sensitivity, and (iv) specificity. The performance assessment for selecting the best model will primarily be based on the AUC value. First, we conducted an assessment on the internal validation set, which comprised 30% of the original data that was initially set aside for validation purposes only. After model selection, we used the Shapley Additive Explanations (SHAP) algorithm from model-agnostic approaches to explain the best-performing model\u003csup\u003e[32\u0026ndash;34]\u003c/sup\u003e. Finally, the dataset from May 2023 to December 2023 was utilized for external validation of the optimal model.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003eStatistical Analysis\u003c/h2\u003e \u003cp\u003eBinary variables were presented as counts and percentages, and their significance was assessed using the Chi-square test or Fisher's exact test. Continuous variables that were normally distributed were compared with a t-test and reported as means\u0026thinsp;\u0026plusmn;\u0026thinsp;SEM. For variables with a non-normal distribution, the Mann\u0026ndash;Whitney U test was applied. A P-value of less than 0.05 was deemed statistically significant. All statistical analyses were conducted in the Beckman Coulter DxAI platform (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.xsmartanalysis.com/beckman/login/\u003c/span\u003e\u003cspan address=\"https://www.xsmartanalysis.com/beckman/login/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003ePatient characteristics\u003c/h2\u003e \u003cp\u003eOur model construction database initially contained 5,057 inpatients. Following a series of exclusions, 2,323 adult inpatients were included in this study, of which 300 patients developed BSI, representing 12.9% of the study population. The patient selection process is illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. The baseline characteristics of the patients are presented in Supplementary Table\u0026nbsp;1.The training and internal validation datasets comprised of 1,626 and 697 patients, respectively. A total of 74 variables, including age, sex, Temperature, White Blood Cell Count (WBC), D-dimer, and other laboratory or clinical parameters related to BSI, were collected for each patient. A comparison of basic information between the two sets were shown in Supplementary Table\u0026nbsp;2.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFor external validation of the model, 259 patients were included, of whom 34 developed BSI (13.13%). The baseline characteristics of the patients are presented in Supplementary Table\u0026nbsp;3.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eVariables of importance\u003c/h2\u003e \u003cp\u003eThe model's accuracy increased as more variables were incorporated. However, increasing the number of variables did not correspond with the practicality needed for clinical application. In order to indentify the most significant features, we employed univariate logistic regression to preliminarily screen the variables associated with BSI within the training set.We identified 27 variables that are crucial for predicting BSI, which were shown in Supplementary Table\u0026nbsp;4\u003c/p\u003e \u003cp\u003eBased on the results of the univariate logistic regression analysis, the individual indicators that were screened (WBC, EOS, EOS%, Neu%, Mon, Mon%, RDW, Hct, PLT, A/G, Alb, CHE, PA, Cr, UA, Urea, Fib, SB, AB, BEf, Lac, TCO2, Cl, Mg, IL-6, hs-CRP, and T) were used separately to predict whether patients had BSI. As shown in Fig.\u0026nbsp;2, the AUC values for Neu%, Cr, Urea, and T exceeded 0.600, while the AUC values for the remaining indicators were all below 0.600. The efficacy of single indicators for predicting BSI was poor.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Taba\" border=\"1\"\u003e \u003ccolgroup cols=\"1\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFIGURE 2. The ROC Curves of Predictive Factors Identified by Univariate Logistic Regression Analysis\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eWe utilized the Boruta algorithm, Lasso regression, and RFE-CV to further reduce the number of variables. As shown in Fig.\u0026nbsp;3, the Boruta algorithm indentified 19 variables such as Hct, Fib, UA, Cl, Alb, hs-CRP, WBC, TCO2, Urea, AB, Cr, Mg, Mon, IL-6, Mon%, BEf, SB, Neu%, and T. The Lasso regression analysis highlighted 15 features that help minimize the model's prediction error: WBC, EOS, PLT, PA, Lac, UA, TCO2, AB, SB, BEf, Na, Cl, hs-CRP, IL-6, and T. Meanwhile, the RFE-CV method selected the top five feature indicators based on their contribution rankings, which are WBC, SB, BEf, IL-6, and T. Ultimately, by taking the intersection of the results from these three algorithms, we identified the 5 key features that contribute the most to the model's predictive capability: WBC, SB, BEf, IL-6, and T.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Tabb\" border=\"1\"\u003e \u003ccolgroup cols=\"1\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFIGURE 3 Selection of key features for BSI. a) Variable Selection Plot of Boruta; b) Variable Selection Plot of Lasso; c) Variable Selection Plot of RFE-CV; d)Venn graph displaying 5 features shared by Boruta, Lasso and RFE-CV.\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eClassification Results\u003c/h2\u003e \u003cp\u003eBased on the selected 5 key features (WBC, SB, BEf, IL-6, and T), we constructed six early prediction models for BSI risk using machine learning algorithms: the LightGBM model, the XGBoost model, the GBDT model, the RF model, the SVM model, and the GNB model. The model construction process involved hyperparameter optimization using grid search techniques.