Construction and validation of prognostic nomogram for hospital-acquired CRE colonization among ICU patients based on ADASYN algorithm

Construction and validation of prognostic nomogram for hospital-acquired CRE colonization among ICU patients based on ADASYN algorithm | 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 Construction and validation of prognostic nomogram for hospital-acquired CRE colonization among ICU patients based on ADASYN algorithm Jianshui Yang, Shuhua Chen, Xiaowem Gong, Zhiping Qi, Xiaona Yin, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6746525/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Objectives To construct a prognostic model for patients with high risk of Carbapenem-resistant Enterobacteriaceae (CRE) colonization in Intensive Care Unit (ICU). Methods A prospective cohort study was performed on 724 ICU patients from Mar 2023 to Jul 2024 at a tertiary hospital. Totally 577 patients were involved by selection and divided into the modeling (n=392) and validation (n=185) cohorts. Multivariate analysis based on adaptive synthetic sampling (ADASYN) was conducted to estimate the risk of CRE colonization, then presented with a visible nomogram. Next, the performances of nomogram were assessed by area under the receiver operating characteristic curve (AUROC), calibration curves, decision curve analysis (DCA), clinical impact curve (CIC). Results The incidence rate of CRE colonization was 10.05% among ICU patients. Multivariate analysis showed Enterobacteriaceae infection (D), days of carbapenem antibiotics (X 1 ) and mechanical ventilation (X 2 ) were independent risk factors of CRE colonization. Logistic regressions based on original data and ADASYN algorithm were described as follows: Logit( P 1 )=1.824D 1 +21.604D 2 +2.482X 1 +1.088X 2 -8.039 and Logit( P 2 )=2.754D 1 +24.355D 2 +2.824X 1 +1.577X 2 -7.057. Internal validation showed ADASYN algorithm model has a higher goodness-of-fit compared with original data model without reducing the discrimination ( P =0.818). DCA of ADASYN algorithm model indicated the threshold risk of positive net benefit ranged 0 to 75%. CIC analysis verified ADASYN algorithm model possessed significant predictive value. A well discrimination with AUROC value of 0.891 (95%CI:0.786~0.995) and goodness-of-fit ( P =0.197) was demonstrated by external validation. Conclusions ADASYN algorithm model can identify patients with high risk of CRE colonization. Optimizing the usage of carbapenem and mechanical ventilation may reduce the risk of CRE colonization. Prognostic nomogram Hospital-acquired CRE colonization ICU patients ADASYN algorithm Figures Figure 1 Figure 2 Figure 3 Figure 4 Introduction With the domestic escalation of antibiotic resistance of Enterobacteriaceae, the resistance rate to carbapenem in China has increased to above 11% so far according to CHINET website[1]. Carbapenem-resistant Enterobacteriaceae (CRE) is highly pathogenic, infectious and difficult to treat[2-3], which increases the economic burden and medical resources waste[4-5]. The transmission of CRE is concealed, and CRE infected people are only a pindling part of carriers. Only 1 of every 100 carriers is identified as an infestor[6]. Patients with CRE colonization have no symptomatic infection. However, this population is the main source of infection by expelling bacteria to external environment on one’s own[7-8]. Therefore, screening for CRE carriers in high risk departments is a standard practice at many medical institutes[9]. But the way of extensive screening seemed inapplicable in general departments in consideration of low positive rate and cost-effectiveness[10], which led to oversights in prevention and control of CRE. Previous study[11] indicated long duration of carbapenem antibiotics and ventilator, and intestinal bacterial infection significantly increased the risk of CRE colonization among Intensive Care Unit (ICU) patients. Due to the low colonization rate of 18.79%, there is an imbalance between colonized and non-colonized population. Traditional statistical analysis have poor predictive effect on imbalanced datasets, with overrating the prediction of majority categories and underrating the prediction of minority categories[12-13], which is not suitable for postoperative model construction of hospital-acquired CRE colonization. Adaptive synthetic sampling (ADASYN) algorithm was developed to ameliorate inclination of imbalanced datasets[14]. The essential idea of ADASYN algorithm is to assign weights to different minority class examples according to the learning difficulty. In detail, the higher weight was inclined to minority class examples with higher learning difficulty. By reducing the learning bias caused by imbalance distribution of original data, ADASYN algorithm can achieve a relatively comprehensive dataset. Previous studies have successfully established models for predicting many diseases based on ADASYN algorithm,including breast cancer[15] and diabetes[16]. To identify high risk patients and provide guidance for the prevention of CRE infection, a prognostic nomogram for hospital-acquired CRE colonization based on ADASYN algorithm was constructed and validated by a prospective cohort study. Participants and methods Patient selection and flowchart of the study A prospective cohort study was performed on totally 724 ICU inpatients from Mar 2023 to Jul 2024 at a tertiary hospital. The inclusion criteria was described as follows: (1) Patients undergoing CRE screening ≥ 2 times; (2) Patients transferred to and out of ICU. And on this basis patients with CRE positive at first screening were excluded. Finally, 577 patients processed to the eligibility criteria shown in the flow diagram (Supplementary Fig. 1), which were divided into the modeling (n = 392) and validation (n = 185) cohorts. This study has been approved by the Ethics Committee of the hospital [Approval ID:2023 (SR) No.030]. Data collection and sample size Based on the literature review and expert consultations, the study included 19 risk factors for data collection. The demographic data prior to the last CRE screening, including age, gender, underlying diseases, malignancies, radiotherapeutic and chemotherapeutical history, different admission to hospital, acute physiology and chronic health evaluation (APACHE II) score, infections status, antibiotics service condition, and invasive procedures were collected by the Zhongzhi Real-Time Nosocomial Infection System. The sample size was evaluated according to the incidence rate of CRE colonization reported by a previous study[ 17 ] in the general ICU at a large grade-A tertiary hospital. According to the formula of smallest sample size = numbers of predictor variables×5 ~ 10×(1 + 15 ~ 20%)/the incidence rate of CRE colonization, 304 ~ 634 samples in the modeling cohort could ensure the validity of the study results. ADASYN algorithm ADASYN algorithm was performed with an R package for ‘Utility-Based Learning’ (UBL). The multiple of over-sampling for CRE colonized population in the modeling cohort was 8.3(350/42) with the number ramping up to 348. Screening strategy and bacteria identification Anal specimens were collected for culture and bacteria identification. The collection was carried out after admission to the ICU and every 7 days until a positive result or consecutively 5 negative results, or discharge. The bacteria were cultured and isolated by professional microbiological technicians, and identification was conducted by MicroScan WalkAway 96 plus. The susceptibility tests were determined by the double disk synergistic method and the double disk synergistic method and the results were determined in accordance with the standards[ 18 ] amended by the Clinical and Laboratory Standards Institute (CLSI) in 2023. Standard bacterial strains of all examined bacteria were incubated, as quality control. Standard bacterial strains including Escherichia coli (ATCC25922), Pseudomonas aeruginosa (ATCC27853), and Staphylococcus aureus (ATCC52923). Statistical analysis Data were summarized with Excel 2017 and statistical analysis was performed by R 4.3.3. Enterobacteriaceae infection was converted into 3 dummy variables, where D 0 ~ D 2 mean Enterobacteriaceae uninfection, Carbapenem-susceptible Enterobacteriaceae (CSE) and CRE infection, respectively. The days of carbapenem antibiotics and mechanical ventilation were classified into 4 categorical by one-week block. The predictive model was presented with a nomogram by the modeling cohort to provide a visual point system to estimate the probability of hospital-acquired CRE colonization. Area under the receiver operating characteristic curve (AUROC) analysis was used to evaluate discriminative ability of the prediction model. Calibration plots were used to evaluate calibrating ability. Hosmer-Lemeshow (H-L) goodness-of-fit test was used to evaluate the model’s fit. Decision curve analysis (DCA) was used to assess the clinical validity and net benefit of the nomogram model. Clinical impact curve (CIC) analysis were plotted to evaluate the clinical applicability and net benefits of the model with the best diagnostic value. AUROC value varies from 0.5 to 1.0, where 0.5 represents random chance and 1.0 indicates a perfect fit. Typically, AUROC values greater than 0.75 suggest a reasonable estimation. A large P value (> 0.05) of H-L goodness-of-fit test indicates good calibration. P value ≤ 0.05 was considered as statistical difference. Results