Construction of a Machine Learning-Based Predictive Model for Complications Associated with Peripheral Intravenous Catheter Placement in Elderly Patients: A Prospective Cohort Study

Construction of a Machine Learning-Based Predictive Model for Complications Associated with Peripheral Intravenous Catheter Placement in Elderly Patients: A Prospective Cohort Study | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Construction of a Machine Learning-Based Predictive Model for Complications Associated with Peripheral Intravenous Catheter Placement in Elderly Patients: A Prospective Cohort Study Xiao Chen, Yuchen Guo, Zhimin Liang, Lingli Li This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8197291/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 5 You are reading this latest preprint version Abstract Aim: To establish a prediction model for complications tightly correlated with peripheral intravenous catheter placement in elderly patients using four machine learning methods, identify risk factors, and comprehensively assess the models to select the optimal one. Design: A Prospective Cohort Study Methods: We collected elderly patients admitted to a tertiary hospital in Chengdu, Sichuan Province, from November 2023 to September 2024, includingbaseline information, laboratory tests, risk assessments, IV catheter treatment, and catheterization procedures. Risk factors for complications were identified through stepwise regression, which was predominantly intended to construct risk prediction models using random forest, logistic regression, gradient boosting, and Bayesian logistic regression. Model's performance was systematically assessed through ROC, sensitivity, specificity, brier and accuracy. Results: This study involved a total of 750 patients, with a peripheral intravenous catheter complications rate of 32.31%. The prediction model comprised seven variables, nutritional risk, comorbidities, self-care ability, vascularcondition, curlededges, daily input volume and pain score. Amongthe four machine learning models, the random forest model displayed the optimal predictive performance. Conclusion: The random forest-based risk prediction model for peripheral intravenous catheter complications in elderly patients demonstrated the best predictive performance and can assist healthcare professionals in identifying high-risk factors. Implications for the Profession and/or Patient Care Enhancing the ability of clinical nurses to reduce the incidence of peripheral intravenous cathetercomplications and improve patient outcomes. Impact : The prediction model demonstrates good ability to identify the risk of peripheral intravenous catheter complications in elderly patients is helpful to provide personalized care. Patient or Public Contribution : Participants were hospitalized elderly patients. First of all, before the investigation and research, a team is formed to discuss the concept, research purpose, method, significance, etc., and determine the research tools. Second, by reasonably explaining the study to patients to seek informed consent from the patient and sign it. Health sciences/Diseases Health sciences/Health care Health sciences/Medical research Health sciences/Risk factors machine learning elderly patients complications of peripheral venous cathters severity of complications prediction model Figures Figure 1 Figure 2 1.0 Introduction As an extensively utilized procedure in intravenous fluid administration and drug therapy, “the insertion of peripheral intravenous catheters (PVCs), remains a prevalent clinical practice” (Nickel B et, al.,2019). Nevertheless, catheter-related adverse reactions still demonstrate alarming incidence. Data from a multicenter prospective clinical study suggest that more than half of hospitalized patients experience varying degrees of catheter-related complications (Chen YM et al.,2022; Miliani K et al.,2017). “In line with United Nations population projections, the proportion of the elderly population aged 65 and above in China is anticipated to augment dramatically, increasing from 14% in 2022 to 30% by 2050” (Anastasia G et al.2022). This demographic shift reveals that China will undergo a rapid and unprecedented aging process over the next three decades. The elderly population will increase demand for medicine and medical services. Even so, the physical condition of the elderly may cause early replacement of intravenous catheters, resulting in increased medical consumables expense. Moreover, mobility limitations and extended hospitalizations may occur as a result of chronic pain and anxiety (Larsen E et al.,2017; Krein SL et al.,2019). In recent years, studies on PVC (premature ventricular contractions) complications have mainly focused on related risk factors and optimizing clinical intervention measures, which have exhibited significant limitations, such as the difference in the level of competence of medical staff and a lack of objective assessment tools. Moreover, thanks to its strong capabilities in data mining and processing, machine learning (ML) has advanced rapidly in disease risk prediction. Taking into account the limitations of the previously discussed models, the primary aim of this study was to develop a practical, implementable clinical prediction model for complications associated with peripheral intravenous catheters in elderly patients. By incorporating potentially significant characteristics, this model displays superior accuracy to anticipate the risk of catheter-related complications occurring. It aims to help clinicians identify high-risk patients, implementation of early intervention measures. 2.0 Material & Methods 2.1 The Study Designs This study was designed as a prospective cohort study. 2.2 Participants The study incorporated 36 variables, the estimated sample size was 750 who received PVC treatment between November 1, 2023, and September 30, 2024. All data were collected at West China Hospital, Sichuan University, and were divided into a “training set and a validation set” (Shah, 2017) in a 7:3 ratio ( Figure 1 ). The inclusion criteria were as follows:(1) age≥60 years; (2) underwent PVC insertion and treatment during hospitalization; (3) obtained informed consent from both the patient and the operating nurse. The exclusion criteria were as follows:(1) elderly patients whose PVC was removed ascribable to transfer to another department or hospital; (2) patients with missing medical records 20% or more. 2.3 Data Collection 2.3.1 Sociodemographic Data and Clinical Information Demographic characteristics and medical history data from the baseline phase were obtained from patients’ medical records and histories. After insertion, catheter-related characteristics and nurse data were obtained from the catheterisation nurses. Treatment-related characteristics were collected daily post-insertion, including medications, catheter status, infusion status, and complication status. The duration from “insertion to removal was calculated as the indwelling time (Carr et al., 2018). These data entiled (1) baseline information: gender, age, patient educational level, department, smoking status, alcohol consumption, BMI, patient vascular condition, number of comorbidities; (2) Laboratory tests and risk assessment: albumin, nutritional risk, thrombosis risk, hemoglobin, platelet count, (3) Operator information: nurse education level, years of experience, professional title, intravenous therapy nurse, number of intravenous therapy training sessions attended in the past year; (4) IV catheter and treatment information: IV catheter type, insertion site, duration of placement, infusion method, infusion rate, tension-free Securement, edge rolling, IV fluid leakage, U-shaped fixation, extension tube, total infusion duration, total infusion volume, daily infusion duration, daily infusion volume, infusion frequency, pain score, and drug properties. 2.3.2 Definition of Outcomes Rooted in a literature review, complications associated with “peripheral intravenous catheters in elderly patients primarily include phlebitis, leakage and extravasation, catheter occlusion” (Marsh et al., 2021), and accidental dislodgement. The Infusion Nursing Society (INS) published the “Infusion Therapy Practice Standards” in 2024(Nickel B et al.,2024), which defines these terms as follows:(1) Phlebitis: Sterile inflammation of a vein; (2) Leakage and extravasation: Non-corrosive/corrosive medications entering the surrounding tissues outside the venous lumen; (3) Tube occlusion: Inability to withdraw blood or slow blood return, inability to inject flushing solution or high resistance, or inability to administer fluids; (4) Accidental tube dislodgement: The PVC catheter is unintentionally removed from the blood vessel by the patient or medical staff. 