\u003c/p\u003e \u003cp\u003eAs shown in Fig.\u0026nbsp;4, the average AUC values for the XGBoost, LightGBM, RF, GBDT, GNB, and SVM models on the internal validation set were 0.782 (95% CI: 0.715\u0026ndash;0.849), 0.700 (95% CI: 0.627\u0026ndash;0.773), 0.772 (95% CI: 0.704\u0026ndash;0.841), 0.723 (95% CI: 0.650\u0026ndash;0.797), 0.562 (95% CI: 0.483\u0026ndash;0.642), and 0.528 (95% CI: 0.446\u0026ndash;0.611), respectively. The XGBoost model had the highest AUC value of 0.782, while the SVM model had the lowest AUC value of 0.528.\u003c/p\u003e \u003cp\u003e,\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Tabc\" border=\"1\"\u003e \u003ccolgroup cols=\"2\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFIGURE 4 ROC Curves of Six Models in the Internal Validation Set\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eAs shown in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e, the RF model had the highest accuracy rate at 0.882; the GNB model had the highest sensitivity at 0.747; and the XGBoost model had the highest specificity at 0.824. Considering the AUC values and the evaluation metrics, the XGBoost model emerged as the best model.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eEvaluation Metrics Results of Six Models\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModels\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAccuracy\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eSensitivity\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eSpecificity\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eXGBoost\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.763\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.633\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.824\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLightGBM\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.830\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.527\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.788\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRF\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.882\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.653\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.799\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGBDT\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.729\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.643\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.738\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGNB\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.470\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.747\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.416\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSVM\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.595\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.510\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.642\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eIn the external validation, the AUROC of the XGBoost model decreased to 0.776, with an accuracy of 0.685, sensitivity of 0.647, and specificity of 0.800. The calibration curve was close to the 45\u0026deg; line, indicating a good fit between the model's predictions and the actual values. The results are shown in Fig.\u0026nbsp;5.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Tabd\" border=\"1\"\u003e \u003ccolgroup cols=\"1\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFIGURE 5 Performance Evaluation of the XGBoost Model. a) ROC Curve of External Validation Set in the XGBoost Model; b) calibration curve of XGBoost Model\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eModel Interpretation and Online Application\u003c/h2\u003e \u003cp\u003eTo better understand the prediction results of the XGBoost model and the basis for decision-making, the SHAP algorithm was used to quantify the contribution of each feature to the model's predictive outcomes. Figure\u0026nbsp;6a displays the ranking of feature contributions in the XGBoost model, with the indicators ranked from highest to lowest contribution being SB, BEf, IL-6, T, and WBC.\u003c/p\u003e \u003cp\u003eFor individual patients, as shown in Fig.\u0026nbsp;6b and c ,the figure uses color coding to represent the impact of features on the prediction. Blue indicates features that negatively influence the prediction (leftward arrows, which correspond to a decrease in SHAP values), and red signifies features that positively affect the prediction (rightward arrows, indicating an increase in SHAP values). The base value represents the average model output for the training set, and the SHAP values for an individual patient's model output are indicated by f(x). In Fig.\u0026nbsp;6b, the f(x) value is below the base value (0.03 compared to 0.20), which suggests the model predicts a low risk of BSI for this patient. In contrast, in Fig.\u0026nbsp;6c, the f(x) value exceeds the base value (0.62 compared to 0.20), leading the model to predict a high risk of BSI for the patient.