Demographic characteristic Among the 577 candidates, CRE positive was observed in 58 patients, with an incidence rate of 10.05%. The 392 patients in the modeling cohort were divided into the colonized group with 42 cases and non-colonized group with 350 controls. Among 185 patients in the validation group, 16 patients were verified as CRE positive. The demographic characteristic of participants in the modeling cohort were described as previously. Prognostic nomogram for hospital-acquired CRE colonization based on ADASYN algorithm Among the modeling cohort, any variables with statistical significance in the univariate test were selected in the multivariate analysis, and only 3 variables including Enterobacteriaceae infection, the days of carbapenem antibiotics and mechanical ventilation were retained. Regarding the occurrence of CRE colonization as the dependent variable, and Enterobacteriaceae infection (D), the days of carbapenem antibiotics (X 1 ) and mechanical ventilation (X 2 ) as predictor variables, the logistic regressions based on original data and ADASYN algorithm were constructed as follows: Logit( P 1 ) = 1.824D 1 + 21.604D 2 + 2.482X 1 + 1.088X 2 -8.039 and Logit( P 2 ) = 2.754D 1 + 24.355D 2 + 2.824 X 1 + 1.577X 2 -7.057. Prognostic nomograms for hospital-acquired CRE colonization based on original data and ADASYN algorithm were shown as Fig. 1 . The darker color means higher risk of hospital-acquired CRE colonization. Comparison of warning efficacy in the internal validation cohort The results of AUROC analysis were shown in Fig. 2 . Both the AUROC value of the original data model and the ADASYN algorithm model in the internal validation cohort were 0.981 (all 95%CI:0.968 ~ 0.994, P = 0.818), which showed excellent discrimination. The ADASYN algorithm nomogram model had a lower sensitivity (92.60% vs 94.30%), a higher specificity (95.20% vs 92.90%) and positive predictive value (PPV, 69.84% vs 61.44%) compared with the original data model. P value of H-L goodness-of-fit test were 0.415 and 0.533 respectively, which indicated well calibration perfermance of the two prognostic nomograms in the internal validation cohort. However, as shown in Fig. 3 , the apparent calibration curve of the ADASYN algorithm model was closer to the 45° ideal line compared to the original data model, indicating that the algorithm strengthened the consistence between observed probability and predicted probability. DCA and CIC of the ADASYN algorithm model The depicted DCA was used to determine whether decisions based on the optimal predictive nomogram had clinical applicability compared to the default strategy. The analysis provide insight into the range of predicted risk for which the model has a high net benefit than simply either treating all (slope line) patients versus treating no (horizontal line) patient, that is to say, a prediction model is only useful at the threshold risk. The graphically DCA based on the internal validation cohort indicated the expected net benefit (blue curve) of the ADASYN algorithm prediction model. Within the threshold risk range of 0 ~ 75%, intervention decisions based on the optimal predictive nomogram are significantly beneficial (Fig. 4 A). The clinical effectiveness of the optimal predictive model was demonstrated by CIC analysis. CIC visually showed that the nomogram had a superior overall net benefit within a wide and practical range of threshold probabilities, which indicated that the risk model possessed significant predictive value. Besides, the number of positive cases predicted by the ADASYN algorithm model was highly matched with the number of true-positive cases when the threshold probability was above 63% (Fig. 4 B). External validation of the ADASYN algorithm model The external validation analysis of the ADASYN algorithm model was carried out on 185 ICU patients. As shown in Table 1 , AUROC value of the ADASYN algorithm model in the external validation cohort were 0.891 (95%CI:0.789 ~ 0.993, P < 0.001). 14 and 149 patients were predicted as CRE positive and negative by the ADASYN algorithm model among the 16 true positive and 169 true negative patients with a sensitivity of 87.50%, and a specificity of 88.17%. Positive predictive value (PPV) of the model was 41.18(14/34). P value of H-L goodness-of-fit test in the external validation cohort was 0.197. Table 1 Performances of the ADASYN algorithm model in external validation cohort CRE colonized patients CRE non-colonized patients Predicted value 34 151 Actual value 16 169 Corrrect value 14 149 Sensitivity(%) 88.17 Specificity(%) 87.50 PPV(%) 41.18 AUROC(95%CI) 0.891(0.789, 0.993) P value of H-L test 0.197 ADASYN, adaptive synthetic sampling; CRE, Carbapenem-resistant Enterobacteriaceae; PPV, positive predictive value; AUROC, area under of the receiver operating characteristic curve; H-L, Hosmer-Lemeshow. Discussion Intestinal CRE colonization, which is associated with profound endogenous infection and substantial economic burden[ 19 – 21 ], is no longer defined as a simply event. A matched case-control study[ 22 ] at a tertiary hospital in northern Israel showed that CRE colonized patients had 1.7 times odds of clinical infection of any cause and increased length of hospital stay compared to controls among 1-year survivors. Progressive exacerbation of CRE colonization lead to serious infection and heavy burden, and CRE screening could identify carriers and forestall further poor progression. Unfortunately, the fact was that screening for CRE carriers had not formulated a popular practice in international or domestic large-scale hospitals[ 23 – 24 ] because of low cost-effectiveness. A retrospective descriptive study[ 25 ] in the Medical Microbiology & Parasitology laboratory of Hospital Universiti Sains Malaysia stated that active screening results in significant cost pressures and therefore should be revisited and revised, especially in low resource settings. In the settings of domestic diagnosis related groups (DRGs) and diagnosis-intervention packet (DIP) payment policies[ 26 ], hospitalization expenses control, especially inspection expenditure had become the concern of all public hospitals. Active screening for CRE, which was considered as an unnecessary and unprofitable inspection in most clinicians, was easily ignored. Therefore, there is an urgent need to reconsider a substitutive screening scheme in distinguishing patients with high risk of CRE colonization. This study was carried out to construct a novel nomogram for hospital-acquired CRE colonization among ICU patients based on ADASYN algorithm. The ADASYN algorithm model had good discrimination and goodness-of-fit both in internal and external validation cohorts. The clinical applicability of the nomogram was determined by DCA and CIC analysis, suggesting that intervention decisions based on the predictive model were clearly beneficial within a wide threshold risk range. The incidence of hospital-acquired CRE colonization among ICU patients was 10.05%, lower than the results(35.8%) reported by Garpvall K[ 27 ], which might be related to different administrative regions and hospital scales[ 28 ]. Our previous study[ 11 ] revealed long-term usage of carbapenem antibiotics and mechanical ventilation and intestinal bacterial infection were the independent risk factors of intestinal CRE colonization for ICU patients. According to nomogram constructed with three variables above, particular attention should be paid to patients who have CRE infection and been treated with carbapenems more than 14 days or mechanical ventilation more than 14 days because of higher probability of intestinal colonization. Long-term usage of carbapenem antibiotics and mechanical ventilation might cause disruption and translocation of intestinal microflora, functional impaired of body barrier and advantage growth of multiple drug resistant organisms (MDROs)[ 29 – 30 ]. The occurrence of Enterobacteriaceae bacterial infection mean the disruption and shift of intestinal microbiota, which increased the risk of intestinal CRE colonization. In the modeling cohort, 10.71% of ICU patients were defined as CRE carriers as well the others without CRE colonizing. There was a significant imbalance in the number of case and control group. Traditional early warning model leads to a high error in the predictive accuracy of minority categories because of low screening specificity when the dataset is highly imbalanced[ 31 ]. The statistical analysis method based on the balanced sample size between classes can not achieve satisfactory results. Therefore, this study constructed a new dataset by introducing an oversampling technology called ADASYN algorithm to solve the imbalance of majority and minority categories. ADASYN algorithm, different from other oversampling methods to simply replicate samples, randomly selected one sample for each minority sample in its nearest neighbor sample, and then randomly selects one point on the two sample lines as the newly synthesized sample[ 32 – 33 ]. ADASYN algorithm paid more attention to interclass distance and local density, especially difficult samples or noise sensitive regions in minority categories, which modulate the sampling strategy more finely and avoided the problem of data size expansion and increased model training complexity caused by oversampling. ADASYN algorithm helped to improve the accuracy and generalization ability of models in real-world applications. The results of this study showed ADASYN