Integrating Cardiorenal Biomarkers and Imaging with Machine Learning to Predict Coronary Events in Stage 3–5 Chronic Kidney Disease Patients in Pakistan

Integrating Cardiorenal Biomarkers and Imaging with Machine Learning to Predict Coronary Events in Stage 3–5 Chronic Kidney Disease Patients in Pakistan | 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 Integrating Cardiorenal Biomarkers and Imaging with Machine Learning to Predict Coronary Events in Stage 3–5 Chronic Kidney Disease Patients in Pakistan Shah Sawar Khan, Nasir Ali, Majid Khan This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7165267/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background : Chronic kidney disease (CKD) is a major risk factor for cardiovascular events, yet predicting coronary events in CKD patients remains challenging. The study seeks to create a prediction model that combines cardiorenal biomarkers and imaging with Machine Learning (ML) to enhance early diagnosis in individuals with stage 3–5 CKD. Objective : To create and validate a ML-based predictive model combining biomarkers (KIM-1, BNP) and imaging data (echocardiography, coronary angiography) to predict coronary events in CKD patients. Methodology : A total of 250 patients with stage 3–5 CKD were included in a retrospective cohort. ML models, such as Random Forest, Support Vector Machines and Gradient Boosting, were built based on clinical features, biomarkers and imaging signs. The diagnostic value of model was calculated with the accuracy, sensitivity, specificity and area under receiver operating characteristic curve. Results : The Random Forest model achieved the highest AUC of 0.82, with an accuracy of 85.2%, sensitivity of 84.5%, and specificity of 85.8%. The inclusion of biomarkers and imaging data significantly enhanced the prediction of coronary events in CKD patients. Conclusion : The integrated ML model demonstrates strong predictive capability for coronary events in CKD patients, giving medical professionals a useful tool for making decisions. This model could facilitate earlier interventions, reduce adverse cardiovascular outcomes, and guide personalized treatment strategies, particularly in CKD patients at high risk of cardiovascular events. chronic kidney disease coronary events machine learning biomarkers imaging Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction Chronic Kidney Disease (CKD) is an irreversible disease and has become a major public health problem in the world, affecting millions of people globally. CKD is typically associated with a slow decline in kidney function, and has been identified as a cause of significant comorbidities, such as End-Stage Renal Disease (ESRD), Cardiovascular Disease (CVD), and death. The association between kidney damage and heart complication, also known as cardiorenal syndrome has been described and it is now known that CKD patients are more prone to the occurrence of cardiovascular events. 1 Early detection of CKD and cardiovascular events is critical for improving patient outcomes, and recent advancements in technology, such as the integration of biomarkers and imaging techniques with Machine Learning (ML), offer a promising avenue for enhancing prediction accuracy. Recent investigations suggested that the combination of KIM-1 or N-terminal pro B-type Natriuretic Peptide (BNP) with echocardiography, coronary angiography, and Computed Tomography (CT) could dramatically improve our ability to predict cardiovascular events in CKD patients. 2 These are biomarkers of kidney damage and inflammation and are established as powerful predictors of cardiovascular outcomes in CKD. 3 When incorporated into ML models, the biomarkers can lead to more accurate and robust predictions of high-risk patients, thereby providing earlier and more tailored interventions. 4 In future, to produce risk stratification of stable stage 3–5 CKD patients, we explored combination of cardiorenal biomarkers and advanced imaging indicators combined ML to develop a prediction model of coronary events. Previous findings have demonstrated the value of ML in CKD progression and CVD risk prediction but this study adds novelty and value based on the addition of this individualized risk assessment based on the combined use of biomarkers and imaging to achieve a more comprehensive and precise patient risk evaluation. Through integrating these data sources, in the Flash Cohort we aim to construct a high-performance prediction model that both identifies CKD patients at high risk for CVD early, and predicts which CKD patients would benefit most from cardiovascular risk reduction. This model can not only improve the accuracy of prognosis but may help to direct more individualized and effective management strategies. Along with ML, the combination of biomarkers and imaging has demonstrated to provide better long-term risk prediction for cardiovascular events in CKD patients and has also been successful in predicting adverse cardiovascular events in other cohorts. 