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Tabe\" border=\"1\"\u003e \u003ccolgroup cols=\"1\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFIGURE 6 Model Interpretation of XGBoost. a) Importance Ranking of Features; b) Example of Low-risk Patient; c) Example of Hight-risk Patient\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eTo enhance the practicality and broad applicability of the constructed model in clinical practice, early risk prediction for patients can be conducted via an online link. The URL for the online prediction tool is: \u003cem\u003e[\u003c/em\u003e\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.xsmartanalysis.com/model/list/predict/model/html?mid=13885\u0026amp;symbol=11im71SWNC211Qj91806](http://www.xsmartanalysis.com/model/list/predict/model/html?mid=13885\u0026amp;symbol=11im71SWNC211Qj91806)\u003c/span\u003e\u003cspan address=\"http://www.xsmartanalysis.com/model/list/predict/model/html?mid=13885\u0026amp;symbol=11im71SWNC211Qj91806](http://www.xsmartanalysis.com/model/list/predict/model/html?mid=13885\u0026amp;symbol=11im71SWNC211Qj91806)\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eThe study findings revealed a 12.91% incidence rate of BSI among hospitalized patients from 2021 to 2023. This rate exceeded that of a 6-year U.S. retrospective study (12.91% vs. 5.90%)\u003csup\u003e[35]\u003c/sup\u003e, likely due to the study hospital's role as a national critical care center, which saw a higher volume of critically ill patients prone to BSI. The period of data collection had coincided with the COVID-19 pandemic, potentially contributing to the elevated BSI rates\u003csup\u003e[36]\u003c/sup\u003e. Finnish research had indicated higher BSI incidence and mortality in the elderly, particularly those over 80\u003csup\u003e[37]\u003c/sup\u003e. Our study population had an older median age (68.0 years for the cohort, 72.0 years for BSI cases), which may have explained the higher incidence. With an aging population, addressing BSI in elderly patients was crucial. It was important to note that pre-admission BSI cases had not been excluded, possibly inflating the incidence rate with community-acquired BSI.\u003c/p\u003e \u003cp\u003eEarly diagnosis of BSI is vital for lowering mortality and enhancing patient outcomes. As artificial intelligence evolves, Machine Learning (ML) algorithms are becoming pivotal in medicine, particularly for BSI diagnosis. Studies like Roimi's achieved an AUC of 0.930 with 50 features\u003csup\u003e[13]\u003c/sup\u003e, Zhang's LSTM model reached 0.892 with over 100 features\u003csup\u003e[15]\u003c/sup\u003e, and Zoabi et al. reported 0.810 with 25 features\u003csup\u003e[38]\u003c/sup\u003e. While more features can improve model performance, extensive data collection complicates practical use, especially in primary care where early BSI diagnosis is challenging. This study initially narrowed 74 predictors to 27 via univariate logistic regression, but single-factor prediction was inadequate. Further analysis led to feature selection using ML methods, including Boruta, Lasso, and RFE-CV, pinpointing 5 key ,indicators for early BSI risk: SB, BEf, IL-6, T, and WBC. WBC\u003csup\u003e[40\u0026ndash;41]\u003c/sup\u003e, IL-6\u003csup\u003e[42\u0026ndash;44]\u003c/sup\u003e, and T\u003csup\u003e[45,46]\u003c/sup\u003e are standard in infectious disease management and are key in BSI diagnosis. Blood gas analysis, often focusing on TCO2 and pH, has seen less research on SB and BEf for early BSI detection. Some studies have indicated that during the early stages of infectious diseases, more pronounced changes occur in SB and BEF. Research suggests that the systemic inflammatory response induced by infection can impair the normal function of the circulatory system, thereby affecting tissue oxygenation. Even when the blood pH of patients has not shown significant fluctuations, the SB level begins to decline in the context of hypoxia \u003csup\u003e[47]\u003c/sup\u003e. When BSI patients experience acid-base balance disorders, BEF exhibits marked abnormalities. Song-Mao Ouyang and colleagues have observed statistically significant differences in BEF values between infected and non-infected patients (P\u0026thinsp;\u0026lt;\u0026thinsp;0.05)\u003csup\u003e[48]\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eThe XGBoost model outperformed others with an AUC of 0.782 and high specificity, aligning closely with the 45\u0026deg; line in calibration curves. It was chosen as the optimal model for further external validation and clinical use.External validation is key to evaluating a model's performance and generalizability. It ensures accuracy and reliability\u003csup\u003e[49]\u003c/sup\u003e. The XGBoost model, tested on a new dataset, achieved an AUC of 0.776 for BSI prediction, demonstrating robust predictive ability. The calibration curve showed a close match between predictions and actual results. For clinical use, the model is accessible online, allowing clinicians to input WBC, SB, T, BEf, and IL-6 values to receive BSI risk predictions.\u003c/p\u003e \u003cp\u003eThis study's goal is to create an early BSI risk prediction model using standard, affordable, and easy-to-administer lab tests. We aimed to pinpoint key tests for early BSI prediction to streamline clinical diagnostics. Machine Learning was employed to develop the prediction model, which is designed for easy use in various healthcare settings. Our analysis revealed five key predictors: WBC, SB, T, BEf, and IL-6. While WBC, IL-6, and T are standard infection markers, SB and BEf's role in BSI prediction is underexplored. Our model underscores the importance of these latter two indicators, suggesting they deserve more research attention. The study has two key limitations. First, all patient data was sourced from one institution, potentially leading to selection bias. The model, developed with a focus on critical patients at Chongqing University Affiliated Central Hospital, may not generalize to other patient groups. Future research should use multi-center data to enhance the model's universality and reliability. Second, the model, available online, lacks prospective clinical validation due to BSI's low incidence. Subsequent research should validate the model with real-world clinical data across various populations, times, and settings to confirm its predictive value for BSI and its utility in clinical decisions.