algorithm strengthened the consistence between observed probability and predicted probability without reducing the discrimination of predictive mode. The consequence of DCA and CIC demonstrated ICU patients would conspicuously benefit from the clinical decisions based on the predictive model within a wide and practical range of threshold probabilities. Clinician could incorporate threshold probabilities as a surrogate of physician preference, which highlighted the superiority of predictive model in supporting clinical decision-making processes[ 34 – 35 ]. What’s more, the parameters of predictive model in the validation cohort were acceptable,which indicated the model was steady, reproducible and generalduty. The results of external validation is dependable because of the normative framework of external validation study design referring to a teaching study[ 36 ]. The strength of this study was mainly that it was the first time to constructed a novel predictive model for identifying patients with high risk of CRE colonization by a prospective cohort study, and meanwhile the model was verified by DCA and CIC compared to traditional regression analysis and clinical scoring system. The prospective study is consistent with temporal logic in causal inference. Moreover, the population who underwent CRE screening in ICU was entirely included in this study and the sample size reached statistical requirement. Nevertheless, we must acknowledge some other limitations of the study. All results were derived from a single center. Besides, the relevant risk factors incorporated in this study may be incomplete, which may lead to the abandonment of some key variables. The predictive model need to be further validated by cohort studies with multicentric data, more comprehensive risk factors and rigorous quality control. Even so, we believed that the proposed model may contribute to promote our understanding of the prevention and control of patients potentially suffering from CRE colonization in ICU. Conclusions In conclusion, this prospective cohort study shows the ADASYN algorithm model does outperform conventional original data model. The ADASYN algorithm model may prove to be clinically useful in precise identification and intervention for patients with high risk of hospital-acquired CRE colonization. Furthermore, by optimizing the usage of carbapenem antibiotics and mechanical ventilation, the risk of CRE colonization may decrease. Declarations Ethics approval and consent to participat e This study has been approved by the ethical committee of Changzhou Cancer Hospital review board [Approval ID:2023 (SR) No.030]. Consent for publication Not applicable. Competing interests The authors declare that they have no competing interests. Fundings This study was supported by the Young Talent Development Program of Changzhou Health Commission (CZQM2023021), Health Young Scientific and Technological Talent Development Program of Changzhou Medical Development Center and Changzhou Medical Association (lcyx2023010), the Applied Basic research Program of Changzhou Technology Division (CJ20241024). Author Contribution Conception and design: QM, SC and JY. Provision of study materials or patients: XG, ZQ, XY. Collection and assembly of data: JY, SC. Data analysis and interpretation: JY, SC and QM. All authors read and approved the final manuscript. Acknowledgement The authors are grateful to the clinical medical institution members, nurses, and physicians to perform this study. Data Availability Data is provided within the manuscript or supplementary information files References Qin X, Ding L, Hao M, et al . Antimicrobial resistance of clinical bacterial isolates in China: current status and trends[J]. JAC Antimicrob Resist, 2024, 6(2):dlae052. Huang N, Jia H, Zhou B, et al . Hypervirulent carbapenem-resistant Klebsiella pneumoniae causing highly fatal meningitis in southeastern China[J]. Front Public Health, 2022, 10:991306. Wu D, Xiao J, Ding J, et al . Predictors of Mortality and Drug Resistance Among Carbapenem-Resistant Enterobacteriaceae-Infected Pancreatic Necrosis Patients[J]. Infect Dis Ther, 2021, 10(3):1665–1676. Zhu Y, Xiao T, Wang Y, et al . Socioeconomic Burden of Bloodstream Infections Caused by Carbapenem-Resistant Enterobacteriaceae[J]. Infect Drug Resist, 2021, 14:5385–5393. Song KH, Kim CJ, Choi NK, et al . Clinical and economic burden of bacteremia due to multidrug-resistant organisms in Korea: a prospective case control study[J]. J Glob Antimicrob Resist, 2022, 31:379–385. Tomczyk S, Zanichelli V, Grayson ML, et al . Control of Carbapenem-resistant Enterobacteriaceae, Acinetobacter baumannii, and Pseudomonas aeruginosa in Healthcare Facilities: A Systematic Review and Reanalysis of Quasi-experimental Studies[J]. Clin Infect Dis, 2019, 68(5):873–884. Wu MM, Yang DH, Pan HY, et al . Risk factors analysis of bloodstream infection caused by colonized carbapenem-resistant Enterobacteriaceae[J]. Zhejiang Med, 2022,44(12):1300–1303. Gomides MDA, Fontes AMS, Silveira AOSM, et al . The importance of active surveillance of carbapenem-resistant Enterobacterales (CRE) in colonization rates in critically ill patients[J]. PLoS One, 2022, 17(1):e0262554. Stein M, Dabaja-Younis H, Kassis I, et al . Evaluation of Pediatric Screening for Resistant Pathogens in an Israeli Tertiary Center[J]. Isr Med Assoc J, 2022, 25(12):836–841. Ho KW, Ng WT, Ip M, et al . Active surveillance of carbapenem-resistant Enterobacteriaceae in intensive care units: Is it cost-effective in a nonendemic region[J]. Am J Infect Control, 2016, 44(4):394–9. Yang JS, Min QF, Gong XW, et al . Pathogenocity and risk factors for intestinal colonization of Carbapenem-resistant Enterobacteriaceae in patients from Intensive Care Unit[J]. Clin J Infect Control, 2024, 23(11):1373–1378. Zheng Y, Zhang C, Liu Y. Risk prediction models of depression in older adults with chronic diseases[J]. J Affect Disord, 2024, 359:182–188. Yu M, Yuan Z, Li R, et al . Interpretable machine learning model to predict surgical difficulty in laparoscopic resection for rectal cancer[J]. Front Oncol, 2024,14:1337219. Hao Y, Li X, Zhang C. Improving prediction model robustness with virtual sample construction for near-infrared spectra analysis[J]. Anal Chim Acta, 2023, 1279:341763. Khan TM, Xu S, Khan ZG, et al . Implementing Multilabeling, ADASYN, and ReliefF Techniques for Classification of Breast Cancer Diagnostic through Machine Learning: Efficient Computer-Aided Diagnostic System[J]. J Healthc Eng, 2021, 2021:5577636. Tasin I, Nabil TU, Islam S, et al . Diabetes prediction using machine learning and explainable AI techniques[J]. Healthc Technol Lett, 2022, 10(1–2):1–10. Liu SD, Wu AH, Li CH, et al . Screening of intestinal carbapenem-resistant Enterobacteriaceae for hospitalized patients of high-risk department and risk factors for positive testing[J]. Chin J Nosocomiol, 2023(04):517–521. Clinical and Laboratory Standards Institute. Performance standards for antimicrobial susceptibility testing[S]. M100-S30.Wayne, PA:CLSI, 2023. Chen HY, Jean SS, Lee YL, et al . Carbapenem-Resistant Enterobacterales in Long-Term Care Facilities: A Global and Narrative Review[J]. Front Cell Infect Microbiol, 2021, 11:601968. Wangchinda W, Thamlikitkul V, Watcharasuwanseree S, et al . Active Surveillance for Carbapenem-Resistant Enterobacterales (CRE) Colonization and Clinical Course of CRE Colonization among Hospitalized Patients at a University Hospital in Thailand[J]. Antibiotics (Basel), 2022, 11(10):1401. Pan F, Chen P, Duan Y, et al . Prevalence of intestinal colonization and nosocomial infection with carbapenem-resistant Enterobacteriales in children: a retrospective study[J]. Front Public Health, 2023, 11:1251609. Hassoun-Kheir N, Hussien K, Karram M, et al . Clinical significance and burden of carbapenem-resistant Enterobacterales colonization acquisition in hospitalized patients[J]. Antimicrob Resist Infect Control, 2023, 12(1):129. Cai M, Zhang H. Research progress for CRE infection in intensive care unit patients[J]. Laboratory Medicine, 2024, 39(09):913–918. Bai HS, Wu XQ, Guo WD, et al . Progress of research on active screening and multidisciplinary collaborative management mode of carbapenem-resistant Enterobacteriaceae[J]. Clin J Nosocomiol, 2023, 33(22):3499–3504. Zaidah AR, Mohammad NI, Suraiya S, et al . High burden of Carbapenem-resistant Enterobacteriaceae (CRE) fecal carriage at a teaching hospital: cost-effectiveness of screening in low-resource setting[J]. Antimicrob Resist Infect Control, 2017, 6:42. Zhu FM. DRG/DIP Payment Reform, Cost Control and Impact Mechanism: A Multi-Period DID Analysis Based on Sampling Datafrom 19 Regions[J]. Chinese Social Security Review, 2023, 7(05):54–67. Garpvall K, Duong V, Linnros S, et al . Admission screening and cohort care decrease carbapenem resistant enterobacteriaceae in Vietnamese pediatric ICU's[J]. Antimicrob Resist Infect Control, 2021, 10(1):128–131. Kotb S, Lyman M, Ismail G, et al . Epidemiology of Carbapenem-resistant Enterobacteriaceae in Egyptian intensive care units using National Healthcare-associated Infections Surveillance Data, 2011–2017[J]. Antimicrob Resist Infect Control, 2020, 9(1):2–6. Zimmermann P, Curtis N. The effect of antibiotics on the composition of the intestinal microbiota - a systematic review[J]. J Infect, 2019, 79(6):471–489. Ge W, Wei W, Shuang P, et al . Nasointestinal Tube in Mechanical Ventilation Patients is More Advantageous[J]. Open Med (Wars), 2019, 14:426–430. Li Y, Xu W, Li W, et al . Research on hybrid intrusion detection method based on the ADASYN and ID3 algorithms[J]. Math Biosci Eng, 2022, 19(2):2030–2042. Ni Z, Zhu Y, Qian Y, et al . Synthetic minority over-sampling technique-enhanced machine learning models for predicting recurrence of postoperative chronic subdural hematoma[J]. Front Neurol, 2024, 15:1305543. Zhang W, Liu J, Jin W, et al . Radiomics from dual-energy CT-derived iodine maps predict lymph node metastasis in head and neck squamous cell carcinoma[J]. Radiol Med, 2024, 129(2):252–267. Vickers AJ, Holland F. Decision curve analysis to evaluate the clinical benefit of prediction models[J]. Spine J, 2021, 21(10):1643–1648. Zhao L, Leng Y, Hu Y, et al . Understanding decision curve analysis in clinical prediction model research[J]. Postgrad Med J, 2024, 100(1185):512–515. Ramspek CL, Jager KJ, Dekker FW, et al . External validation of prognostic models: what, why, how, when and where[J]. Clin Kidney J, 2020, 14(1):49–58. Additional Declarations No competing interests reported. Supplementary Files SupplementaryFigure1.