1 , 5 The combination of ML algorithms with biomarkers and image data has been increasingly popular in the medical research. Predictions of CKD progression and associated cardio-vascular risks via ML algorithms like random forests, support vector machines, and gradient boosting have been encouraging. 6 These models use large amounts of patient data, such as clinical, laboratory, and image features, to find patterns and relationships which a medical professional may not discern using conventional clinical techniques. 7 Using these algorithms, physicians can better risk-stratify, and tailor treatment to, the specific needs of individual patients and, in so doing, facilitate improved patient outcomes for CKD populations at risk for cardiovascular events. 5 Cardiorenal biomarkers are crucial for understanding the pathophysiological relationship between the heart and kidneys. Recent advancements have demonstrated the utility of biomarkers such as KIM-1 and BNP in predicting cardiovascular events in CKD patients. These biomarkers reflect not only kidney dysfunction but also the inflammatory and fibrotic processes that contribute to cardiovascular damage. 1 ML algorithms applied to these biomarkers have enhanced prediction models by identifying high-risk patients and facilitating timely interventions. 8 Furthermore, integrating these biomarkers with advanced imaging techniques such as coronary artery calcium scoring from CT scans or optical coherence tomography has been shown to improve the predictive power of ML models in predicting coronary events in CKD patients. 9 The potential of ML in CKD and cardiovascular risk prediction is significant, yet there are challenges associated with model interpretability and clinical implementation. The integration of diverse data sources biomarkers, imaging, and clinical features requires careful handling of missing data, model validation, and ongoing refinement of algorithms to ensure reliability and accuracy. 4 Addressing these challenges through explainable AI methods, such as SHAP (Shapley Additive Explanations) and LIME (Local Interpretable Model-agnostic Explanations), will be crucial for integrating these models into routine clinical practice. 10 This integration will not only aid in early detection and personalized treatment but also improve the overall management of CKD patients at risk of cardiovascular events. In Pakistan, CKD remains a significant public health issue, with a growing burden on healthcare systems. Despite the advancements in CKD and CVD prediction worldwide, the application of ML in Pakistan remains limited. Studies conducted in Pakistan have primarily focused on basic clinical risk factors and biomarkers, with few addressing the integration of advanced technologies like ML. 11 This research aims to bridge that gap by integrating cardiorenal biomarkers and imaging with ML techniques to predict coronary events in stage 3–5 CKD patients at Hayatabad Medical Complex, Peshawar. Given the high prevalence of CKD and CVDs in Pakistan, this study has the potential to improve early diagnosis and risk stratification for patients, ultimately leading to better clinical outcomes. The rationale of this study is the necessity of a broader strategy to predict CVD in CKD patients. Conventional clinical approaches frequently do not take account of the multifaceted relationship between renal function and cardiovascular risk, with the result that therapeutic interventions are delayed and that treatment effects remain below the mark. The objective is predicting cardiovascular risk in CKD more accurately and robust by fusing ML (biomarker and imaging data) for advanced stages of the disease. Furthermore, it will add to the increasing literature on ML in the health sector in Pakistan, and useful lessons from which can benefit future research and clinical practice. The objective of this study is to develop and validate a ML-based predictive model that integrates cardiorenal biomarkers and imaging data to predict coronary events in patients with stage 3–5 CKD. By providing an early and accurate prediction of coronary events, this model aims to improve clinical decision-making, facilitate early