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eBSI \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Bloodstream Infection\u003c/p\u003e\n\u003cp\u003eMODS \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Multiple Organ Dysfunction Syndrome\u003c/p\u003e\n\u003cp\u003eRFE-CV \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Recursive Feature Elimination with Cross-validation\u003c/p\u003e\n\u003cp\u003eSHAP \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Shapley Additive Explanations\u003c/p\u003e\n\u003cp\u003eML \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Machine learning\u003c/p\u003e\n\u003cp\u003eCDC \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Disease Control and Prevention\u003c/p\u003e\n\u003cp\u003eNHSN \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; National Health Safety Network\u003c/p\u003e\n\u003cp\u003eLCBI \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Laboratory Confirmed Bloodstream Infection\u003c/p\u003e\n\u003cp\u003eLightGBM \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Light Gradient Boosting\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eXGBoost \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Extreme Gradient Boosting\u003c/p\u003e\n\u003cp\u003eGBDT \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Gradient Boosting Decision Tree\u003c/p\u003e\n\u003cp\u003eRF \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Random Forest\u003c/p\u003e\n\u003cp\u003eSVM \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Support Vector Machine\u003c/p\u003e\n\u003cp\u003eGNB \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Gaussian Naive Bayes\u003c/p\u003e\n\u003cp\u003eMCHC \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Mean Corpuscular Hemoglobin Concentration\u003c/p\u003e\n\u003cp\u003eRDW \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Red Cell Distribution Width\u003c/p\u003e\n\u003cp\u003eMCH \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Mean Corpuscular Hemoglobin\u003c/p\u003e\n\u003cp\u003eMCV \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Mean Corpuscular Volume\u003c/p\u003e\n\u003cp\u003ePLT \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Platelet Count\u003c/p\u003e\n\u003cp\u003eWBC \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;White Blood Cell Count\u003c/p\u003e\n\u003cp\u003eNeu \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Neutrophil Count\u003c/p\u003e\n\u003cp\u003eNeu% \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Neutrophils Percentage\u003c/p\u003e\n\u003cp\u003eEos \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Direct Eosinophil Count\u003c/p\u003e\n\u003cp\u003eEos% \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Eosinophils Percentage\u003c/p\u003e\n\u003cp\u003eMon \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Monocyte Count\u003c/p\u003e\n\u003cp\u003eMon% \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Monocytes Percentage\u003c/p\u003e\n\u003cp\u003eBaso \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Direct Basophil Count\u003c/p\u003e\n\u003cp\u003eBaso% \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Basophils Percentage\u003c/p\u003e\n\u003cp\u003eLym \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Lymphocyte Count\u003c/p\u003e\n\u003cp\u003eLym% \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Lymphocyte Percentage\u003c/p\u003e\n\u003cp\u003eRBC \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Red Blood Cell Count\u003c/p\u003e\n\u003cp\u003eHb \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Hemoglobin\u003c/p\u003e\n\u003cp\u003eHct \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Hematocrit\u003c/p\u003e\n\u003cp\u003ePlateletcrit \u0026nbsp; \u0026nbsp; \u0026nbsp;Plateletcrit\u003c/p\u003e\n\u003cp\u003eAFU \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026alpha;-fucosidase\u003c/p\u003e\n\u003cp\u003eALP \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Alkaline Phosphatase\u003c/p\u003e\n\u003cp\u003eALT \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Alanine Aminotransferase\u003c/p\u003e\n\u003cp\u003eAlb \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Albumin\u003c/p\u003e\n\u003cp\u003eCHE \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Cholinesterase\u003c/p\u003e\n\u003cp\u003eGCA \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Glycocholic Acid\u003c/p\u003e\n\u003cp\u003eLDH \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Lactate Dehydrogenase\u003c/p\u003e\n\u003cp\u003ePA \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Prealbumin\u003c/p\u003e\n\u003cp\u003eTBA \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Total Bile Acids\u003c/p\u003e\n\u003cp\u003eTP \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Total Protein\u003c/p\u003e\n\u003cp\u003eTBiL \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Total Bilirubin\u003c/p\u003e\n\u003cp\u003eA/G \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Alb/Glob Ratio\u003c/p\u003e\n\u003cp\u003e5-n \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;5-nucleotidase\u003c/p\u003e\n\u003cp\u003eGGT \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Gamma-glutamyl Transferase\u003c/p\u003e\n\u003cp\u003eCr \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Creatinine\u003c/p\u003e\n\u003cp\u003eCys-C \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Cystatin