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6746525","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":471954778,"identity":"a571241a-7f50-4169-ac53-a24c40c730c9","order_by":0,"name":"Jianshui Yang","email":"","orcid":"","institution":"Department of Infection Control, Changzhou Cancer Hospita","correspondingAuthor":false,"prefix":"","firstName":"Jianshui","middleName":"","lastName":"Yang","suffix":""},{"id":471954784,"identity":"56e2928d-47d6-40cd-ab42-9ab77695baf5","order_by":1,"name":"Shuhua Chen","email":"","orcid":"","institution":"Intensive Care Unit, Changzhou Cancer Hospital","correspondingAuthor":false,"prefix":"","firstName":"Shuhua","middleName":"","lastName":"Chen","suffix":""},{"id":471954785,"identity":"30b0e36a-0cf0-42cf-834f-1f3cc44792fc","order_by":2,"name":"Xiaowem Gong","email":"","orcid":"","institution":"Department of Infection Control, Changzhou Cancer Hospita","correspondingAuthor":false,"prefix":"","firstName":"Xiaowem","middleName":"","lastName":"Gong","suffix":""},{"id":471954786,"identity":"3670bc68-085d-452d-b3f9-eec501a94144","order_by":3,"name":"Zhiping Qi","email":"","orcid":"","institution":"Department of Infection Control, Changzhou Cancer Hospita","correspondingAuthor":false,"prefix":"","firstName":"Zhiping","middleName":"","lastName":"Qi","suffix":""},{"id":471954787,"identity":"dbf97da9-eb78-44f7-bb0e-8052ce178f89","order_by":4,"name":"Xiaona Yin","email":"","orcid":"","institution":"Intensive Care Unit, Changzhou Cancer Hospital","correspondingAuthor":false,"prefix":"","firstName":"Xiaona","middleName":"","lastName":"Yin","suffix":""},{"id":471954788,"identity":"94e8e602-25f2-4233-93a0-b421c7c214b0","order_by":5,"name":"Qifen Min","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA0ElEQVRIiWNgGAWjYBACfvnHBww+VNjU2x9vIFKLZENaQuGMM2kJDGcOEKnF4ECOwWfelsMJDDcSiHXZgQOGG2c2MOcxzny88QZDjU00QR2MjQ3JBh93sBUzS6cVWzAcS8ttIKSFmZnhmOHMMzyMbdI5ZhKMDYcJa2FjY2z/zdsmwdgjeYZILTw8zAzGvG0GiTMkeIjUIiHBxmA440yCsQEP0C8JxPjF/gb/B2BU/pczYD+88caHGhvCWpCBgUQCKcohWkjVMQpGwSgYBSMDAACXtkFqznrOYQAAAABJRU5ErkJggg==","orcid":"","institution":"Department of Infection Control, Changzhou Cancer Hospita","correspondingAuthor":true,"prefix":"","firstName":"Qifen","middleName":"","lastName":"Min","suffix":""}],"badges":[],"createdAt":"2025-05-26 03:08:14","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6746525/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6746525/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":85346568,"identity":"46f92349-649a-4fbe-89d5-82666ce5d23f","added_by":"auto","created_at":"2025-06-25 02:11:14","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":204940,"visible":true,"origin":"","legend":"\u003cp\u003eNomograms for the prediction of hospital-acquired CRE colonization based on original data (left) and Adasyn algorithm (right) in the modeling cohort. Predictor variables including Enterobacteriaceae infection, The days of carbapenem antibiotics and mechanical ventilation. Enterobacteriaceae infection was converted into 3 dummy variables including Enterobacteriaceae uninfection, CSE and CRE infection. The days of carbapenem antibiotics and mechanical ventilation were classified into 4 categorical by one-week block. The score for a patient was measured by summation of color depth assigned to each factor. The patient’s risk was determined by drawing a perpendicular line depending on overall color depth. CRE, Carbapenem-resistant Enterobacteriaceae; Adasyn, adaptive synthetic sampling; CSE, Carbapenem-susceptible Enterobacteriaceae.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-6746525/v1/a266180b0de83af2b9901e21.png"},{"id":85347210,"identity":"523f4eef-88d1-468c-8f92-60c074c0916b","added_by":"auto","created_at":"2025-06-25 02:19:14","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":214394,"visible":true,"origin":"","legend":"\u003cp\u003eComparison of discrimination by AUROC of the purposed nomograms for predicting hospital-acquired CRE colonization based on original data (left) and Adasyn algorithm (right) in the internal validation cohort. \u003cem\u003eP \u003c/em\u003evalue of heterogeneity test for AUROC was 0.818, which meant undifferentiated discrimination. AUROC, area under of the receiver operating characteristic curve; CRE, Carbapenem-resistant Enterobacteriaceae; Adasyn, adaptive synthetic sampling.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-6746525/v1/81e5cf67da5b92c899bc2bb6.png"},{"id":85345608,"identity":"6a5c9bab-9250-40a6-b7b8-e84cf9d2ff5b","added_by":"auto","created_at":"2025-06-25 02:03:14","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":198475,"visible":true,"origin":"","legend":"\u003cp\u003eCalibration curve of the purposed nomograms for predicting hospital-acquired CRE colonization based on original data (left) and Adasyn algorithm (right) in the internal validation cohort. The 45° ideal line stands for perfect prediction. The other dotted line represents estimations of predictive model vs observed values meanwhile the solid line (on behalf of bias) shows the corrected estimates via employing 1000 bootstrap samples. CRE, Carbapenem-resistant Enterobacteriaceae; Adasyn, adaptive synthetic sampling.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-6746525/v1/b937001dc5d96e5515d252fa.png"},{"id":85345612,"identity":"67265ca4-8dc6-4e0c-aaa8-ba660e6dc614","added_by":"auto","created_at":"2025-06-25 02:03:14","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":187384,"visible":true,"origin":"","legend":"\u003cp\u003eThe results of DCA (A) and CIC (B) of the Adasyn algorithm model for predicting hospital-acquired CRE colonization in the internal validation cohort. \u003cstrong\u003e(A)\u003c/strong\u003e The DCA curve of the Adasyn algorithm model. The curve revealed the purposed nomogram possessed a higher net benefit than treating either no or all patients within a expansive threshold risk range of 0~75%. \u003cstrong\u003e(B)\u003c/strong\u003e The CIC curve of the Adasyn algorithm model. The red curve (the number of individuals at high risk) indicates the number of persons who are classified as positive (high risk) by the predictive model at each threshold probability, and the blue curve (the number of individuals at high risk with outcomes) is the number of true positives at each threshold probability. DCA, decision curve analysis; CIC, clinical impact curve; CRE, Carbapenem-resistant Enterobacteriaceae; Adasyn, adaptive synthetic sampling.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-6746525/v1/fe4067763a00f811aec04382.png"},{"id":90639792,"identity":"e78517f5-6c75-4e33-aa8d-8972e0dc05c9","added_by":"auto","created_at":"2025-09-05 06:17:07","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1525886,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6746525/v1/24fa6788-5ade-4e52-b5a3-427d33f79719.pdf"},{"id":85345605,"identity":"42c909f5-e9d3-4f50-9d22-1ad104adf609","added_by":"auto","created_at":"2025-06-25 02:03:14","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":240106,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryFigure1.docx","url":"https://assets-eu.researchsquare.com/files/rs-6746525/v1/923b4ff948c9b1e192bbdbe7.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Construction and validation of prognostic nomogram for hospital-acquired CRE colonization among ICU patients based on ADASYN algorithm","fulltext":[{"header":"Introduction","content":"\u003cp\u003eWith the domestic escalation of antibiotic resistance of Enterobacteriaceae, the resistance rate to\u0026nbsp;carbapenem in China has increased to above 11% so far according to CHINET website[1].\u0026nbsp;Carbapenem-resistant Enterobacteriaceae (CRE) is highly pathogenic, infectious and difficult to treat[2-3], which increases the economic burden and medical resources waste[4-5]. The transmission of CRE is concealed, and CRE infected people are only a pindling part of carriers. Only 1 of every 100 carriers is identified as an infestor[6]. Patients with CRE colonization have no symptomatic infection. However, this population is the main source of infection by expelling bacteria to external environment on one’s own[7-8]. Therefore, screening for CRE carriers in high risk departments is a standard practice at many medical institutes[9]. But the way of extensive screening seemed inapplicable in general departments in consideration of low positive rate and cost-effectiveness[10], which led to oversights in prevention and control of CRE. Previous study[11] indicated long duration of carbapenem antibiotics and ventilator, and intestinal bacterial infection significantly increased the risk of CRE colonization among Intensive Care Unit (ICU) patients. Due to the low colonization rate of 18.79%, there is an imbalance between colonized and non-colonized population. Traditional statistical analysis have poor predictive effect on imbalanced datasets, with overrating the prediction of majority categories and underrating the prediction of minority categories[12-13], which is not suitable for postoperative model construction of\u0026nbsp;hospital-acquired CRE colonization.