2.4 Statistical analysis All raw data were organised in Excel and imported into SPSS 29.0 for statistical analysis. The mean ± standard deviation was used for normally distributed continuous variables. Independent samples t-tests were employed for intergroup comparisons; non-normally distributed continuous variables were described using the median (interquartile range) and analysed using the Mann-Whitney U test (Hirasawa et al., 2025). Categorical variables were expressed as frequencies (percentages), and intergroup differences were assessed using the x² test, with p < 0.05 indicating statistical significance. Stepwise regression was used to screen for characteristic variables. Rooted in the R4.4.2 language platform, “we constructed random forest (RF), logistic regression (LR), extreme gradient boosting (XGBoost), Bayesian Logistic Regression (BLR), machine learning models were constructed using the random Forest, glm, xgboost, and brms packages, respectively, for multivariate modelling analysis” (Evi Diana Omar et al., 2024). The models were assessed using the area under the receiver operating characteristic curve (AUC), sensitivity, specificity, Brier score, and accuracy. 2.5 Ethical Considerations . This study was approved by the Ethics Committee of West China Hospital of Sichuan University (No. 2023 1940). All methods were completed in accordance with the Declaration of Helsinki and/or the appropriate guidelines and regulations. All patient-approved dimensions were obtained before the start of the study. 3.0 Results 3.1 General Information and Characteristics Patients were distributed across 11 departments (7 in surgery and 4 in internal medicine), with a predominance of males (54.67%) ( Figure 1 ). The average age was 69.96±7.14years, with a PVC complication rate of 32.31%. The patients were randomly divided into a training group (526 cases, complication rate 31.75%) and a test group (224 cases, complication rate 33.48%) in a 7:3 ratio. No remarkable group distinctions were observed about gender, age, educational level, department, smoking, alcohol consumption, BMI, vascular condition, complications, nutritional risk, thrombotic risk, self-care ability, hemoglobin, albumin, and platelet levels ( p >0.05). 97.73% nurses had a bachelor’s degree or higher, while 83.07% were at the junior level or below, with an average work experience of 8.65±5.66 years. No noticeable group distinctions among investigated nurses were detected with reference to education level, professional title, or work experience. The work experience was mostly 25～96 hours (89.06%), with the primary placement sites being the back of the hand (21.86%) and the forearm (48.80%). The primary administration method was intravenous infusion (74.00%), with an average infusion duration of 17.73±17.46 hours and an average total infusion volume of 2908.86±2606.36ml. There were no significant differences between the two groups of patients in terms of PVC model, insertion site, indwelling time, administration method, infusion rate, dressing application method, edge rolling, leakage, U-shaped fixation, extension tube, administration frequency, pain score, total administration duration, total administration volume, and drug properties ( p >0.05) ( Table 1 ) 3.2 Features included in the model Using the occurrence of PVC complications as the dependent variable, 37 candidate variables were included in a stepwise regression analysis, ultimately retaining 7 variables: nutritional risk, comorbidities, self-care ability, vascular condition, dressing edge rolling, daily fluid volume, and pain score ( Table 2 ). 3.3 Model C onstruction and E valuation Using RF, LR, XGBoost, and BLR, a predictive model for peripheral intravenous catheter complications was constructed for elderly patients, comprising seven risk factors as feature variables. The performance of the four models on the test set ( Table 3 , Figures 2 ). As experimental findings illustrated, RF demonstrated optimal diagnostic performance (AUC = 0.864, sensitivity = 0.913, specificity = 0.707, PPV = 0.861, NPV = 0.803, AUPRC = 0.777, Brier = 0.133, accuracy = 0.835). 4.0 Discussion 4.1 Incidence of PVCs C omplications in E lderly P atients. As people age, the blood vessels gradually lose their elasticity, becoming hardened and thickened, thereby increasing the risk of complications such as phlebitis and drug extravasation, which ultimately affect the efficacy of venous therapy (Fu, J et al., 2022). Ya-mei Chen et al(Chen YM et al.,2022) conducted a cross-sectional survey of 5,345 patients in nine tertiary hospitals in Suzhou, China. They found that the incidence of complications was 54.05%, higher than our research findings. This difference may be due to the study not restricting patient age. Emmanuelle Gras et al(Gras E et al.,2023) investigated 849 PVCs in 322 patients aged 70 years or older in the geriatric department of a French hospital. As the above analyses demonstrated, the incidence of complications was 13.80%, which may be associated with a fundamental fact that the study population was limited to patients aged 70 years or older, some of whom were home-dwelling, and their frequency of venous therapy was lower than that of hospitalized patients. Our study confirms the high incidence of PVC complications in elderly patients, highlighting the necessity of implementing preventive measures and management strategies tailored to this population. 4.2 Risk F actors for c omplications of PVC s in E lderly P atients As our research outcomes illustrated, nutritional risk, comorbidities, self-care ability, vascular condition, dressing edge rolling, daily fluid volume, and pain scores were significantly associated with the incidence of peripheral intravenous catheter complications in elderly patients. Favourable nutritional status can exert a significant influence on elderly patients’ tolerance to venous therapy and treatment outcomes. On the other hand, weak vascular tissue can suffer from insufficient nutrients. Under such circumstances, vessels are more likely to sustain mechanical and chemical damage, further increasing the risk of complications (Jia Z et al., 2020). Clinical practitioners should integrate nutritional risk screening for older adults into routine nursing assessments. As shown by several retrospective studies already, the presence of diabetes and vascular complications and kidney disease, along with age-related skin and connective tissue changes, dramatically increases the risk of PVC complications in elderly patients (Bonte F et al.,2019; Enes S M S et al.,2016). These findings were reported by Simin D et al. (2019), which may be due to prolonged intravenous therapy that impairs the quality of the vascular wall, local inflammation, or micro-thrombosis formation (Marsh N et al., 2023). The presence of other diseases confirms the role of comorbidity in AD. Thus, patients with comorbid chronic diseases should undergo enhanced multidisciplinary collaborative management, with a focus on controlling blood glucose levels, optimising circulatory function, and improving microcirculatory status to minimise damage to vascular endothelial cells caused by metabolic abnormalities and hemodynamic disorders. Specific individuals with impaired self-care abilities commonly experience reduced activity, leading to slowed venous return and prolonged retention of medications in local veins. Studies have shown that early mobilisation of the catheterised limb, maintaining normal daily activities, and adequate hydration can significantly delay the development of catheter-related complications (Chen Y et al., 2020). Consequently, it is advisable to perform puncture on limbs with desirable functional status. Moreover, patients are encouraged to engage in moderate functional exercises. Vascular aging functions as a key feature of aging, and repeated venous punctures increase patient discomfort and the risk of complications (Bahl A et al.,2024). Da Silva's research team (Silva D A S et al.,2023) (n=3,552) conducted a large-scale clinical study, reaching a confirmed conclusion that specialized teams can materialize lower repeated punctures (P < 0.01) and decrease catheter consumption by 30%. The introduction of vascular visualization technology can greatly improve the puncture success rate (P < 0.05) (Kuo C et al.,2025), especially in patients with hard punctures, while measuring the diameter of vessels (error < 1 mm). However, local hospitals that are not at the main level typically lack sufficient equipment and personnel. Specifically, only a handful of university hospitals utilize vascular visualization technology for central venous catheterization in their intensive care units/emergency departments. Meanwhile, it is mostly not at all utilised for PVC insertion in vulnerable populations. Consequently, it is vital to train in vascular visualization technology. According to the INS guidelines, an ideal fixation device should keep the catheter stable while allowing for visibility of the site. At the moment, all use 3M transparent dressings. These dressings are highly transparent, allowing the user to see changes at the puncture site and around it. However, the sweating of the patient will eventually cause them to curl and peel off which reduces its fixation efficiency and causes catheter dislodgement. When this situation is in place, various complications may arise. (Huang et al.,2021) Therefore, for future academic pursuits, it would be better to focus on the curling and peeling off of transparent dressings in hot, sweaty conditions. According to a logistic regression analysis, patients who had pain were more prone to develop PVC