C\u003c/p\u003e\n\u003cp\u003eUA \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Uric Acid\u003c/p\u003e\n\u003cp\u003e\u0026alpha;1-MG \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026alpha;1-microglobulin\u003c/p\u003e\n\u003cp\u003e\u0026beta;2-MG \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026beta;2-microglobulin\u003c/p\u003e\n\u003cp\u003ePT \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Prothrombin Time\u003c/p\u003e\n\u003cp\u003eAPTT \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Activated Partial Thromboplastin Time\u003c/p\u003e\n\u003cp\u003eFib \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Fibrinogen\u003c/p\u003e\n\u003cp\u003eTT \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Thrombin Time\u003c/p\u003e\n\u003cp\u003eINR \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; International Normalized Ratio\u003c/p\u003e\n\u003cp\u003eD-D \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;D-dimer\u003c/p\u003e\n\u003cp\u003ePT% \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Prothrombin Activity\u003c/p\u003e\n\u003cp\u003eAB \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Actual Bicarbonate\u003c/p\u003e\n\u003cp\u003eAG \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Anion Gap\u003c/p\u003e\n\u003cp\u003eBEf \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Base Excess of Extracellular Fluid\u003c/p\u003e\n\u003cp\u003eBOP \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Blood Osmotic Pressure\u003c/p\u003e\n\u003cp\u003eFiO2 \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Fraction of Inspired Oxygen\u003c/p\u003e\n\u003cp\u003ePCO2 \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Partial Pressure of Carbon Dioxide\u003c/p\u003e\n\u003cp\u003epH \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Pondus Hydrogeni\u003c/p\u003e\n\u003cp\u003eSB \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Standard Bicarbonate\u003c/p\u003e\n\u003cp\u003eTCO2 \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Total Carbon Dioxide Partial Pressure\u003c/p\u003e\n\u003cp\u003eLac \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Lactic Acid\u003c/p\u003e\n\u003cp\u003ePO2 \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Oxygen Partial Pressure\u003c/p\u003e\n\u003cp\u003eNa \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Sodium\u003c/p\u003e\n\u003cp\u003eK \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Potassium\u003c/p\u003e\n\u003cp\u003eCl \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Chlorine\u003c/p\u003e\n\u003cp\u003eMg \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Magnesium\u003c/p\u003e\n\u003cp\u003eP \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Phosphorus\u003c/p\u003e\n\u003cp\u003ePCT \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Procalcitonin\u003c/p\u003e\n\u003cp\u003eIL-6 \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Interleukin-6\u003c/p\u003e\n\u003cp\u003ehs-CRP \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;High-sensitive C-reactive Protein\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\u003eFunding\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by the Science and Technology Research Project of the Chongqing Municipal Education Commission (grant number: KJZD-M202300101),\u0026nbsp;\u003c/p\u003e\n\u003cp\u003ethe Emergency Medicine Chongqing Key Laboratory Talent Development Innovation Joint Fund Project (grant number: 2024RCCX06) and the Wu Jieping Medical Foundation (grant number: 320.6750.2024-23-1 1). The statements made herein are solely the responsibility of the authors.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDisclosure of potential conflicts of interest\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no confict of interest.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eContributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eXiefei Hu:\u0026nbsp;Conceptualization, Data Curation, Methodology, Software, Writing- Original draft preparation, Writing- Reviewing and Editing.\u0026nbsp;Shenshen Zhi: Conceptualization, case data collection and article design.\u0026nbsp;Yuming Chen and Yang Li:\u0026nbsp;Conceptualization, Methodology.\u0026nbsp;Haiping Fan, Haorong Li, Zihao Meng and Jiaxin Xie:\u0026nbsp;Data Curation, Methodology, Software.\u0026nbsp;Shu Tang and Wei Li:\u0026nbsp;Overall planning. All authors contributed to the article and approved the submitted version.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCorresponding author\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCorrespondence to\u0026nbsp;Shu Tang and Wei Li:\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics declarations\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe studies involving humans were approved by the Ethics Committee of Chongqing Emergency Medical Center and Chongqing University Central Hospital (Approval Ethics Review No.RS202410). The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written\u0026nbsp;\u003c/p\u003e\n\u003cp\u003einformed consent to participate in this study\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for the 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 no competing interests.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eLamy B, Sundqvist M, Idelevich E A. Bloodstream infections - Standard and progress in pathogen diagnostics[J]. Clinical microbiology and infection, 2020, 26(2): 142-150.