\u0026nbsp;Adaptive synthetic sampling (ADASYN) algorithm was developed to ameliorate inclination of imbalanced datasets[14]. The essential idea of\u0026nbsp;ADASYN algorithm\u0026nbsp;is\u0026nbsp;to assign weights to different minority class examples according to the learning difficulty. In detail, the higher weight was inclined to minority class examples with higher learning difficulty. By reducing the learning bias caused by imbalance distribution of original data, ADASYN algorithm can achieve a relatively comprehensive dataset. Previous studies have successfully established models for predicting many diseases based on ADASYN algorithm,including breast cancer[15] and diabetes[16].\u003c/p\u003e\n\u003cp\u003eTo identify high risk patients and provide guidance for the prevention of CRE infection, a \u0026nbsp;prognostic nomogram for hospital-acquired CRE colonization based on ADASYN algorithm was constructed and validated by a prospective cohort study.\u003c/p\u003e"},{"header":"Participants and methods","content":"\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003ePatient selection and flowchart of the study\u003c/h2\u003e \u003cp\u003eA prospective cohort study was performed on totally 724 ICU inpatients from Mar 2023 to Jul 2024 at a tertiary hospital. The inclusion criteria was described as follows: (1) Patients undergoing CRE screening\u0026thinsp;\u0026ge;\u0026thinsp;2 times; (2) Patients transferred to and out of ICU. And on this basis patients with CRE positive at first screening were excluded.\u003c/p\u003e \u003cp\u003eFinally, 577 patients processed to the eligibility criteria shown in the flow diagram (Supplementary Fig.\u0026nbsp;1), which were divided into the modeling (n\u0026thinsp;=\u0026thinsp;392) and validation (n\u0026thinsp;=\u0026thinsp;185) cohorts. This study has been approved by the Ethics Committee of the hospital [Approval ID:2023 (SR) No.030].\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eData collection and sample size\u003c/h3\u003e\n\u003cp\u003eBased on the literature review and expert consultations, the study included 19 risk factors for data collection. The demographic data prior to the last CRE screening, including age, gender, underlying diseases, malignancies, radiotherapeutic and chemotherapeutical history, different admission to hospital, acute physiology and chronic health evaluation (APACHE II) score, infections status, antibiotics service condition, and invasive procedures were collected by the Zhongzhi Real-Time Nosocomial Infection System.\u003c/p\u003e \u003cp\u003eThe sample size was evaluated according to the incidence rate of CRE colonization reported by a previous study[\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e] in the general ICU at a large grade-A tertiary hospital. According to the formula of smallest sample size\u0026thinsp;=\u0026thinsp;numbers of predictor variables\u0026times;5\u0026thinsp;~\u0026thinsp;10\u0026times;(1\u0026thinsp;+\u0026thinsp;15\u0026thinsp;~\u0026thinsp;20%)/the incidence rate of CRE colonization, 304\u0026thinsp;~\u0026thinsp;634 samples in the modeling cohort could ensure the validity of the study results.\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eADASYN algorithm\u003c/h2\u003e \u003cp\u003eADASYN algorithm was performed with an R package for \u0026lsquo;Utility-Based Learning\u0026rsquo; (UBL). The multiple of over-sampling for CRE colonized population in the modeling cohort was 8.3(350/42) with the number ramping up to 348.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eScreening strategy and bacteria identification\u003c/h3\u003e\n\u003cp\u003eAnal specimens were collected for culture and bacteria identification. The collection was carried out after admission to the ICU and every 7 days until a positive result or consecutively 5 negative results, or discharge. The bacteria were cultured and isolated by professional microbiological technicians, and identification was conducted by MicroScan WalkAway 96 plus. The susceptibility tests were determined by the double disk synergistic method and the double disk synergistic method and the results were determined in accordance with the standards[\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e] amended by the Clinical and Laboratory Standards Institute (CLSI) in 2023. Standard bacterial strains of all examined bacteria were incubated, as quality control. Standard bacterial strains including \u003cem\u003eEscherichia coli\u003c/em\u003e (ATCC25922), \u003cem\u003ePseudomonas aeruginosa\u003c/em\u003e (ATCC27853), and \u003cem\u003eStaphylococcus aureus\u003c/em\u003e (ATCC52923).\u003c/p\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003eStatistical analysis\u003c/h2\u003e \u003cp\u003eData were summarized with Excel 2017 and statistical analysis was performed by R 4.3.3. Enterobacteriaceae infection was converted into 3 dummy variables, where D\u003csub\u003e0\u003c/sub\u003e\u0026thinsp;~\u0026thinsp;D\u003csub\u003e2\u003c/sub\u003e mean Enterobacteriaceae uninfection, Carbapenem-susceptible Enterobacteriaceae (CSE) and CRE infection, respectively. The days of carbapenem antibiotics and mechanical ventilation were classified into 4 categorical by one-week block. The predictive model was presented with a nomogram by the modeling cohort to provide a visual point system to estimate the probability of hospital-acquired CRE colonization. Area under the receiver operating characteristic curve (AUROC) analysis was used to evaluate discriminative ability of the prediction model. Calibration plots were used to evaluate calibrating ability. Hosmer-Lemeshow (H-L) goodness-of-fit test was used to evaluate the model\u0026rsquo;s fit. Decision curve analysis (DCA) was used to assess the clinical validity and net benefit of the nomogram model. Clinical impact curve (CIC) analysis were plotted to evaluate the clinical applicability and net benefits of the model with the best diagnostic value. AUROC value varies from 0.5 to 1.0, where 0.5 represents random chance and 1.0 indicates a perfect fit. Typically, AUROC values greater than 0.75 suggest a reasonable estimation. A large \u003cem\u003eP\u003c/em\u003e value (\u0026gt;\u0026thinsp;0.05) of H-L goodness-of-fit test indicates good calibration. \u003cem\u003eP\u003c/em\u003e value\u0026thinsp;\u0026le;\u0026thinsp;0.05 was considered as statistical difference.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eDemographic characteristic\u003c/h2\u003e \u003cp\u003eAmong the 577 candidates, CRE positive was observed in 58 patients, with an incidence rate of 10.05%. The 392 patients in the modeling cohort were divided into the colonized group with 42 cases and non-colonized group with 350 controls. Among 185 patients in the validation group, 16 patients were verified as CRE positive. The demographic characteristic of participants in the modeling cohort were described as previously.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003ePrognostic nomogram for hospital-acquired CRE colonization based on ADASYN algorithm\u003c/h2\u003e \u003cp\u003eAmong the modeling cohort, any variables with statistical significance in the univariate test were selected in the multivariate analysis, and only 3 variables including Enterobacteriaceae infection, the days of carbapenem antibiotics and mechanical ventilation were retained. Regarding the occurrence of CRE colonization as the dependent variable, and Enterobacteriaceae infection (D), the days of carbapenem antibiotics (X\u003csub\u003e1\u003c/sub\u003e) and mechanical ventilation (X\u003csub\u003e2\u003c/sub\u003e) as predictor variables, the logistic regressions based on original data and ADASYN algorithm were constructed as follows: Logit(\u003cem\u003eP\u003c/em\u003e\u003csub\u003e1\u003c/sub\u003e)\u0026thinsp;=\u0026thinsp;1.824D\u003csub\u003e1\u003c/sub\u003e\u0026thinsp;+\u0026thinsp;21.604D\u003csub\u003e2\u003c/sub\u003e\u0026thinsp;+\u0026thinsp;2.482X\u003csub\u003e1\u003c/sub\u003e\u0026thinsp;+\u0026thinsp;1.088X\u003csub\u003e2\u003c/sub\u003e-8.039 and Logit(\u003cem\u003eP\u003c/em\u003e\u003csub\u003e2\u003c/sub\u003e)\u0026thinsp;=\u0026thinsp;2.754D\u003csub\u003e1\u003c/sub\u003e\u0026thinsp;+\u0026thinsp;24.355D\u003csub\u003e2\u003c/sub\u003e\u0026thinsp;+\u0026thinsp;2.824 X\u003csub\u003e1\u003c/sub\u003e\u0026thinsp;+\u0026thinsp;1.577X\u003csub\u003e2\u003c/sub\u003e-7.057. Prognostic nomograms for hospital-acquired CRE colonization based on original data and ADASYN algorithm were shown as Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. The darker color means higher risk of hospital-acquired CRE colonization.