complications (OR=7.13) (Welyczko N et al et al., 2020). Studies related to this phenomenon have evidently confirmed that active inquiry regarding abnormal sensations can end complications and pain in 24 hours (Lee S et al., 2019). Removing the tube can reduce the risk from 19.7% to 5.3%. Therefore, it is recommended to implement a dynamic monitoring system centered on pain: elderly patients should be assessed every 1 to 2 hours through the use of the “four-step observation method”, which can not only centre on distinguishing whether pain is of organic or psychogenic origin, but also dynamically record intensity, and probe into localization characteristics. The routine nursing protocol for assessing pain can help detect complications early and improve the safety of intravenous therapy significantly. According to Trivedi et al. in 2022, the risk of complications with peripheral intravenous catheters increases with large volume infusions. The processes concerned may include: (i) an upsurge in the activity of shear stress system owing to hemodynamic alterations; (ii) osmotic departure of intravascular liquid component owing to large volume liquid administration, damages of endothelial cells. Hence, it is encouraged to prioritize alternative options. PICC lines are universally employed in clinical practice owing to their advantages of high safety, high success rate, and long retention time. Nonetheless, they also carry a risk of impaired blood flow and a high incidence of thrombosis. Thus, appropriate venous access options should be selected in clinical practice based on the patient’s condition. 4.3 Strengths and Limitations This study employed four machine learning algorithms-RF, LR, XGBoost, and BLR-to model the risk prediction of peripheral intravenous catheter complications in elderly patients (Liu et al., 2021). The evaluation metrics selected were AUC, AUPRC, sensitivity, specificity, PPV, NPV, accuracy, and Brier score, and the performance of the four algorithms was compared. To the best of our knowledge, we pioneered the development and validation of a predictive model for peripheral intravenous catheter-related complications in elderly patients using machine-learning algorithms. Random forest achieved comparatively desirable predictive accuracy and outperformed other models in terms of predictive efficiency. Nonetheless, our study has several limitations. First, since the data were collected from a single study, further research is needed to validate the generalizability of our model. Second, while the model can assist clinicians in identifying high-risk patients for complications promptly, it does not provide specific treatment recommendations. Multicenter data and treatment information are needed to provide more conclusive results and better guidance. 5.0 Conclusion In this study, we proposed and validated a machine learning-based predictive model for predicting the development of peripheral venous catheter-related complications in elderly patients. As our research findings evidently demonstrated, the random forest model can accurately predict the development of peripheral venous catheter-related complications. Identifying high-risk populations can provide timely opportunities for close monitoring and intervention. This model is excepted to promote early diagnosis and influence the treatment decision-making process. Declarations Author Contributions Xiao Chen,Yuchen Guo,Zhimin Liang did the conception and design, or acquisition of data, or analysis and interpretation of data. Xiao Chen,Lingli Li involved in drafting the manuscript or revising it critically for important intellectual content. Xiao Chen,Yuchen Guo,Zhimin Liang,Lingli Li given final approval of the version to be published. Xiao Chen, Yuchen Guo Agreed did the Project Administration. Acknowledgements The authors thank and thank all clinicians and staff who collected the data. Conflicts of Interest There’s no Conflict of Interest Among the Author’s. Data Availability Statement The data are not publicly available due to privacy or ethical restrictions. But available Upon Request to the Corresponding Author. Funding The research received no funding or any form of grant References Nickel B. Peripheral intravenous access: applying infusion therapy standards of practice to improve patient safety. Crit Care Nurse.2019;39(1):61-71. doi:10.4037/ccn2019790. Chen YM, Fan XW, Liu MH, et al. Risk factors for peripheral venous catheter failure: a prospective cohort study of 5345 patients. J Vasc Access. 2022;23(6):911-921.doi:10.1177/11297298211015035. Miliani K, Taravella R, Thillard D, et al. Peripheral venous catheter-related adverse events: evaluation from a multicentre epidemiological study in France (the CATHEVAL Project). PLoS One.2017;12(1):e0168637. doi:10.1371/journal.pone0168637 Anastasia G. World Population Prospects 2022: Summary of Results. 2022 [cited 2023 Jun 18]. In:United Nations Department of Economic and Social Affairs, Population Division [Internet]. Available from: https://www.un.org/development/desa/pd/content/World-Population-Prospects-2022 Larsen E, Keogh S, Marsh N, Rickard C. Experiences of peripheral IV insertion in hospital: a qualitative study. Br J Nurs. 2017;26(19): S18-S25. doi:10.12968/bjon.2017.26.19.S18. Slosse C, Manneville F, Ricci L, et al. Development and evaluation of an algorithm for peripheral venous catheter placement (ALCOV): protocol for a quasi-experimental study[J]. BMJ OPEN, 2024,14(6):10. Krein SL, Saint S, Trautner BW, et al. Patient-reported complications related to peripherally inserted central catheters: a multicentre prospective cohort study. BMJ Qual Saf. 2019;28(7):574-581. doi:10.1136/bmjqs-2018-008726 Resnick O, Ahmad W A, Bancovsky D, et al. Predicting factors for complications in peripheral intravenous catheters in the pediatric population[J]. ACTA PAEDIATRICA, 2021,110(5):1639-1644 Gong Z, Zhang J, Hou J, et al. Drug Extravasation in a Large General Hospital in Hunan, China: A Retrospective Survey[J]. RISK MANAGEMENT AND HEALTHCARE POLICY, 2021,14:4931-4938 Lones M A. Avoiding common machine learning pitfalls[J]. PATTERNS, 2024,5(10):16. Wang Y, Lei L, Ji M, et al. Predicting postoperative delirium after microvascular decompression surgery with machine learning[J]. JOURNAL OF CLINICAL ANESTHESIA, 2020,66:6. Su F, Chao J, Liu P, et al. Prognostic models for breast cancer: based on logistics regression and Hybrid Bayesian Network[J]. BMC MEDICAL INFORMATICS AND DECISION MAKING, 2023,23(1):15. Nickel B, Gorski L, Kleidon T, et al. Infusion Therapy Standards of Practice, 9th Edition[J]. JOURNAL OF INFUSION NURSING, 2024,47:291. Fu, J.; Zhao, L.; Tian, Y.; Liu, Y.; Zhang, H.; Li, H. Impact of Terminal Tip Location of Midline Catheters on Catheter‐related Complications and Indwelling Duration in Elderly Patients. Nursing Open, 2022. Chen Y, Fan X, Liu M, et al. Risk factors for peripheral venous catheter failure: A prospective cohort study of 5345 patients[J]. JOURNAL OF VASCULAR ACCESS, 2022,23(6):911-921. Gras E, Jean A, Rocher V, et al. Incidence of and risk factors for local complications of peripheral venous catheters in patients older than 70 years: Empirical research quantitative[J]. JOURNAL OF CLINICAL NURSING, 2023,32(15-16):5000-5009. Jia Z, Moheb M E, Nordestgaard A, et al. The Geriatric Nutritional Risk Index is a powerful predictor of adverse outcome in the elderly emergency surgery patient[J]. JOURNAL OF TRAUMA AND ACUTE CARE SURGERY, 2020,89(2):397-404. Bonte F, Girard D, Archambault J, et al. Skin Changes During Ageing[J]. Subcell Biochem, 2019,91:249-280. Enes S M S, Opitz S P, de Faro A R M D, et al. Phlebitis associated with peripheral intravenous catheters in adults admitted to hospital in the Western Brazilian Amazon[J]. REVISTA DA ESCOLA DE ENFERMAGEM DA USP, 2016,50(2):261-269. Simin D, Milutinovic D, Turkulov V, et al. Incidence, severity and risk factors of peripheral intravenous cannula-induced complications: An observational prospective study[J]. JOURNAL OF CLINICAL NURSING, 2019,28(9-10):1585-1599. Marsh N, Larsen E N, O'Brien C, et al. Comparing the use of midline catheters versus peripherally inserted central catheters for patients requiring peripherally compatible therapies: A pilot randomised controlled trial (the compact trial) [J]. INFECTION DISEASE & HEALTH, 2023,28(4):259-264. Chen Y, Chen H, Yang J, et al. Patterns and risk factors of peripherally inserted central venous catheter-related symptomatic thrombosis events in patients with malignant tumors receiving chemotherapy.J Vasc Surg Venous Lymphat Disord. 2020;8(6):919-929. doi:10.1016/j.jvsv.2020.01.010. Bahl A, Alsbrooks K, Zazyczny K A, et al. An Improved Definition and SAFE Rule for Predicting Difficult Intravascular Access (DIVA) in Hospitalized Adults[J]. JOURNAL OF INFUSION NURSING, 2024,47(2):96-107. Silva D A S, de Lima T R, Goncalves L. "Academia da Saude" program: mapping evidence from the largest health promotion community program in Brazil[J]. FRONTIERS IN PUBLIC HEALTH, 2023,11:9. Kuo C, Lee W, Ke Y. Ultrasound-guided peripheral intravenous access in adults: A randomized crossovercontrolled trial[J]. INTERNATIONAL EMERGENCY NURSING, 2025,79:8. Huang, L.