\u003c/li\u003e\n\u003cli\u003eShanghai Society for Microbiology, Clinical Microbiology Professional Committee, Shanghai Medical Association, Critical Care Medicine Specialty Branch, Shanghai Medical Association, Critical Care Medicine Specialty Branch. Expert Consensus on Clinical Laboratory Testing Pathways for Bloodstream Infections [J]. Chinese Journal of Infectious Diseases, 2022, 40(08): 457-475.\u003c/li\u003e\n\u003cli\u003eVincent J L, Sakr Y, Singer M, et al. Prevalence and Outcomes of Infection Among Patients in Intensive Care Units in 2017[J]. Jama, 2020, 323(15): 1478-1487.\u003c/li\u003e\n\u003cli\u003eLin K, Zhang H C, Zhao Y H, et al. The direct application of plasma droplet digital PCR in the ultra-early pathogen detection and warning during sepsis: Case reports[J]. Journal of infection and public health, 2022, 15(4): 450-454.\u003c/li\u003e\n\u003cli\u003eRudd K E, Johnson S C, Agesa K M, et al. Global, regional, and national sepsis incidence and mortality, 1990-2017: analysis for the Global Burden of Disease Study[J]. Lancet (London, England), 2020, 395(10219): 200-211.\u003c/li\u003e\n\u003cli\u003eXie J, Wang H, Kang Y, et al. The Epidemiology of Sepsis in Chinese ICUs: A National Cross-Sectional Survey[J]. Critical care medicine, 2020, 48(3): e209-e18.\u003c/li\u003e\n\u003cli\u003eOverbeek R, Leitl C J, Stoll S E, et al. The Value of Next-Generation Sequencing in Diagnosis and Therapy of Critically Ill Patients with Suspected Bloodstream Infections: A Retrospective Cohort Study[J]. Journal of clinical medicine, 2024, 13(2).\u003c/li\u003e\n\u003cli\u003eSchenz J, Weigand M A, Uhle F. Molecular and biomarker-based diagnostics in early sepsis: current challenges and future perspectives[J]. Expert review of molecular diagnostics, 2019, 19(12): 1069-1078.\u003c/li\u003e\n\u003cli\u003eWarren B G, Yarrington M E, Polage C R, et al. Evaluation of hospital blood culture utilization rates to identify opportunities for diagnostic stewardship[J]. Infection control and hospital epidemiology, 2023, 44(2): 200-205.\u003c/li\u003e\n\u003cli\u003eSwanson K, Wu E, Zhang A, et al. From patterns to patients: Advances in clinical machine learning for cancer diagnosis, prognosis, and treatment[J]. Cell, 2023, 186(8): 1772-1791.\u003c/li\u003e\n\u003cli\u003ePeiffer-Smadja N, Rawson T M, Ahmad R, et al. Machine learning for clinical decision support in infectious diseases: a narrative review of current applications[J]. Clinical microbiology and infection, 2020, 26(5): 584-595.\u003c/li\u003e\n\u003cli\u003eShibue K. Artificial Intelligence and Machine Learning in Clinical Medicine[J]. The New England journal of medicine, 2023, 388(25): 2398.\u003c/li\u003e\n\u003cli\u003eRoimi M, Neuberger A, Shrot A, et al. Early diagnosis of bloodstream infections in the intensive care unit using machine-learning algorithms[J]. Intensive care medicine, 2020, 46(3): 454-462.\u003c/li\u003e\n\u003cli\u003eLi X, Xu X, Xie F, et al. A Time-Phased Machine Learning Model for Real-Time Prediction of Sepsis in Critical Care[J]. Critical care medicine, 2020, 48(10): e884-e888.\u003c/li\u003e\n\u003cli\u003eZhang D, Yin C, Hunold K M, et al. An interpretable deep-learning model for early prediction of sepsis in the emergency department[J]. Patterns (New York, NY), 2021, 2(2): 100196.\u003c/li\u003e\n\u003cli\u003eChicco D, Oneto L, Tavazzi E. Eleven quick tips for data cleaning and feature engineering[J]. PLoS computational biology, 2022, 18(12): e1010718.\u003c/li\u003e\n\u003cli\u003eMaletzky A, B\u0026ouml;ck C, Tschoellitsch T, et al. Lifting Hospital Electronic Health Record Data Treasures: Challenges and Opportunities[J]. JMIR medical informatics, 2022, 10(10): e38557.\u003c/li\u003e\n\u003cli\u003eYue S, Li S, Huang X, et al. Machine learning for the prediction of acute kidney injury in patients with sepsis[J]. Journal of translational medicine, 2022, 20(1): 215.\u003c/li\u003e\n\u003cli\u003eZhou H, Xin Y, Li S. A diabetes prediction model based on Boruta feature selection and ensemble learning[J]. BMC bioinformatics, 2023, 24(1): 224.\u003c/li\u003e\n\u003cli\u003eKong C, Zhu Y, Xie X, et al. Six potential biomarkers in septic shock: a deep bioinformatics and prospective observational study[J]. Frontiers in immunology, 2023, 14: 1184700.\u003c/li\u003e\n\u003cli\u003eZhou T, Ren Z, Ma Y, et al. Early identification of bloodstream infection in hemodialysis patients by machine learning[J]. Heliyon, 2023, 9(7): e18263.\u003c/li\u003e\n\u003cli\u003eTang G, Qi L, Sun Z, et al. Evaluation and analysis of incidence and risk factors of lower extremity venous thrombosis after urologic surgeries: A prospective two-center cohort study using LASSO-logistic regression[J]. International journal of surgery (London, England), 2021, 89: 105948.\u003c/li\u003e\n\u003cli\u003eHan Y, Huang L, Zhou F. A dynamic recursive feature elimination framework (dRFE) to further refine a set of OMIC biomarkers[J]. Bioinformatics (Oxford, England), 2021, 37(15): 2183-2189.\u003c/li\u003e\n\u003cli\u003eAl Abir F, Shovan S M, Hasan M A M, et al. Biomarker identification by reversing the learning mechanism of an autoencoder and recursive feature elimination[J]. Molecular omics, 2022, 18(7): 652-661.\u003c/li\u003e\n\u003cli\u003eZhang Z, Wang S, Zhu Z, et al. Identification of potential feature genes in non-alcoholic fatty liver disease using bioinformatics analysis and machine learning strategies[J]. Computers in biology and medicine, 2023, 157: 106724.\u003c/li\u003e\n\u003cli\u003eAmin M N, Salami B A, Zahid M, et al. Investigating the Bond Strength of FRP Laminates with Concrete Using LIGHT GBM and SHAPASH Analysis[J]. Polymers, 2022, 14(21).\u003c/li\u003e\n\u003cli\u003eMoore A, Bell M. XGBoost, A Novel Explainable AI Technique, in the Prediction of Myocardial Infarction: A UK Biobank Cohort Study[J]. Clinical Medicine Insights Cardiology, 2022, 16: 11795468221133611.