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eComparison of warning efficacy in the internal validation cohort\u003c/h2\u003e \u003cp\u003eThe results of AUROC analysis were shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e. Both the AUROC value of the original data model and the ADASYN algorithm model in the internal validation cohort were 0.981 (all 95%CI:0.968\u0026thinsp;~\u0026thinsp;0.994, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.818), which showed excellent discrimination. The ADASYN algorithm nomogram model had a lower sensitivity (92.60% vs 94.30%), a higher specificity (95.20% vs 92.90%) and positive predictive value (PPV, 69.84% vs 61.44%) compared with the original data model.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cem\u003eP\u003c/em\u003e value of H-L goodness-of-fit test were 0.415 and 0.533 respectively, which indicated well calibration perfermance of the two prognostic nomograms in the internal validation cohort. However, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e, the apparent calibration curve of the ADASYN algorithm model was closer to the 45\u0026deg; ideal line compared to the original data model, indicating that the algorithm strengthened the consistence between observed probability and predicted probability.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eDCA and CIC of the ADASYN algorithm model\u003c/h2\u003e \u003cp\u003eThe depicted DCA was used to determine whether decisions based on the optimal predictive nomogram had clinical applicability compared to the default strategy. The analysis provide insight into the range of predicted risk for which the model has a high net benefit than simply either treating all (slope line) patients versus treating no (horizontal line) patient, that is to say, a prediction model is only useful at the threshold risk. The graphically DCA based on the internal validation cohort indicated the expected net benefit (blue curve) of the ADASYN algorithm prediction model. Within the threshold risk range of 0\u0026thinsp;~\u0026thinsp;75%, intervention decisions based on the optimal predictive nomogram are significantly beneficial (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eA).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe clinical effectiveness of the optimal predictive model was demonstrated by CIC analysis. CIC visually showed that the nomogram had a superior overall net benefit within a wide and practical range of threshold probabilities, which indicated that the risk model possessed significant predictive value. Besides, the number of positive cases predicted by the ADASYN algorithm model was highly matched with the number of true-positive cases when the threshold probability was above 63% (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eB).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eExternal validation of the ADASYN algorithm model\u003c/h2\u003e \u003cp\u003eThe external validation analysis of the ADASYN algorithm model was carried out on 185 ICU patients. As shown in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e, AUROC value of the ADASYN algorithm model in the external validation cohort were 0.891 (95%CI:0.789\u0026thinsp;~\u0026thinsp;0.993, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001). 14 and 149 patients were predicted as CRE positive and negative by the ADASYN algorithm model among the 16 true positive and 169 true negative patients with a sensitivity of 87.50%, and a specificity of 88.17%. Positive predictive value (PPV) of the model was 41.18(14/34). \u003cem\u003eP\u003c/em\u003e value of H-L goodness-of-fit test in the external validation cohort was 0.197.\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\u003ePerformances of the ADASYN algorithm model in external validation cohort\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\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 \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCRE colonized patients\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCRE non-colonized patients\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePredicted value\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e34\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e151\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eActual value\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e16\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e169\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCorrrect value\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e14\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e149\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSensitivity(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e88.17\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSpecificity(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e87.50\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePPV(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e \u003cp\u003e41.18\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAUROC(95%CI)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e \u003cp\u003e0.891(0.789, 0.993)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eP\u003c/em\u003e value of H-L test\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e \u003cp\u003e0.197\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"3\"\u003e\u003cb\u003eADASYN, adaptive synthetic sampling; CRE, Carbapenem-resistant Enterobacteriaceae; PPV, positive predictive value; AUROC, area under of the receiver operating characteristic curve; H-L, Hosmer-Lemeshow.\u003c/b\u003e\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eIntestinal CRE colonization, which is associated with profound endogenous infection and substantial economic burden[\u003cspan additionalcitationids=\"CR20\" citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e], is no longer defined as a simply event. A matched case-control study[\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e] at a tertiary hospital in northern Israel showed that CRE colonized patients had 1.7 times odds of clinical infection of any cause and increased length of hospital stay compared to controls among 1-year survivors. Progressive exacerbation of CRE colonization lead to serious infection and heavy burden, and CRE screening could identify carriers and forestall further poor progression. Unfortunately, the fact was that screening for CRE carriers had not formulated a popular practice in international or domestic large-scale hospitals[\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e] because of low cost-effectiveness. A retrospective descriptive study[\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e] in the Medical Microbiology \u0026amp; Parasitology laboratory of Hospital Universiti Sains Malaysia stated that active screening results in significant cost pressures and therefore should be revisited and revised, especially in low resource settings. In the settings of domestic diagnosis related groups (DRGs) and diagnosis-intervention packet (DIP) payment policies[\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e], hospitalization expenses control, especially inspection expenditure had become the concern of all public hospitals. Active screening for CRE, which was considered as an unnecessary and unprofitable inspection in most clinicians, was easily ignored. Therefore, there is an urgent need to reconsider a substitutive screening scheme in distinguishing patients with high risk of CRE colonization.\u003c/p\u003e \u003cp\u003eThis study was carried out to construct a novel nomogram for hospital-acquired CRE colonization among ICU patients based on ADASYN algorithm. The ADASYN algorithm model had good discrimination and goodness-of-fit both in internal and external validation cohorts. The clinical applicability of the nomogram was determined by DCA and CIC analysis, suggesting that intervention decisions based on the predictive model were clearly beneficial within a wide threshold risk range.\u003c/p\u003e \u003cp\u003eThe incidence of hospital-acquired CRE colonization among ICU patients was 10.05%, lower than the results(35.8%) reported by Garpvall K[\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e], which might be related to different administrative regions and hospital scales[\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]. Our previous study[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e] revealed long-term usage of carbapenem antibiotics and mechanical ventilation and intestinal bacterial infection were the independent risk factors of intestinal CRE colonization for ICU patients. According to nomogram constructed with three variables above, particular attention should be paid to patients who have CRE infection and been treated with carbapenems more than 14 days or mechanical ventilation more than 14 days because of higher probability of intestinal colonization. Long-term usage of carbapenem antibiotics and mechanical ventilation might cause disruption and translocation of intestinal microflora, functional impaired of body barrier and advantage growth of multiple drug resistant organisms (MDROs)[\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]. The occurrence of Enterobacteriaceae bacterial infection mean the disruption and shift of intestinal microbiota, which increased the risk of intestinal CRE colonization.