-S.; Huang, Y.; Hu, J. Current Practices of Peripheral Intravenous Catheter Fixation in Pediatric Patients and Factors Influencing Pediatric Nurses’ Knowledge, Attitude and Practice Concerning Peripheral Intravenous Catheter Fixation: A Cross-Sectional Study.BMC Nursing, 2021. Welyczko N. Peripheral intravenous cannulation: reducing pain and local complications[J]. Br J Nurs, 2020,29(8):S12-S19. Lee S, Kim K, Kim J. A Model of Phlebitis Associated with Peripheral Intravenous Catheters in Orthopedic Inpatients[J]. INTERNATIONAL JOURNAL OF ENVIRONMENTAL RESEARCH AND PUBLIC HEALTH, 2019,16(18):11. Trivedi, M. R.; Bochiya, G. P.; Modi, J. V. A Prospective Study of Peripheral Inserted Venous Catheter Related Local Complications. International Surgery Journal, 2022. Carr, P. J., Rippey, J. C. R., Cooke, M. L., Higgins, N. S., Trevenen, M., Foale, A., & Rickard, C. M. (2018). From insertion to removal: A multicenter survival analysis of an admitted cohort with peripheral intravenous catheters inserted in the emergency department. Infection Control & Hospital Epidemiology , 39 (10), 1216–1221. https://doi.org/10.1017/ice.2018.190 Evi Diana Omar, Mat, H., Karim, A., Ridwan Sanaudi, Ibrahim, F., Mohd Azahadi Omar, Hafiz, Z., Vivek Jayaraj, & Bak Leong Goh. (2024). Comparative Analysis of Logistic Regression, Gradient Boosted Trees, SVM, and Random Forest Algorithms for Prediction of Acute Kidney Injury Requiring Dialysis After Cardiac Surgery. International Journal of Nephrology and Renovascular Disease , Volume 17 , 197–204. https://doi.org/10.2147/ijnrd.s461028 Hirasawa, R., Kazuhiro Oinuma, Hagiwara, S., Sato, T., Yuya Kawarai, Miura, Y., Nakamura, J., & Seiji Ohtori. (2025). Collared fully hydroxyapatite-coated femoral components reduce early periprosthetic femoral fractures in total hip arthroplasty with the direct anterior approach. The Bone & Joint Journal , 107-B (10), 1011–1019. https://doi.org/10.1302/0301-620x.107b10.bjj-2024-1494.r1 Liu, J., Wu, J., Liu, S., Li, M., Hu, K., & Li, K. (2021). Predicting mortality of patients with acute kidney injury in the ICU using XGBoost model. PLOS ONE , 16 (2), e0246306. https://doi.org/10.1371/journal.pone.0246306 Marsh, N., Larsen, E. N., Takashima, M., Kleidon, T., Keogh, S., Ullman, A. J., Mihala, G., Chopra, V., & Rickard, C. M. (2021). Peripheral intravenous catheter failure: A secondary analysis of risks from 11,830 catheters. International Journal of Nursing Studies , 124 (1), 104095. https://doi.org/10.1016/j.ijnurstu.2021.104095 Shah, T. (2017, December 6). About Train, Validation and Test Sets in Machine Learning . Medium; TDS Archive. https://medium.com/data-science/train-validation-and-test-sets-72cb40cba9e7 Tables Tables 1 to 3 are available in the Supplementary Files section. Additional Declarations No competing interests reported. Supplementary Files TABLE1PatientPVCsdata.docx Cite Share Download PDF Status: Under Review Version 1 posted Reviewers invited by journal 30 Apr, 2026 Editor assigned by journal 07 Apr, 2026 Editor invited by journal 30 Dec, 2025 Submission checks completed at journal 27 Dec, 2025 First submitted to journal 27 Dec, 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-8197291","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":633487439,"identity":"5a383cdf-0139-422f-a343-d6c0670cd15f","order_by":0,"name":"Xiao Chen","email":"","orcid":"","institution":"Sichuan University, Sichuan University","correspondingAuthor":false,"prefix":"","firstName":"Xiao","middleName":"","lastName":"Chen","suffix":""},{"id":633487440,"identity":"b67e0937-0228-4e80-8ef9-070cc4c203b3","order_by":1,"name":"Yuchen Guo","email":"","orcid":"","institution":"West China Hospital, Sichuan University","correspondingAuthor":false,"prefix":"","firstName":"Yuchen","middleName":"","lastName":"Guo","suffix":""},{"id":633487441,"identity":"d6ba53bd-2815-4277-95df-c4133c628fdd","order_by":2,"name":"Zhimin Liang","email":"","orcid":"","institution":"Sichuan University, Sichuan University","correspondingAuthor":false,"prefix":"","firstName":"Zhimin","middleName":"","lastName":"Liang","suffix":""},{"id":633487442,"identity":"1ee8fb90-ef9e-49bb-a176-4454deb6399c","order_by":3,"name":"Lingli Li","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA1klEQVRIiWNgGAWjYBACAwYGNuYfBv/l+EG8B2ABYrQwFDAbSzYwMDYkEK/lA3PihgPEajGXSD72uMCAjXHz8R7zBwkVNsYM7IePbsCnxXJGWrrxDAMeZrMzZwwbEs6kmTHwpKXdwOuwM2fMJHgMJNjMbuQYNiS2HbZhkOAxI6Dl/DegFgMe4xlEaznewybNY5AgYSAB0WJGhJY2c8MZBgcMJM4cK5wB9IsxG0G/HGZ+9uDDnwP1/e3NGz58qLAx7Gc/fAyvFkzARpryUTAKRsEoGAXYAAD9sEsJAu+o7wAAAABJRU5ErkJggg==","orcid":"","institution":"West China Hospital, Sichuan University","correspondingAuthor":true,"prefix":"","firstName":"Lingli","middleName":"","lastName":"Li","suffix":""}],"badges":[],"createdAt":"2025-11-24 23:23:12","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8197291/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8197291/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":108978294,"identity":"a92b71df-1ab3-46c7-b2d4-475a9b90cc90","added_by":"auto","created_at":"2026-05-11 11:36:00","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":63233,"visible":true,"origin":"","legend":"\u003cp\u003eScheme of the development and validation of the prediction mod\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-8197291/v1/1f31bd830ec1e2119a5d4509.png"},{"id":108977969,"identity":"f82d294d-33c4-40f2-9db8-9a5c0c1a5af6","added_by":"auto","created_at":"2026-05-11 11:33:33","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":54369,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ea\u003c/strong\u003e: ROC curves of four machine learning models;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2b\u003c/strong\u003e: Sensitivity, specificity, PPV, NPV,and accuracy of four machine learning models;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2c\u003c/strong\u003e: Confusion matrix of four machine learning models on the test set.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-8197291/v1/2b4012a5d4703c522d0a3dcb.png"},{"id":109204640,"identity":"0a03bc78-b444-4f14-9e36-b447f4411f5d","added_by":"auto","created_at":"2026-05-13 15:01:40","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":274739,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8197291/v1/4620de0c-92b2-409c-8bdb-8278a6d98a95.pdf"},{"id":108972088,"identity":"549eb03c-0fe6-47ce-b803-83f3df4a5c81","added_by":"auto","created_at":"2026-05-11 10:34:17","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":38346,"visible":true,"origin":"","legend":"","description":"","filename":"TABLE1PatientPVCsdata.docx","url":"https://assets-eu.researchsquare.com/files/rs-8197291/v1/0fa0f3c521b616100f88e015.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"\u003cp\u003eConstruction of a Machine Learning-Based Predictive Model for Complications Associated with Peripheral Intravenous Catheter Placement in Elderly Patients: A Prospective Cohort Study\u003c/p\u003e","fulltext":[{"header":"1.0 Introduction","content":"\u003cp\u003eAs an extensively utilized procedure in intravenous fluid administration and drug therapy, “the insertion of peripheral intravenous catheters (PVCs), remains a prevalent clinical practice” (Nickel B et, al.,2019). Nevertheless,\u0026nbsp;catheter-related adverse reactions\u0026nbsp;still demonstrate alarming incidence. Data\u0026nbsp;from a multicenter prospective clinical study\u0026nbsp;suggest that\u0026nbsp;more than half of hospitalized\u0026nbsp;patients experience varying degrees of catheter-related complications\u0026nbsp;(Chen YM et al.,2022;\u0026nbsp;Miliani K et al.,2017).\u003c/p\u003e\n\u003cp\u003e“In line with United Nations population projections, the proportion of the elderly population aged 65 and above in China is anticipated to augment dramatically, increasing from 14% in 2022 to 30% by 2050” (Anastasia G et al.2022). This\u0026nbsp;demographic shift\u0026nbsp;reveals that\u0026nbsp;China will undergo a rapid and unprecedented aging process over the next three decades. The elderly population will increase demand for medicine and medical services. Even so, the physical condition of the elderly may cause early replacement of intravenous catheters, resulting in increased medical consumables expense. Moreover, mobility limitations and extended hospitalizations may occur as a result of chronic pain and anxiety (Larsen E et al.,2017; Krein SL et al.,2019).\u003c/p\u003e\n\u003cp\u003eIn recent years, studies on PVC (premature ventricular contractions) complications have mainly focused on related risk factors and optimizing clinical intervention measures, which have exhibited significant limitations, such as the difference in the level of competence of medical staff and a lack of objective assessment tools. Moreover, thanks to its strong capabilities in data mining and processing, machine learning (ML) has advanced rapidly in disease risk prediction.\u003c/p\u003e\n\u003cp\u003eTaking into account the limitations of the previously discussed models, the primary aim of this study was to develop a practical, implementable clinical prediction model for complications associated with peripheral intravenous catheters in elderly patients. By incorporating potentially significant characteristics, this model displays superior accuracy to anticipate the risk of catheter-related complications occurring. It aims to help clinicians identify high-risk patients, implementation of early intervention measures.