\u003c/li\u003e\n\u003cli\u003eSeto H, Oyama A, Kitora S, et al. Gradient boosting decision tree becomes more reliable than logistic regression in predicting probability for diabetes with big data[J]. Scientific reports, 2022, 12(1): 15889.\u003c/li\u003e\n\u003cli\u003eBlanchet L, Vitale R, van Vorstenbosch R, et al. Constructing bi-plots for random forest: Tutorial[J]. Analytica chimica acta, 2020, 1131: 146-155.\u003c/li\u003e\n\u003cli\u003eHao P Y, Chiang J H, Chen Y D. Possibilistic classification by support vector networks[J]. Neural networks, 2022, 149: 40-56.\u003c/li\u003e\n\u003cli\u003eLiu T, Zhang X, Chen R, et al. Development, comparison, and validation of four intelligent, practical machine learning models for patients with prostate-specific antigen in the gray zone[J]. Frontiers in oncology, 2023, 13: 1157384.\u003c/li\u003e\n\u003cli\u003eLi X, Zhao Y, Zhang D, et al. Development of an interpretable machine learning model associated with heavy metals\u0026apos; exposure to identify coronary heart disease among US adults via SHAP: Findings of the US NHANES from 2003 to 2018[J]. Chemosphere, 2023, 311(Pt 1): 137039.\u003c/li\u003e\n\u003cli\u003eHu C, Li L, Huang W, et al. Interpretable Machine Learning for Early Prediction of Prognosis in Sepsis: A Discovery and Validation Study[J]. Infectious diseases and therapy, 2022, 11(3): 1117-1132.\u003c/li\u003e\n\u003cli\u003eEjiyi C J, Qin Z, Ukwuoma C C, et al. Comparative performance analysis of Boruta, SHAP, and Borutashap for disease diagnosis: A study with multiple machine learning algorithms[J]. Network (Bristol, England), 2024: 1-38.\u003c/li\u003e\n\u003cli\u003eRhee C, Dantes R, Epstein L, et al. Incidence and Trends of Sepsis in US Hospitals Using Clinical vs Claims Data, 2009-2014[J]. Jama, 2017, 318(13): 1241-1249.\u003c/li\u003e\n\u003cli\u003eValik J K, Hedberg P, Holmberg F, et al. Impact of the COVID-19 pandemic on the incidence and mortality of hospital-onset bloodstream infection: a cohort study[J]. BMJ quality \u0026amp; safety, 2022, 31(5): 379-382.\u003c/li\u003e\n\u003cli\u003eKontula K S K, Skogberg K, Ollgren J, et al. Population-Based Study of Bloodstream Infection Incidence and Mortality Rates, Finland, 2004-2018[J]. Emerging infectious diseases, 2021, 27(10): 2560-9.\u003c/li\u003e\n\u003cli\u003eZoabi Y, Kehat O, Lahav D, et al. Predicting bloodstream infection outcome using machine learning[J]. Scientific reports, 2021, 11(1): 20101.\u003c/li\u003e\n\u003cli\u003eAn R, Ou Y, Pang L, et al. Epidemiology and Risk Factors of Community-Associated Bloodstream Infections in Zhejiang Province, China, 2017-2020[J]. Infection and drug resistance, 2023, 16: 1579-1590.\u003c/li\u003e\n\u003cli\u003eFlorin T A, Ambroggio L, Brokamp C, et al. Biomarkers and Disease Severity in Children With Community-Acquired Pneumonia[J]. Pediatrics, 2020, 145(6).\u003c/li\u003e\n\u003cli\u003eTang Y H, Jeng M J, Wang H H, et al. Risk factors and predictive markers for early and late-onset neonatal bacteremic sepsis in preterm and term infants[J]. Journal of the Chinese Medical Association : JCMA, 2022, 85(4): 507-513.\u003c/li\u003e\n\u003cli\u003eYang X, Zeng J, Yu X, et al. PCT, IL-6, and IL-10 facilitate early diagnosis and pathogen classifications in bloodstream infection[J]. Annals of clinical microbiology and antimicrobials, 2023, 22(1): 103.\u003c/li\u003e\n\u003cli\u003eZhu Q, Li H, Zheng S, et al. IL-6 and IL-10 Are Associated With Gram-Negative and Gram-Positive Bacteria Infection in Lymphoma[J]. Frontiers in immunology, 2022, 13: 856039.\u003c/li\u003e\n\u003cli\u003eNiu D, Huang Q, Yang F, et al. Serum biomarkers to differentiate Gram-negative, Gram-positive and fungal infection in febrile patients[J]. Journal of medical microbiology, 2021, 70(7).\u003c/li\u003e\n\u003cli\u003eO\u0026apos;Grady N P, Alexander E, Alhazzani W, et al. Society of Critical Care Medicine and the Infectious Diseases Society of America Guidelines for Evaluating New Fever in Adult Patients in the ICU[J]. Critical care medicine, 2023, 51(11): 1570-1586.\u003c/li\u003e\n\u003cli\u003eDoman M, Thy M, Dessajan J, et al. Temperature control in sepsis[J]. Frontiers in medicine, 2023, 10: 1292468.\u003c/li\u003e\n\u003cli\u003eXie Y, Yang Y, Han Y, et al. Association Between Arterial Blood Gas Variation and Intraocular Pressure in Healthy Subjects Exposed to Acute Short-Term Hypobaric Hypoxia[J]. Translational vision science \u0026amp; technology, 2019, 8(6): 22.\u003c/li\u003e\n\u003cli\u003eOuyang S M, Zhu H Q, Xie Y N, et al. Temporal changes in laboratory markers of survivors and non-survivors of adult inpatients with COVID-19[J]. BMC infectious diseases, 2020, 20(1): 952.\u003c/li\u003e\n\u003cli\u003eChoi M H, Kim D, Park Y, et al. Development and validation of artificial intelligence models to predict urinary tract infections and secondary bloodstream infections in adult patients[J]. Journal of infection and public health, 2024, 17(1): 10-17.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-medical-informatics-and-decision-making","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"midm","sideBox":"Learn more about [BMC Medical Informatics and Decision Making](http://bmcmedinformdecismak.