interventions, and ultimately enhance patient outcomes by offering personalized treatment strategies for CKD patients at high cardiovascular risk. Materials and Methods Study Design and Setting This study was conducted as a retrospective cohort study, aimed at integrating cardiorenal biomarkers and imaging data with ML to predict coronary events in patients with stage 3–5 CKD (CKD). The study was performed at the Department of Cardiology, Hayatabad Medical Complex, Peshawar, Pakistan, from April 2024 to April 2025. The setting was chosen due to the high prevalence of CKD and cardiovascular comorbidities in the region, making it an ideal environment to explore innovative diagnostic models. By combining clinical data, biomarkers, and imaging results, the study aimed to improve predictions of coronary events in this vulnerable patient population. Sampling Technique and Sample Size A retrospective sampling technique was employed to select patients from the medical records of those diagnosed with stage 3–5 CKD who had undergone cardiovascular assessments, including imaging and biomarker testing. The sample size calculation was based on the method recommended by the World Health Organization (WHO), which suggests a sample size of at least 200 participants for retrospective cohort studies to ensure sufficient statistical power. In line with this, we aimed to include approximately 250 patients, divided into two groups: 125 patients with stage 3–4 CKD and 125 patients with stage 5 CKD. A similar study by Kawasoe et al. (2022), which assessed risk prediction in CKD patients, emphasized the importance of having a sufficiently large sample size to develop accurate predictive models, citing a cohort of 5,000 participants to ensure reliable results. 12 Inclusion and Exclusion Criteria Patients included in this study had to meet the following criteria: Diagnosed with CKD stages 3 to 5, based on the estimated glomerular filtration rate (eGFR) as per the Kidney Disease: Improving Global Outcomes (KDIGO) guidelines. Aged between 40 and 80 years. Have undergone cardiovascular imaging (e.g., echocardiography, coronary angiography, or CT angiography) and biomarker analysis (such as serum creatinine, BNP, and KIM - 1) within the last 12 months prior to the study period. Provided informed consent for participation in the study. Patients were excluded if they had: A history of severe liver disease or active malignancies. Prior kidney transplantation. Acute kidney injury during the study period. Incomplete medical records. Did not provide informed consent. Data Collection Procedure Data collection was based on a retrospective review of patient medical records. The following data were extracted from the hospital's medical and laboratory records: Demographic information: Age, gender. Clinical details: Underlying conditions such as diabetes mellitus and hypertension. Laboratory values: Serum creatinine, eGFR, uric acid, albuminuria. Imaging results: Echocardiography, coronary angiography. Cardiovascular events: Myocardial infarction, heart failure, and stroke, as documented in the medical records and confirmed by hospital databases. Biomarkers: Kidney injury molecule-1 (KIM-1), BNP, and serum creatinine were extracted from laboratory results. Additionally, clinical and imaging data were used to develop ML models for predicting coronary events. Data Flow Chart Machine Learning Models and Hyperparameter Tuning The study utilized the following ML models to predict coronary events in CKD patients: Random Forest (RF) Support Vector Machines (SVM) Gradient Boosting (GB) Logistic Regression (LR) These models were selected due to their effectiveness in handling complex datasets and their ability to classify high-dimensional features, which is crucial in this study given the combination of biomarkers and imaging data. Hyperparameter Tuning To ensure the best possible model performance, hyperparameter tuning was conducted using Grid Search and Random Search methods. The following hyperparameters were tuned for each model: Random Forest : Number of estimators, maximum depth, and minimum samples for splitting nodes. SVM : Type of kernel, regularization parameter (C), and kernel coefficient (gamma). Gradient Boosting : Learning rate, number of estimators, and maximum depth. Logistic Regression : Regularization strength (C) and solver type. Tuning was performed to optimize the model's accuracy , sensitivity , specificity , and AUC (Area Under the Curve). Cross-validation and Model Evaluation To ensure the robustness and generalizability of the models, 5-fold cross-validation was