\u003c/p\u003e"},{"header":"2.0 Material \u0026 Methods","content":"\u003cp\u003e\u003cstrong\u003e2.1 The Study Designs\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was designed as a prospective cohort study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.2 Participants\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe study incorporated 36 variables, the estimated sample size was 750 who received PVC treatment between November 1, 2023, and September 30, 2024. All data were collected at West China Hospital, Sichuan University, and were divided into a “training set and a validation set”\u0026nbsp;(Shah, 2017)\u0026nbsp;in a 7:3 ratio (\u003cstrong\u003eFigure 1\u003c/strong\u003e).\u003c/p\u003e\n\u003cp\u003eThe inclusion criteria were as follows:(1) age≥60 years; (2) underwent PVC insertion and treatment during hospitalization; (3) obtained\u0026nbsp;informed consent from both the patient and the operating nurse. The\u0026nbsp;exclusion criteria were as follows:(1) elderly\u0026nbsp;patients\u0026nbsp;whose PVC\u0026nbsp;was\u0026nbsp;removed\u0026nbsp;ascribable to\u0026nbsp;transfer to another department or hospital; (2) patients\u0026nbsp;with\u0026nbsp;missing\u0026nbsp;medical\u0026nbsp;records 20%\u0026nbsp;or\u0026nbsp;more.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.3 Data Collection\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.3.1 Sociodemographic Data and Clinical Information\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDemographic characteristics and medical history data from the baseline phase were obtained from patients’ medical records and histories. After insertion, catheter-related characteristics and nurse data were obtained from the catheterisation nurses. Treatment-related characteristics were collected daily post-insertion, including medications, catheter status, infusion status, and complication status. The duration from “insertion to removal was calculated as the indwelling time\u0026nbsp;(Carr et al., 2018). These\u0026nbsp;data entiled (1) baseline\u0026nbsp;information: gender, age, patient educational\u0026nbsp;level,\u0026nbsp;department, smoking\u0026nbsp;status, alcohol\u0026nbsp;consumption, BMI, patient\u0026nbsp;vascular condition, number\u0026nbsp;of\u0026nbsp;comorbidities; (2) Laboratory\u0026nbsp;tests and risk assessment: albumin, nutritional\u0026nbsp;risk, thrombosis\u0026nbsp;risk, hemoglobin, platelet\u0026nbsp;count, (3) Operator\u0026nbsp;information: nurse education\u0026nbsp;level, years\u0026nbsp;of experience, professional\u0026nbsp;title, intravenous\u0026nbsp;therapy\u0026nbsp;nurse, number\u0026nbsp;of\u0026nbsp;intravenous therapy training sessions attended in the past year; (4) IV\u0026nbsp;catheter and treatment information: IV\u0026nbsp;catheter type, insertion\u0026nbsp;site, duration\u0026nbsp;of placement, infusion\u0026nbsp;method, infusion\u0026nbsp;rate, tension-free Securement,\u0026nbsp;edge\u0026nbsp;rolling,\u0026nbsp;IV fluid leakage, U-shaped fixation, extension\u0026nbsp;tube, total\u0026nbsp;infusion\u0026nbsp;duration, total\u0026nbsp;infusion volume, daily\u0026nbsp;infusion duration, daily\u0026nbsp;infusion volume, infusion\u0026nbsp;frequency, pain\u0026nbsp;score, and\u0026nbsp;drug properties.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.3.2\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003eDefinition of Outcomes\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eRooted in a literature review, complications associated with “peripheral intravenous catheters in elderly patients primarily include phlebitis, leakage and extravasation, catheter occlusion”\u0026nbsp;(Marsh et al., 2021), and\u0026nbsp;accidental\u0026nbsp;dislodgement. The\u0026nbsp;Infusion Nursing Society (INS) published\u0026nbsp;the “Infusion Therapy Practice Standards”\u0026nbsp;in 2024(Nickel B et al.,2024), which\u0026nbsp;defines these terms as follows:(1) Phlebitis: Sterile\u0026nbsp;inflammation of a vein; (2) Leakage\u0026nbsp;and\u0026nbsp;extravasation: Non-corrosive/corrosive medications entering the surrounding tissues outside the venous lumen; (3) Tube\u0026nbsp;occlusion: Inability\u0026nbsp;to withdraw blood or slow blood return, inability\u0026nbsp;to inject flushing solution or high\u0026nbsp;resistance, or\u0026nbsp;inability to administer fluids; (4) Accidental\u0026nbsp;tube dislodgement: The\u0026nbsp;PVC catheter is unintentionally removed from the blood vessel by the patient or medical staff.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.4\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003eStatistical analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll raw data were organised in Excel and imported into SPSS 29.0 for statistical analysis. The mean ± standard deviation was used for normally distributed continuous variables. Independent samples t-tests were employed for intergroup comparisons; non-normally distributed continuous variables were described using the median (interquartile range) and analysed using the Mann-Whitney U test\u0026nbsp;(Hirasawa et al., 2025). Categorical\u0026nbsp;variables\u0026nbsp;were expressed as frequencies (percentages), and\u0026nbsp;intergroup differences were assessed using the x² test, with p \u0026lt; 0.05 indicating statistical significance. Stepwise\u0026nbsp;regression was used to screen for characteristic\u0026nbsp;variables. Rooted in\u0026nbsp;the R4.4.2 language platform, “we\u0026nbsp;constructed random\u0026nbsp;forest (RF), logistic\u0026nbsp;regression (LR), extreme\u0026nbsp;gradient\u0026nbsp;boosting (XGBoost), Bayesian\u0026nbsp;Logistic\u0026nbsp;Regression (BLR), machine\u0026nbsp;learning models were constructed using the random\u0026nbsp;Forest, glm, xgboost, and\u0026nbsp;brms packages, respectively, for\u0026nbsp;multivariate modelling analysis”\u0026nbsp;(Evi Diana Omar et al., 2024). The\u0026nbsp;models were\u0026nbsp;assessed\u0026nbsp;using the area under the receiver operating characteristic\u0026nbsp;curve (AUC), sensitivity, specificity, Brier\u0026nbsp;score, and accuracy.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.5 Ethical Considerations\u003c/strong\u003e.\u003c/p\u003e\n\u003cp\u003eThis study was approved by the Ethics Committee of West China Hospital of Sichuan University (No. 2023 1940). All methods were completed in accordance with the Declaration of Helsinki and/or the appropriate guidelines and regulations. All patient-approved dimensions were obtained before the start of the study.\u003c/p\u003e"},{"header":"3.0 Results","content":"\u003cp\u003e\u003cstrong\u003e3.1 General Information and Characteristics\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePatients were distributed across 11 departments (7 in surgery and 4 in internal medicine), with a predominance of males (54.67%) (\u003cstrong\u003eFigure 1\u003c/strong\u003e). The average age was 69.96±7.14years, with a PVC complication rate of 32.31%. The patients were randomly divided into a training group (526 cases, complication rate 31.75%) and a test group (224 cases, complication rate 33.48%) in a 7:3 ratio. No remarkable group distinctions were observed about gender, age, educational level, department, smoking, alcohol consumption, BMI, vascular condition, complications, nutritional risk, thrombotic risk, self-care ability, hemoglobin, albumin, and platelet levels (\u003cem\u003ep\u003c/em\u003e\u0026gt;0.05).\u0026nbsp;97.73%\u0026nbsp;nurses had a bachelor’s degree or\u0026nbsp;higher,\u0026nbsp;while\u0026nbsp;83.07%\u0026nbsp;were at the junior level or below, with\u0026nbsp;an average work experience of\u0026nbsp;8.65±5.66 years. No noticeable group distinctions among investigated nurses were detected with reference to education level, professional title, or work experience. The work experience\u0026nbsp;was mostly 25～96 hours\u0026nbsp;(89.06%), with\u0026nbsp;the primary placement sites being the back of the hand\u0026nbsp;(21.86%) and\u0026nbsp;the forearm\u0026nbsp;(48.80%). The\u0026nbsp;primary administration method was\u0026nbsp;intravenous infusion\u0026nbsp;(74.00%), with\u0026nbsp;an average infusion duration of 17.73±17.46 hours and an average total infusion volume of\u0026nbsp;2908.86±2606.36ml. There\u0026nbsp;were no\u0026nbsp;significant differences between the two groups of patients in terms of PVC model, insertion\u0026nbsp;site, indwelling\u0026nbsp;time, administration\u0026nbsp;method, infusion\u0026nbsp;rate, dressing\u0026nbsp;application method, edge\u0026nbsp;rolling, leakage, U-shaped fixation, extension\u0026nbsp;tube, administration\u0026nbsp;frequency, pain\u0026nbsp;score, total\u0026nbsp;administration duration, total\u0026nbsp;administration volume, and\u0026nbsp;drug\u0026nbsp;properties\u0026nbsp;(\u003cem\u003ep\u003c/em\u003e\u0026gt;0.05)\u0026nbsp;(\u003cstrong\u003eTable\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003e1\u003c/strong\u003e)\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.2\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003eFeatures included in the model\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eUsing the occurrence of PVC complications as the dependent variable, 37 candidate variables were included in a stepwise regression analysis, ultimately retaining 7 variables: nutritional risk, comorbidities, self-care ability, vascular condition, dressing edge rolling, daily fluid volume, and pain score (\u003cstrong\u003eTable\u003c/strong\u003e\u003cstrong\u003e2\u003c/strong\u003e).