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/midm/default.aspx","title":"BMC Medical Informatics and Decision Making","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Bloodstream Infection, Risk Prediction, Real-world, Model Construction","lastPublishedDoi":"10.21203/rs.3.rs-5859635/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5859635/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eBloodstream Infection (BSI) is a severe systemic infectious disease that can lead to sepsis and Multiple Organ Dysfunction Syndrome (MODS), resulting in high mortality rates and posing a major public health burden globally. Early identification of BSI is crucial for effective intervention, reducing mortality, and improving patient outcomes. However, existing diagnostic methods are flawed by low specificity, long detection times and high demands on testing platforms. The development of artificial intelligence provides a new approach for early disease identification. This study aims to explore the optimal combination of routine laboratory data and clinical monitoring indicators, and to utilize machine learning algorithms to construct an early, rapid, and universally applicable BSI risk prediction model, to assist in the early diagnosis of BSI in clinical practice.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eClinical data of 2582 suspected BSI patients admitted to the Chongqing University Central Hospital, from January 1, 2021 to December 31, 2023 were collected for this study. The data were divided into a modeling dataset and an external validation dataset based on chronological order, while the modeling dataset was further divided into a training set and an internal validation set. The occurrence rate of BSI, distribution of pathogens, and microbial primary reporting time were analyzed within the training set. During the feature selection stage, univariate regression and ML algorithms were applied. First, Univariate logistic regression was used to screen for predictive factors of BSI. Then, the Boruta algorithm, Lasso regression, and Recursive Feature Elimination with Cross-validation (RFE-CV) were employed to determine the optimal combination of predictors for predicting BSI. Based on the optimal combination, six machine learning algorithms were used to construct an early BSI risk prediction model. The best model was selected by models\u0026rsquo; performance, and the Shapley Additive Explanations (SHAP) method was used to explain the model. The external validation set was used to evaluate the predictive performance and generalizability of the selected model, and the research findings were ultimately applied in clinical practice.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eThe incidence of BSI among inpatients at the Chongqing University Central Hospital was 12.91%. Following further feature selection, a set of 5 variables was determined, including white blood cell count, standard bicarbonate, base excess of extracellular fluid, interleukin-6, and body temperature. BSI early risk prediction models were constructed using six machine learning algorithms, with the XGBoost model demonstrating the best performance, achieving an AUC value of 0.782 in the internal validation set and an AUC value of 0.776 in the external validation set. This model is made publicly available as an online webpage tool for clinical use.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eThis study successfully identified a set of 5 features by analyzing routine laboratory data clinical monitoring indicators among hospitalized patients. Based on this set, a machine learning-based early risk prediction model for BSI was constructed. The model is capable of early and rapid differentiation between BSI and non-BSI patients. The inclusion of minimal risk prediction factors enhances its applicability in clinical settings, particularly at the primary care level. To further improve the model\u0026rsquo;s real-world applicability and more convenient for clinical use, the online application of the model could greatly improve the efficiency of BSI diagnosis and reducing patients\u0026rsquo; mortality.\u003c/p\u003e","manuscriptTitle":"Development and Application of an Early Prediction Model for Risk of Bloodstream Infection based on Real-world Study","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-02-03 09:14:41","doi":"10.21203/rs.3.rs-5859635/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-02-03T11:37:00+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-02-03T11:25:33+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-01-27T12:58:54+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Medical Informatics and Decision Making","date":"2025-01-19T13:06:59+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-medical-informatics-and-decision-making","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"midm","sideBox":"Learn more about [BMC Medical Informatics and Decision Making](http://bmcmedinformdecismak.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/midm/default.aspx","title":"BMC Medical Informatics and Decision Making","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"a1f5417c-c435-4317-8b22-447d292d5dbb","owner":[],"postedDate":"February 3rd, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2025-05-19T16:09:43+00:00","versionOfRecord":{"articleIdentity":"rs-5859635","link":"https://doi.org/10.1186/s12911-025-03020-9","journal":{"identity":"bmc-medical-informatics-and-decision-making","isVorOnly":false,"title":"BMC Medical Informatics and Decision Making"},"publishedOn":"2025-05-14 15:57:02","publishedOnDateReadable":"May 14th, 2025"},"versionCreatedAt":"2025-02-03 09:14:41","video":"","vorDoi":"10.1186/s12911-025-03020-9","vorDoiUrl":"https://doi.org/10.1186/s12911-025-03020-9","workflowStages":[]},"version":"v1","identity":"rs-5859635","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5859635","identity":"rs-5859635","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}