\u003c/p\u003e \u003cp\u003eIn the modeling cohort, 10.71% of ICU patients were defined as CRE carriers as well the others without CRE colonizing. There was a significant imbalance in the number of case and control group. Traditional early warning model leads to a high error in the predictive accuracy of minority categories because of low screening specificity when the dataset is highly imbalanced[\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]. The statistical analysis method based on the balanced sample size between classes can not achieve satisfactory results. Therefore, this study constructed a new dataset by introducing an oversampling technology called ADASYN algorithm to solve the imbalance of majority and minority categories. ADASYN algorithm, different from other oversampling methods to simply replicate samples, randomly selected one sample for each minority sample in its nearest neighbor sample, and then randomly selects one point on the two sample lines as the newly synthesized sample[\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]. ADASYN algorithm paid more attention to interclass distance and local density, especially difficult samples or noise sensitive regions in minority categories, which modulate the sampling strategy more finely and avoided the problem of data size expansion and increased model training complexity caused by oversampling. ADASYN algorithm helped to improve the accuracy and generalization ability of models in real-world applications. The results of this study showed ADASYN algorithm strengthened the consistence between observed probability and predicted probability without reducing the discrimination of predictive mode.\u003c/p\u003e \u003cp\u003eThe consequence of DCA and CIC demonstrated ICU patients would conspicuously benefit from the clinical decisions based on the predictive model within a wide and practical range of threshold probabilities. Clinician could incorporate threshold probabilities as a surrogate of physician preference, which highlighted the superiority of predictive model in supporting clinical decision-making processes[\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e]. What\u0026rsquo;s more, the parameters of predictive model in the validation cohort were acceptable,which indicated the model was steady, reproducible and generalduty. The results of external validation is dependable because of the normative framework of external validation study design referring to a teaching study[\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eThe strength of this study was mainly that it was the first time to constructed a novel predictive model for identifying patients with high risk of CRE colonization by a prospective cohort study, and meanwhile the model was verified by DCA and CIC compared to traditional regression analysis and clinical scoring system. The prospective study is consistent with temporal logic in causal inference. Moreover, the population who underwent CRE screening in ICU was entirely included in this study and the sample size reached statistical requirement. Nevertheless, we must acknowledge some other limitations of the study. All results were derived from a single center. Besides, the relevant risk factors incorporated in this study may be incomplete, which may lead to the abandonment of some key variables. The predictive model need to be further validated by cohort studies with multicentric data, more comprehensive risk factors and rigorous quality control. Even so, we believed that the proposed model may contribute to promote our understanding of the prevention and control of patients potentially suffering from CRE colonization in ICU.\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eIn conclusion, this prospective cohort study shows the ADASYN algorithm model does outperform conventional original data model. The ADASYN algorithm model may prove to be clinically useful in precise identification and intervention for patients with high risk of hospital-acquired CRE colonization. Furthermore, by optimizing the usage of carbapenem antibiotics and mechanical ventilation, the risk of CRE colonization may decrease.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e \u003ch2\u003e \u003cb\u003eEthics approval and consent to participat\u003c/b\u003ee\u003c/h2\u003e \u003cp\u003e This study has been approved by the ethical committee of Changzhou Cancer Hospital review board [Approval ID:2023 (SR) No.030].\u003c/p\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003eConsent for publication\u003c/strong\u003e \u003cp\u003eNot applicable.\u003c/p\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003eCompeting interests\u003c/strong\u003e \u003cp\u003eThe authors declare that they have no competing interests.\u003c/p\u003e \u003c/p\u003e\u003ch2\u003eFundings\u003c/h2\u003e \u003cp\u003eThis study was supported by the Young Talent Development Program of Changzhou Health Commission (CZQM2023021), Health Young Scientific and Technological Talent Development Program of Changzhou Medical Development Center and Changzhou Medical Association (lcyx2023010), the Applied Basic research Program of Changzhou Technology Division (CJ20241024).\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eConception and design: QM, SC and JY. Provision of study materials or patients: XG, ZQ, XY. Collection and assembly of data: JY, SC. Data analysis and interpretation: JY, SC and QM. All authors read and approved the final manuscript.\u003c/p\u003e\u003ch2\u003eAcknowledgement\u003c/h2\u003e\u003cp\u003eThe authors are grateful to the clinical medical institution members, nurses, and physicians to perform this study.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eData is provided within the manuscript or supplementary information files\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eQin X, Ding L, Hao M, \u003cem\u003eet al\u003c/em\u003e. Antimicrobial resistance of clinical bacterial isolates in China: current status and trends[J]. JAC Antimicrob Resist, 2024, 6(2):dlae052.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHuang N, Jia H, Zhou B, \u003cem\u003eet al\u003c/em\u003e. Hypervirulent carbapenem-resistant Klebsiella pneumoniae causing highly fatal meningitis in southeastern China[J]. Front Public Health, 2022, 10:991306.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWu D, Xiao J, Ding J, \u003cem\u003eet al\u003c/em\u003e. Predictors of Mortality and Drug Resistance Among Carbapenem-Resistant Enterobacteriaceae-Infected Pancreatic Necrosis Patients[J]. Infect Dis Ther, 2021, 10(3):1665\u0026ndash;1676.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhu Y, Xiao T, Wang Y, \u003cem\u003eet al\u003c/em\u003e. Socioeconomic Burden of Bloodstream Infections Caused by Carbapenem-Resistant Enterobacteriaceae[J]. Infect Drug Resist, 2021, 14:5385\u0026ndash;5393.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSong KH, Kim CJ, Choi NK, \u003cem\u003eet al\u003c/em\u003e. Clinical and economic burden of bacteremia due to multidrug-resistant organisms in Korea: a prospective case control study[J]. J Glob Antimicrob Resist, 2022, 31:379\u0026ndash;385.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTomczyk S, Zanichelli V, Grayson ML, \u003cem\u003eet al\u003c/em\u003e. Control of Carbapenem-resistant Enterobacteriaceae, Acinetobacter baumannii, and Pseudomonas aeruginosa in Healthcare Facilities: A Systematic Review and Reanalysis of Quasi-experimental Studies[J]. Clin Infect Dis, 2019, 68(5):873\u0026ndash;884.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWu MM, Yang DH, Pan HY, \u003cem\u003eet al\u003c/em\u003e. Risk factors analysis of bloodstream infection caused by colonized carbapenem-resistant Enterobacteriaceae[J]. Zhejiang Med, 2022,44(12):1300\u0026ndash;1303.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGomides MDA, Fontes AMS, Silveira AOSM, \u003cem\u003eet al\u003c/em\u003e. The importance of active surveillance of carbapenem-resistant Enterobacterales (CRE) in colonization rates in critically ill patients[J]. PLoS One, 2022, 17(1):e0262554.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStein M, Dabaja-Younis H, Kassis I, \u003cem\u003eet al\u003c/em\u003e. Evaluation of Pediatric Screening for Resistant Pathogens in an Israeli Tertiary Center[J]. Isr Med Assoc J, 2022, 25(12):836\u0026ndash;841.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHo KW, Ng WT, Ip M, \u003cem\u003eet al\u003c/em\u003e. Active surveillance of carbapenem-resistant Enterobacteriaceae in intensive care units: Is it cost-effective in a nonendemic region[J]. Am J Infect Control, 2016, 44(4):394\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYang JS, Min QF, Gong XW, \u003cem\u003eet al\u003c/em\u003e. Pathogenocity and risk factors for intestinal colonization of Carbapenem-resistant Enterobacteriaceae in patients from Intensive Care Unit[J]. Clin J Infect Control, 2024, 23(11):1373\u0026ndash;1378.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZheng Y, Zhang C, Liu Y. Risk prediction models of depression in older adults with chronic diseases[J]. J Affect Disord, 2024, 359:182\u0026ndash;188.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYu M, Yuan Z, Li R, \u003cem\u003eet al\u003c/em\u003e. Interpretable machine learning model to predict surgical difficulty in laparoscopic resection for rectal cancer[J]. Front Oncol, 2024,14:1337219.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHao Y, Li X, Zhang C. Improving prediction model robustness with virtual sample construction for near-infrared spectra analysis[J]. Anal Chim Acta, 2023, 1279:341763.