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.3\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003eModel\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003eC\u003c/strong\u003e\u003cstrong\u003eonstruction and\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003eE\u003c/strong\u003e\u003cstrong\u003evaluation\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eUsing RF, LR, XGBoost, and BLR, a predictive model for peripheral intravenous catheter complications was constructed for elderly patients, comprising seven risk factors as feature variables. The performance of the four models on the test set (\u003cstrong\u003eTable\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;3\u003c/strong\u003e\u003cstrong\u003e,\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;Figures\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003e2\u003c/strong\u003e). As experimental findings illustrated, RF demonstrated optimal diagnostic performance (AUC = 0.864, sensitivity = 0.913, specificity = 0.707, PPV = 0.861, NPV = 0.803, AUPRC = 0.777, Brier = 0.133, accuracy = 0.835).\u003c/p\u003e"},{"header":"4.0 Discussion","content":"\u003cp\u003e\u003cstrong\u003e4.1\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003eIncidence of\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003ePVCs\u003c/strong\u003e\u003cstrong\u003eC\u003c/strong\u003e\u003cstrong\u003eomplications in\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003eE\u003c/strong\u003e\u003cstrong\u003elderly\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003eP\u003c/strong\u003e\u003cstrong\u003eatients.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAs people age, the blood vessels gradually lose their elasticity, becoming hardened and thickened, thereby increasing the risk of complications such as phlebitis and drug extravasation, which ultimately affect the efficacy of venous therapy (Fu, J et al.,\u0026nbsp;2022). Ya-mei Chen et al(Chen YM et al.,2022)\u0026nbsp;conducted\u0026nbsp;a cross-sectional survey of 5,345 patients in nine tertiary hospitals in Suzhou, China. They found\u0026nbsp;that the incidence of complications was 54.05%, higher\u0026nbsp;than\u0026nbsp;our research findings. This\u0026nbsp;difference may be\u0026nbsp;due to the study not restricting patient age. Emmanuelle\u0026nbsp;Gras et al(Gras E et al.,2023)\u0026nbsp;investigated\u0026nbsp;849\u0026nbsp;PVCs in 322 patients aged 70\u0026nbsp;years\u0026nbsp;or older\u0026nbsp;in the geriatric department of a French hospital. As the above analyses demonstrated,\u0026nbsp;the incidence\u0026nbsp;of complications was 13.80%, which\u0026nbsp;may be\u0026nbsp;associated with a fundamental fact\u0026nbsp;that the study population was limited to\u0026nbsp;patients aged 70 years or older, some\u0026nbsp;of whom were home-dwelling, and\u0026nbsp;their frequency of venous therapy was lower\u0026nbsp;than that of hospitalized patients. Our\u0026nbsp;study confirms the high incidence of PVC complications in elderly patients, highlighting\u0026nbsp;the necessity of implementing preventive measures and management strategies tailored to this population.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e4.2\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003eRisk\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003eF\u003c/strong\u003e\u003cstrong\u003eactors for\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003ec\u003c/strong\u003e\u003cstrong\u003eomplications of\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003ePVC\u003c/strong\u003e\u003cstrong\u003es in\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003eE\u003c/strong\u003e\u003cstrong\u003elderly\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003eP\u003c/strong\u003e\u003cstrong\u003eatients\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAs our research outcomes illustrated, nutritional risk, comorbidities, self-care ability, vascular condition, dressing edge rolling, daily fluid volume, and pain scores were significantly associated with the incidence of peripheral intravenous catheter complications in elderly patients. Favourable nutritional status can exert a significant influence on elderly patients’ tolerance to venous therapy and treatment outcomes. On the other hand, weak vascular tissue can suffer from insufficient nutrients. Under such circumstances, vessels are more likely to sustain mechanical and chemical damage, further increasing the risk of complications (Jia Z et al., 2020). Clinical practitioners should integrate nutritional risk screening for older adults into routine nursing assessments.\u003c/p\u003e\n\u003cp\u003eAs shown by several retrospective studies already, the presence of diabetes and vascular complications and kidney disease, along with age-related skin and connective tissue changes, dramatically increases the risk of PVC complications in elderly patients (Bonte F et al.,2019; Enes S M S et al.,2016). These findings were reported by Simin D et al. (2019), which may be due to prolonged intravenous therapy that impairs the quality of the vascular wall, local inflammation, or micro-thrombosis formation (Marsh N et al., 2023). The presence of other diseases confirms the role of comorbidity in AD. Thus, patients with comorbid chronic diseases should undergo enhanced multidisciplinary collaborative management, with a focus on controlling blood glucose levels, optimising circulatory function, and improving microcirculatory status to minimise damage to vascular endothelial cells caused by metabolic abnormalities and hemodynamic disorders.\u003c/p\u003e\n\u003cp\u003eSpecific individuals with impaired self-care abilities commonly experience reduced activity, leading to slowed venous return and prolonged retention of medications in local veins. Studies have shown that early mobilisation of the catheterised limb, maintaining normal daily activities, and adequate hydration can significantly delay the development of catheter-related complications (Chen Y et al., 2020). Consequently, it is advisable to perform puncture on limbs with desirable functional status. Moreover, patients are encouraged to engage in moderate functional exercises.\u003c/p\u003e\n\u003cp\u003eVascular aging functions as a key feature of aging, and repeated venous punctures increase patient discomfort and the risk of complications (Bahl A et al.,2024).\u0026nbsp;Da Silva's research team (Silva D A S et al.,2023)\u0026nbsp;(n=3,552) conducted a large-scale clinical study, reaching a confirmed conclusion that specialized teams can materialize lower repeated punctures (P \u0026lt; 0.01) and decrease catheter consumption by 30%. The introduction of vascular visualization technology can greatly improve the puncture success rate (P \u0026lt; 0.05) (Kuo C et al.,2025), especially in patients with hard punctures, while measuring the diameter of vessels (error \u0026lt; 1 mm). However, local hospitals that are not at the main level typically lack sufficient equipment and personnel. Specifically, only a handful of university hospitals utilize vascular visualization technology for central venous catheterization in their intensive care units/emergency departments. Meanwhile, it is mostly not at all utilised for PVC insertion in vulnerable populations. Consequently, it is vital to train in vascular visualization technology.\u003c/p\u003e\n\u003cp\u003eAccording to the INS guidelines, an ideal fixation device should keep the catheter stable while allowing for visibility of the site. At the moment, all use 3M transparent dressings. These dressings are highly transparent, allowing the user to see changes at the puncture site and around it. However, the sweating of the patient will eventually cause them to curl and peel off which reduces its fixation efficiency and causes catheter dislodgement. When this situation is in place, various complications may arise. (Huang et al.,2021) Therefore, for future academic pursuits, it would be better to focus on the curling and peeling off of transparent dressings in hot, sweaty conditions.\u003c/p\u003e\n\u003cp\u003eAccording to a logistic regression analysis, patients who had pain were more prone to develop PVC complications (OR=7.13) (Welyczko N et al et al., 2020). Studies related to this phenomenon have evidently confirmed that active inquiry regarding abnormal sensations can end complications and pain in 24 hours (Lee S et al., 2019). Removing the tube can reduce the risk from 19.7% to 5.3%. \u0026nbsp;Therefore, it is recommended to implement a dynamic monitoring system centered on pain: elderly patients should be assessed every 1 to 2 hours through the use of the “four-step observation method”, which can not only centre on distinguishing whether pain is of organic or psychogenic origin, but also dynamically record intensity, and probe into localization characteristics. The routine nursing protocol for assessing pain can help detect complications early and improve the safety of intravenous therapy significantly.\u003c/p\u003e\n\u003cp\u003eAccording to Trivedi et al. in 2022, the risk of complications with peripheral intravenous catheters increases with large volume infusions. The processes concerned may include: (i) an upsurge in the activity of shear stress system owing to hemodynamic alterations; (ii) osmotic departure of intravascular liquid component owing to large volume liquid administration, damages of endothelial cells. Hence, it is encouraged to prioritize alternative options. PICC lines are universally employed in clinical practice owing to their advantages of high safety, high success rate, and long retention time. Nonetheless, they also carry a risk of impaired blood flow and a high incidence of thrombosis. Thus, appropriate venous access options should be selected in clinical practice based on the patient’s condition.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e4.3 Strengths and Limitations\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study employed four machine learning algorithms-RF, LR, XGBoost, and BLR-to model the risk prediction of peripheral intravenous catheter complications in elderly patients\u0026nbsp;(Liu et al., 2021). The evaluation metrics selected were AUC, AUPRC, sensitivity, specificity, PPV, NPV, accuracy, and Brier score, and the performance of the four algorithms was compared. To the best of our knowledge, we pioneered the development and validation of a predictive model for peripheral intravenous catheter-related complications in elderly patients using machine-learning algorithms. Random forest achieved comparatively desirable predictive accuracy and outperformed other models in terms of predictive efficiency.\u003c/p\u003e\n\u003cp\u003eNonetheless, our study has several limitations. First, since the data were collected from a single study, further research is needed to validate the generalizability of our model. Second, while the model can assist clinicians in identifying high-risk patients for complications promptly, it does not provide specific treatment recommendations. Multicenter data and treatment information are needed to provide more conclusive results and better guidance.\u003c/p\u003e"},{"header":"5.0 Conclusion","content":"\u003cp\u003eIn this study, we proposed and validated a machine learning-based predictive model for predicting the development of peripheral venous catheter-related complications in elderly patients. As our research findings evidently demonstrated, the random forest model can accurately predict the development of peripheral venous catheter-related complications. Identifying high-risk populations can provide timely opportunities for close monitoring and intervention. This model is excepted to promote early diagnosis and influence the treatment decision-making process.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAuthor Contributions\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eXiao Chen,Yuchen Guo,Zhimin Liang\u0026nbsp;did the conception and design, or acquisition of data, or analysis and interpretation of data.\u0026nbsp;Xiao\u0026nbsp;Chen,Lingli Li involved in drafting the manuscript or revising it critically for important intellectual content.\u0026nbsp;Xiao\u0026nbsp;Chen,Yuchen\u0026nbsp;Guo,Zhimin\u0026nbsp;Liang,Lingli Li given final approval of the version to be published.\u0026nbsp;Xiao\u0026nbsp;Chen,\u0026nbsp;Yuchen\u0026nbsp;Guo\u0026nbsp;Agreed did the Project Administration.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors thank and thank all clinicians and staff who collected the data.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflicts of Interest\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThere’s no Conflict of Interest Among the Author’s.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Availability Statement\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;The data are not publicly available due to privacy or ethical restrictions. But available Upon Request to the Corresponding Author.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe research received no funding or any form of grant\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eNickel B. Peripheral intravenous access: applying infusion therapy standards of practice to improve patient safety. Crit Care Nurse.2019;39(1):61-71. doi:10.4037/ccn2019790.\u003c/li\u003e\n\u003cli\u003eChen YM, Fan XW, Liu MH, et al. Risk factors for peripheral venous catheter failure: a prospective cohort study of 5345 patients. J Vasc Access. 2022;23(6):911-921.doi:10.1177/11297298211015035.\u003c/li\u003e\n\u003cli\u003eMiliani K, Taravella R, Thillard D, et al. Peripheral venous catheter-related adverse events: evaluation from a multicentre epidemiological study in France (the CATHEVAL Project). PLoS One.2017;12(1):e0168637. doi:10.1371/journal.pone0168637\u003c/li\u003e\n\u003cli\u003eAnastasia G. World Population Prospects 2022: Summary of Results. 2022 [cited 2023 Jun 18]. In:United Nations Department of Economic and Social Affairs, Population Division [Internet]. Available from: https://www.un.org/development/desa/pd/content/World-Population-Prospects-2022\u003c/li\u003e\n\u003cli\u003eLarsen E, Keogh S, Marsh N, Rickard C. Experiences of peripheral IV insertion in hospital: a qualitative study. Br J Nurs. 2017;26(19): S18-S25. doi:10.12968/bjon.2017.26.19.S18.\u003c/li\u003e\n\u003cli\u003eSlosse C, Manneville F, Ricci L, et al. Development and evaluation of an algorithm for peripheral venous catheter placement (ALCOV): protocol for a quasi-experimental study[J]. BMJ OPEN, 2024,14(6):10.\u003c/li\u003e\n\u003cli\u003eKrein SL, Saint S, Trautner BW, et al. Patient-reported complications related to peripherally inserted central catheters: a multicentre prospective cohort study. BMJ Qual Saf. 2019;28(7):574-581. doi:10.1136/bmjqs-2018-008726\u003c/li\u003e\n\u003cli\u003eResnick O, Ahmad W A, Bancovsky D, et al. Predicting factors for complications in peripheral intravenous catheters in the pediatric population[J]. ACTA PAEDIATRICA, 2021,110(5):1639-1644\u003c/li\u003e\n\u003cli\u003eGong Z, Zhang J, Hou J, et al. Drug Extravasation in a Large General Hospital in Hunan, China: A Retrospective Survey[J]. RISK MANAGEMENT AND HEALTHCARE POLICY, 2021,14:4931-4938\u003c/li\u003e\n\u003cli\u003eLones M A. Avoiding common machine learning pitfalls[J]. PATTERNS, 2024,5(10):16.\u003c/li\u003e\n\u003cli\u003eWang Y, Lei L, Ji M, et al. Predicting postoperative delirium after microvascular decompression surgery with machine learning[J]. JOURNAL OF CLINICAL ANESTHESIA, 2020,66:6.\u003c/li\u003e\n\u003cli\u003eSu F, Chao J, Liu P, et al. Prognostic models for breast cancer: based on logistics regression and Hybrid Bayesian Network[J]. BMC MEDICAL INFORMATICS AND DECISION MAKING, 2023,23(1):15.\u003c/li\u003e\n\u003cli\u003eNickel B, Gorski L, Kleidon T, et al. Infusion Therapy Standards of Practice, 9th Edition[J]. JOURNAL OF INFUSION NURSING, 2024,47:291.\u003c/li\u003e\n\u003cli\u003eFu, J.; Zhao, L.; Tian, Y.; Liu, Y.; Zhang, H.; Li, H. Impact of Terminal Tip Location of Midline Catheters on Catheter‐related Complications and Indwelling Duration in Elderly Patients. Nursing Open, 2022.\u003c/li\u003e\n\u003cli\u003eChen Y, Fan X, Liu M, et al. Risk factors for peripheral venous catheter failure: A prospective cohort study of 5345 patients[J]. JOURNAL OF VASCULAR ACCESS, 2022,23(6):911-921.\u003c/li\u003e\n\u003cli\u003eGras E, Jean A, Rocher V, et al. Incidence of and risk factors for local complications of peripheral venous catheters in patients older than 70 years: Empirical research quantitative[J]. JOURNAL OF CLINICAL NURSING, 2023,32(15-16):5000-5009.\u003c/li\u003e\n\u003cli\u003eJia Z, Moheb M E, Nordestgaard A, et al. The Geriatric Nutritional Risk Index is a powerful predictor of adverse outcome in the elderly emergency surgery patient[J]. JOURNAL OF TRAUMA AND ACUTE CARE SURGERY, 2020,89(2):397-404.\u003c/li\u003e\n\u003cli\u003eBonte F, Girard D, Archambault J, et al. Skin Changes During Ageing[J]. Subcell Biochem, 2019,91:249-280.\u003c/li\u003e\n\u003cli\u003eEnes S M S, Opitz S P, de Faro A R M D, et al. Phlebitis associated with peripheral intravenous catheters in adults admitted to hospital in the Western Brazilian Amazon[J]. REVISTA DA ESCOLA DE ENFERMAGEM DA USP, 2016,50(2):261-269.\u003c/li\u003e\n\u003cli\u003eSimin D, Milutinovic D, Turkulov V, et al. Incidence, severity and risk factors of peripheral intravenous cannula-induced complications: An observational prospective study[J]. JOURNAL OF CLINICAL NURSING, 2019,28(9-10):1585-1599.\u003c/li\u003e\n\u003cli\u003eMarsh N, Larsen E N, O\u0026apos;Brien C, et al. Comparing the use of midline catheters versus peripherally inserted central catheters for patients requiring peripherally compatible therapies: A pilot randomised controlled trial (the compact trial) [J]. INFECTION DISEASE \u0026amp; HEALTH, 2023,28(4):259-264.