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKhan TM, Xu S, Khan ZG, \u003cem\u003eet al\u003c/em\u003e. Implementing Multilabeling, ADASYN, and ReliefF Techniques for Classification of Breast Cancer Diagnostic through Machine Learning: Efficient Computer-Aided Diagnostic System[J]. J Healthc Eng, 2021, 2021:5577636.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTasin I, Nabil TU, Islam S, \u003cem\u003eet al\u003c/em\u003e. Diabetes prediction using machine learning and explainable AI techniques[J]. Healthc Technol Lett, 2022, 10(1\u0026ndash;2):1\u0026ndash;10.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu SD, Wu AH, Li CH, \u003cem\u003eet al\u003c/em\u003e. Screening of intestinal carbapenem-resistant Enterobacteriaceae for hospitalized patients of high-risk department and risk factors for positive testing[J]. Chin J Nosocomiol, 2023(04):517\u0026ndash;521.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eClinical and Laboratory Standards Institute. Performance standards for antimicrobial susceptibility testing[S]. M100-S30.Wayne, PA:CLSI, 2023.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen HY, Jean SS, Lee YL, \u003cem\u003eet al\u003c/em\u003e. Carbapenem-Resistant Enterobacterales in Long-Term Care Facilities: A Global and Narrative Review[J]. Front Cell Infect Microbiol, 2021, 11:601968.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWangchinda W, Thamlikitkul V, Watcharasuwanseree S, \u003cem\u003eet al\u003c/em\u003e. Active Surveillance for Carbapenem-Resistant Enterobacterales (CRE) Colonization and Clinical Course of CRE Colonization among Hospitalized Patients at a University Hospital in Thailand[J]. Antibiotics (Basel), 2022, 11(10):1401.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePan F, Chen P, Duan Y, \u003cem\u003eet al\u003c/em\u003e. Prevalence of intestinal colonization and nosocomial infection with carbapenem-resistant Enterobacteriales in children: a retrospective study[J]. Front Public Health, 2023, 11:1251609.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHassoun-Kheir N, Hussien K, Karram M, \u003cem\u003eet al\u003c/em\u003e. Clinical significance and burden of carbapenem-resistant Enterobacterales colonization acquisition in hospitalized patients[J]. Antimicrob Resist Infect Control, 2023, 12(1):129.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCai M, Zhang H. Research progress for CRE infection in intensive care unit patients[J]. Laboratory Medicine, 2024, 39(09):913\u0026ndash;918.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBai HS, Wu XQ, Guo WD, \u003cem\u003eet al\u003c/em\u003e. Progress of research on active screening and multidisciplinary collaborative management mode of carbapenem-resistant Enterobacteriaceae[J]. Clin J Nosocomiol, 2023, 33(22):3499\u0026ndash;3504.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZaidah AR, Mohammad NI, Suraiya S, \u003cem\u003eet al\u003c/em\u003e. High burden of Carbapenem-resistant Enterobacteriaceae (CRE) fecal carriage at a teaching hospital: cost-effectiveness of screening in low-resource setting[J]. Antimicrob Resist Infect Control, 2017, 6:42.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhu FM. DRG/DIP Payment Reform, Cost Control and Impact Mechanism: A Multi-Period DID Analysis Based on Sampling Datafrom 19 Regions[J]. Chinese Social Security Review, 2023, 7(05):54\u0026ndash;67.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGarpvall K, Duong V, Linnros S, \u003cem\u003eet al\u003c/em\u003e. Admission screening and cohort care decrease carbapenem resistant enterobacteriaceae in Vietnamese pediatric ICU's[J]. Antimicrob Resist Infect Control, 2021, 10(1):128\u0026ndash;131.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKotb S, Lyman M, Ismail G, \u003cem\u003eet al\u003c/em\u003e. Epidemiology of Carbapenem-resistant Enterobacteriaceae in Egyptian intensive care units using National Healthcare-associated Infections Surveillance Data, 2011\u0026ndash;2017[J]. Antimicrob Resist Infect Control, 2020, 9(1):2\u0026ndash;6.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZimmermann P, Curtis N. The effect of antibiotics on the composition of the intestinal microbiota - a systematic review[J]. J Infect, 2019, 79(6):471\u0026ndash;489.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGe W, Wei W, Shuang P, \u003cem\u003eet al\u003c/em\u003e. Nasointestinal Tube in Mechanical Ventilation Patients is More Advantageous[J]. Open Med (Wars), 2019, 14:426\u0026ndash;430.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi Y, Xu W, Li W, \u003cem\u003eet al\u003c/em\u003e. Research on hybrid intrusion detection method based on the ADASYN and ID3 algorithms[J]. Math Biosci Eng, 2022, 19(2):2030\u0026ndash;2042.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNi Z, Zhu Y, Qian Y, \u003cem\u003eet al\u003c/em\u003e. Synthetic minority over-sampling technique-enhanced machine learning models for predicting recurrence of postoperative chronic subdural hematoma[J]. Front Neurol, 2024, 15:1305543.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang W, Liu J, Jin W, \u003cem\u003eet al\u003c/em\u003e. Radiomics from dual-energy CT-derived iodine maps predict lymph node metastasis in head and neck squamous cell carcinoma[J]. Radiol Med, 2024, 129(2):252\u0026ndash;267.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVickers AJ, Holland F. Decision curve analysis to evaluate the clinical benefit of prediction models[J]. Spine J, 2021, 21(10):1643\u0026ndash;1648.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhao L, Leng Y, Hu Y, \u003cem\u003eet al\u003c/em\u003e. Understanding decision curve analysis in clinical prediction model research[J]. Postgrad Med J, 2024, 100(1185):512\u0026ndash;515.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRamspek CL, Jager KJ, Dekker FW, \u003cem\u003eet al\u003c/em\u003e. External validation of prognostic models: what, why, how, when and where[J]. Clin Kidney J, 2020, 14(1):49\u0026ndash;58.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Prognostic nomogram, Hospital-acquired CRE colonization, ICU patients, ADASYN algorithm","lastPublishedDoi":"10.21203/rs.3.rs-6746525/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6746525/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eObjectives\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo construct a prognostic model for patients with high risk of Carbapenem-resistant Enterobacteriaceae (CRE) colonization in Intensive Care Unit (ICU).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA prospective cohort study was performed on 724 ICU patients from Mar 2023 to Jul 2024 at a tertiary hospital. Totally 577 patients were involved by selection and divided into the modeling (n=392) and validation (n=185) cohorts. Multivariate analysis based on adaptive synthetic sampling (ADASYN) was conducted to estimate the risk of CRE colonization, then presented with a visible nomogram. Next, the performances of nomogram were assessed by area under the receiver operating characteristic curve (AUROC), calibration curves, decision curve analysis (DCA), clinical impact curve (CIC).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe incidence rate of CRE colonization was 10.05% among ICU patients. Multivariate analysis showed Enterobacteriaceae infection (D), days of carbapenem antibiotics (X\u003csub\u003e1\u003c/sub\u003e) and mechanical ventilation (X\u003csub\u003e2\u003c/sub\u003e) were independent risk factors of CRE colonization. Logistic regressions based on original data and ADASYN algorithm were described as follows: Logit(\u003cem\u003eP\u003c/em\u003e\u003csub\u003e1\u003c/sub\u003e)=1.824D\u003csub\u003e1\u003c/sub\u003e+21.604D\u003csub\u003e2\u003c/sub\u003e+2.482X\u003csub\u003e1\u003c/sub\u003e+1.088X\u003csub\u003e2\u003c/sub\u003e-8.039 and Logit(\u003cem\u003eP\u003c/em\u003e\u003csub\u003e2\u003c/sub\u003e)=2.754D\u003csub\u003e1\u003c/sub\u003e+24.355D\u003csub\u003e2\u003c/sub\u003e+2.824X\u003csub\u003e1\u003c/sub\u003e+1.577X\u003csub\u003e2\u003c/sub\u003e-7.057. Internal validation showed ADASYN algorithm model has a higher goodness-of-fit compared with original data model without reducing the discrimination (\u003cem\u003eP\u003c/em\u003e=0.818). DCA of ADASYN algorithm model indicated the threshold risk of positive net benefit ranged 0 to 75%. CIC analysis verified ADASYN algorithm model possessed significant predictive value. A well discrimination with AUROC value of 0.891 (95%CI:0.786~0.995) and goodness-of-fit (\u003cem\u003eP\u003c/em\u003e=0.197) was demonstrated by external validation.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eADASYN algorithm model can identify patients with high risk of CRE colonization. Optimizing the usage of carbapenem and mechanical ventilation may reduce the risk of CRE colonization.\u003c/p\u003e","manuscriptTitle":"Construction and validation of prognostic nomogram for hospital-acquired CRE colonization among ICU patients based on ADASYN algorithm","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-06-25 02:03:09","doi":"10.21203/rs.3.rs-6746525/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"4c15ddce-46c1-46f8-9a47-e887c91e49e8","owner":[],"postedDate":"June 25th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-03-30T16:39:44+00:00","versionOfRecord":[],"versionCreatedAt":"2025-06-25 02:03:09","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6746525","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6746525","identity":"rs-6746525","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}