\u003c/li\u003e\n\u003cli\u003eChen Y, Chen H, Yang J, et al. Patterns and risk factors of peripherally inserted central venous catheter-related symptomatic thrombosis events in patients with malignant tumors receiving chemotherapy.J Vasc Surg Venous Lymphat Disord. 2020;8(6):919-929. doi:10.1016/j.jvsv.2020.01.010.\u003c/li\u003e\n\u003cli\u003eBahl A, Alsbrooks K, Zazyczny K A, et al. An Improved Definition and SAFE Rule for Predicting Difficult Intravascular Access (DIVA) in Hospitalized Adults[J]. JOURNAL OF INFUSION NURSING, 2024,47(2):96-107.\u003c/li\u003e\n\u003cli\u003eSilva D A S, de Lima T R, Goncalves L. \u0026quot;Academia da Saude\u0026quot; program: mapping evidence from the largest health promotion community program in Brazil[J]. FRONTIERS IN PUBLIC HEALTH, 2023,11:9.\u003c/li\u003e\n\u003cli\u003eKuo C, Lee W, Ke Y. Ultrasound-guided peripheral intravenous access in adults: A randomized crossovercontrolled trial[J]. INTERNATIONAL EMERGENCY NURSING, 2025,79:8.\u003c/li\u003e\n\u003cli\u003eHuang, L.-S.; Huang, Y.; Hu, J. Current Practices of Peripheral Intravenous Catheter Fixation in Pediatric Patients and Factors Influencing Pediatric Nurses\u0026rsquo; Knowledge, Attitude and Practice Concerning Peripheral Intravenous Catheter Fixation: A Cross-Sectional Study.BMC Nursing, 2021.\u003c/li\u003e\n\u003cli\u003eWelyczko N. Peripheral intravenous cannulation: reducing pain and local complications[J]. Br J Nurs, 2020,29(8):S12-S19.\u003c/li\u003e\n\u003cli\u003eLee S, Kim K, Kim J. A Model of Phlebitis Associated with Peripheral Intravenous Catheters in Orthopedic Inpatients[J]. INTERNATIONAL JOURNAL OF ENVIRONMENTAL RESEARCH AND PUBLIC HEALTH, 2019,16(18):11.\u003c/li\u003e\n\u003cli\u003eTrivedi, M. R.; Bochiya, G. P.; Modi, J. V. A Prospective Study of Peripheral Inserted Venous Catheter Related Local Complications. International Surgery Journal, 2022.\u003c/li\u003e\n\u003cli\u003eCarr, P. J., Rippey, J. C. R., Cooke, M. L., Higgins, N. S., Trevenen, M., Foale, A., \u0026amp; Rickard, C. M. (2018). From insertion to removal: A multicenter survival analysis of an admitted cohort with peripheral intravenous catheters inserted in the emergency department. \u003cem\u003eInfection Control \u0026amp; Hospital Epidemiology\u003c/em\u003e, \u003cem\u003e39\u003c/em\u003e(10), 1216\u0026ndash;1221. https://doi.org/10.1017/ice.2018.190\u003c/li\u003e\n\u003cli\u003eEvi Diana Omar, Mat, H., Karim, A., Ridwan Sanaudi, Ibrahim, F., Mohd Azahadi Omar, Hafiz, Z., Vivek Jayaraj, \u0026amp; Bak Leong Goh. (2024). Comparative Analysis of Logistic Regression, Gradient Boosted Trees, SVM, and Random Forest Algorithms for Prediction of Acute Kidney Injury Requiring Dialysis After Cardiac Surgery. \u003cem\u003eInternational Journal of Nephrology and Renovascular Disease\u003c/em\u003e, \u003cem\u003eVolume 17\u003c/em\u003e, 197\u0026ndash;204. https://doi.org/10.2147/ijnrd.s461028\u003c/li\u003e\n\u003cli\u003eHirasawa, R., Kazuhiro Oinuma, Hagiwara, S., Sato, T., Yuya Kawarai, Miura, Y., Nakamura, J., \u0026amp; Seiji Ohtori. (2025). Collared fully hydroxyapatite-coated femoral components reduce early periprosthetic femoral fractures in total hip arthroplasty with the direct anterior approach. \u003cem\u003eThe Bone \u0026amp; Joint Journal\u003c/em\u003e, \u003cem\u003e107-B\u003c/em\u003e(10), 1011\u0026ndash;1019. https://doi.org/10.1302/0301-620x.107b10.bjj-2024-1494.r1\u003c/li\u003e\n\u003cli\u003eLiu, J., Wu, J., Liu, S., Li, M., Hu, K., \u0026amp; Li, K. (2021). Predicting mortality of patients with acute kidney injury in the ICU using XGBoost model. \u003cem\u003ePLOS ONE\u003c/em\u003e, \u003cem\u003e16\u003c/em\u003e(2), e0246306. https://doi.org/10.1371/journal.pone.0246306\u003c/li\u003e\n\u003cli\u003eMarsh, N., Larsen, E. N., Takashima, M., Kleidon, T., Keogh, S., Ullman, A. J., Mihala, G., Chopra, V., \u0026amp; Rickard, C. M. (2021). Peripheral intravenous catheter failure: A secondary analysis of risks from 11,830 catheters. \u003cem\u003eInternational Journal of Nursing Studies\u003c/em\u003e, \u003cem\u003e124\u003c/em\u003e(1), 104095. https://doi.org/10.1016/j.ijnurstu.2021.104095\u003c/li\u003e\n\u003cli\u003eShah, T. (2017, December 6). \u003cem\u003eAbout Train, Validation and Test Sets in Machine Learning\u003c/em\u003e. Medium; TDS Archive. https://medium.com/data-science/train-validation-and-test-sets-72cb40cba9e7\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003eTables 1 to 3 are available in the Supplementary Files section.\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"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":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"machine learning, elderly patients, complications of peripheral venous cathters, severity of complications, prediction model","lastPublishedDoi":"10.21203/rs.3.rs-8197291/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8197291/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eAim: \u003c/strong\u003eTo establish a prediction model for complications tightly correlated with peripheral intravenous catheter placement in elderly patients using four machine learning methods, identify risk factors, and comprehensively assess the models to select the optimal one.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDesign: \u003c/strong\u003eA Prospective Cohort Study\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods: \u003c/strong\u003eWe collected elderly patients admitted to a tertiary hospital in Chengdu, Sichuan Province, from November 2023 to September 2024, includingbaseline information, laboratory tests, risk assessments, IV catheter treatment, and catheterization procedures. Risk factors for complications were identified through stepwise regression, which was predominantly intended to construct risk prediction models using random forest, logistic regression, gradient boosting, and Bayesian logistic regression. Model's performance was systematically assessed through ROC, sensitivity, specificity, brier and accuracy.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults: This\u003c/strong\u003e study involved a total of 750 patients, with a peripheral intravenous catheter complications rate of 32.31%. The prediction model comprised seven variables, nutritional risk, comorbidities, self-care ability, vascularcondition, curlededges, daily input volume and pain score. Amongthe four machine learning models, the random forest model displayed the optimal predictive performance.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusion: \u003c/strong\u003eThe random forest-based risk prediction model for peripheral intravenous catheter complications in elderly patients demonstrated the best predictive performance and can assist healthcare professionals in identifying high-risk factors.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eImplications for the Profession and/or Patient Care \u003c/strong\u003eEnhancing the ability of clinical nurses to reduce the incidence of peripheral intravenous cathetercomplications and improve patient outcomes.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eImpact\u003c/strong\u003e: The prediction model demonstrates good ability to identify the risk of peripheral intravenous catheter complications in elderly patients is helpful to provide personalized care.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ePatient or Public Contribution\u003c/strong\u003e: Participants were hospitalized elderly patients. First of all, before the investigation and research, a team is formed to discuss the concept, research purpose, method, significance, etc., and determine the research tools. Second, by reasonably explaining the study to patients to seek informed consent from the patient and sign it.\u003c/p\u003e","manuscriptTitle":"Construction of a Machine Learning-Based Predictive Model for Complications Associated with Peripheral Intravenous Catheter Placement in Elderly Patients: A Prospective Cohort Study","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-05-11 10:34:13","doi":"10.21203/rs.3.rs-8197291/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewersInvited","content":"","date":"2026-04-30T15:46:04+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-04-07T16:00:17+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-12-30T12:06:35+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-12-27T10:37:51+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2025-12-27T10:27:46+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"83741416-8bbb-4b9b-ac3a-ee3bf5c1c700","owner":[],"postedDate":"May 11th, 2026","published":true,"recentEditorialEvents":[{"type":"reviewersInvited","content":"16","date":"2026-04-30T15:46:04+00:00","index":"","fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":67848597,"name":"Health sciences/Diseases"},{"id":67848598,"name":"Health sciences/Health care"},{"id":67848599,"name":"Health sciences/Medical research"},{"id":67848600,"name":"Health sciences/Risk factors"}],"tags":[],"updatedAt":"2026-05-11T10:34:13+00:00","versionOfRecord":[],"versionCreatedAt":"2026-05-11 10:34:13","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8197291","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8197291","identity":"rs-8197291","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}