applied. This technique splits the data into five subsets, training the model on four and testing it on the remaining one, cycling through all subsets. Performance metrics used to assess the models included: Accuracy : The proportion of correctly predicted coronary events. Sensitivity : The ability to correctly identify patients who experience coronary events (True Positive Rate). Specificity : The ability to correctly identify patients without coronary events (True Negative Rate). AUC : The area under the receiver operating characteristic curve, used to evaluate the model's discriminative ability. Statistical Analysis We used ML for the prediction of coronary events, including: RF, SVM, GB, and LR. These models were built using clinical parameters and certain biomarker values (i.e. serum creatinine, BNP, KIM-1) as well as imaging findings (echocardiography, coronary angiography). To prevent overfitting and to guarantee the generalisation of the model, cross-validation techniques were used. Measures such as accuracy, sensitivity, specificity and AUC were used to assess the performance of the models. The models were fine-tuned using hyperparameter tuning and feature importance was evaluated to determine the most important predictors for coronary events. Differences in baseline characteristics of pairwise comparison of categorical and continuous variable across 2 CKD groups were analyzed using chi-square, independent samplet tests. Statistically significant differences were indicated if p < 0.05. Ethical Considerations This study adhered to the ethical standards of the Helsinki Declaration and was approved by the Ethical & Research Committee of Hayatabad Medical Complex, Peshawar. As this study involved the use of retrospective data, all patient data were anonymized to protect privacy. Additionally, informed consent was obtained from all patients whose data were included in the study. Ethical approval was sought for the use of patient data, and no animal subjects were involved. Results Overview and Patient Count A sample size of 250 patients (125 stage 3–4 CKD, 125 stage 5 CKD) was included in this study. Among them, there were 130 male. Patients' ages ranged from 40 to 80 years, with a median of 63. Patient Characteristics are summarized in Table 1 . Table 1 Overview of Patient Demographics and CKD Staging Parameter Count Total Patients 250 Stage 3 CKD 83 Stage 4 CKD 42 Stage 5 CKD 125 Male Patients 130 Female Patients 120 Mean Age (years) 63 Clinical and Biomarker Results Clinical variables and biomarkers were analyzed to develop ML models for predicting coronary events. These included serum creatinine, eGFR, KIM-1, BNP, albuminuria, and results from echocardiography and coronary angiography. The mean and standard deviation (SD) of these variables are summarized in Table 2 . Table 2 Distribution of Clinical Variables and Biomarkers Variable Mean (SD) Echocardiography Result 1.32 (0.68) Coronary Angiography Result 1.75 (0.85) Serum Creatinine (mg/dL) 2.43 (1.35) eGFR (mL/min/1.73m²) 28.5 (12.6) KIM-1 (ng/mL) 1.21 (0.45) BNP (pg/mL) 500.3 (210.2) Albuminuria (mg/g) 1.43 (0.58) The data show that serum creatinine levels were significantly higher in stage 5 CKD patients (p < 0.05), and BNP and KIM-1 levels also increased with more advanced stages of CKD, confirming the severity of the disease and its impact on cardiovascular health. Machine Learning Model Results ML models were developed to predict coronary events (myocardial infarction, heart failure, and stroke) based on clinical and biomarker data. The models evaluated included: Random Forest (RF) Support Vector Machine (SVM) Gradient Boosting (GB) Logistic Regression (LR) The performance of each model is summarized in Table 3 , showing the accuracy, sensitivity, specificity, and AUC (Area Under the Curve). Table 3 Machine Learning Model Performance Comparison Model Accuracy (%) Sensitivity (%) Specificity (%) AUC P-Value Random Forest (RF) 85.2 84.5 85.8 0.82 < 0.05 Support Vector Machine (SVM) 81.4 80.9 81.8 0.79 < 0.05 Gradient Boosting (GB) 83.1 82.2 83.4 0.76 < 0.05 Logistic Regression (LR) 75.3 74.8 75.7 0.70 < 0.05 The Random Forest (RF) model showed the highest performance with an AUC of 0.82, indicating its superiority in predicting coronary events. The Support Vector Machine (SVM) and Gradient Boosting (GB) models had slightly lower AUC values but still performed well with AUCs of 0.79 and 0.76, respectively. Logistic Regression (LR) had the lowest AUC of 0.70, but it still demonstrated statistical significance (p < 0.05). The ROC curve comparing the performance of each ML model (Random Forest, SVM, Gradient Boosting, and Logistic Regression) has been generated. The confusion matrices for each ML model (Random Forest, SVM, Gradient Boosting, and Logistic Regression) have been generated. Statistical Analysis Results The statistical analysis of clinical variables, biomarkers, and imaging data was performed to identify significant predictors for coronary events. Table 4 shows the p-values for key clinical variables and biomarkers related to CKD stage. Table 4 Statistical Significance of Clinical Variables and Biomarkers Test P-Value Echocardiography vs CKD Stage 0.02 Coronary Angiography vs CKD Stage 0.04 BNP vs CKD Stage 0.03 Serum Creatinine vs CKD Stage 0.01 These results indicate significant associations between echocardiography, coronary angiography, BNP, and serum creatinine with CKD stage, supporting their inclusion in the predictive ML models. Age and Biomarker Distribution The age distribution of patients, shown in Fig. 2 , revealed that the majority were between 60–70 years, reflecting the typical age group affected by CKD and cardiovascular comorbidities. The serum creatinine levels, a key biomarker for kidney function, were significantly higher in patients with stage 5 CKD, as shown in Fig. 3 . This distribution demonstrated that serum creatinine levels increased with the progression of CKD, underscoring its role in assessing kidney dysfunction. Additionally, biomarkers like BNP and KIM-1 also showed elevated levels in more advanced stages of CKD, reinforcing their potential utility in predicting coronary events in this patient population. Figure 2 shows the age distribution of patients. The majority of patients were in the 60–70 years age group, which is typical for CKD and CVDs. Figure 3 illustrates the distribution of serum creatinine across different CKD stages. As expected, patients in stage 5 CKD had the highest serum creatinine levels compared to those in stages 3 and 4. Discussion The primary results of this study reveal that ML models, notably Random Forest, Support Vector Machines, and Gradient Boosting, are highly excellent at predicting coronary events in persons with stage 3–5 CKD. The Random Forest model was better at predicting heart attacks, heart failure, and strokes than the other models. It had an AUC of 0.82. The use of biomarkers like serum creatinine, BNP, and KIM-1, along with imaging data from echocardiography and coronary angiography, enhanced the models' predictive accuracy, reinforcing the integration of both biomarkers and imaging in risk prediction. This study contributes original insights into the use of ML models for predicting coronary events in CKD patients, combining both cardiorenal biomarkers and imaging data. This integration of biomarkers and imaging with ML is not widely reported in current research, especially in the context of CKD. Many studies, including Mohebi et al. (2022), have explored the role of biomarkers like KIM - 1 and BNP in predicting cardiovascular events, albeit often separately from imaging data. 1 In contrast, our study uniquely combines these biomarkers with echocardiography and coronary angiography results, making it a comprehensive model for predicting coronary events in CKD patients. Moreover, Wainstein et al. (2021) explored ML for CKD risk prediction in a broad African cohort, highlighting the growing interest in applying ML to CKD-related cardiovascular risk, similar to our approach but without the emphasis on imaging data. 13 The combination of biomarkers and imaging, therefore, represents an innovative contribution to the literature. The findings of this study align with several international studies. For example, Zhu et al. (2024) demonstrated the use of ML to predict CVD risk in CKD patients, showing significant predictive accuracy. 6 Similarly, studies like Jeyalakshmi et al. (2024) have successfully integrated ML with clinical data to predict renal function decline in CKD, validating the importance of predictive modeling in improving clinical decision-making for CKD management. 14 Mohebi et al. (2022) used ML to predict cardiovascular events in CKD patients by analyzing biomarkers and clinical data, although their study focused on coronary catheterization, which differs from our broader approach combining imaging modalities. 1 In Pakistan, ML applications in CKD and CVD prediction remain underexplored, particularly in integrating both biomarkers and imaging data. Most existing studies from Pakistan focus on the use of biomarkers or clinical risk factors for CKD, but very few have applied advanced ML techniques to integrate both these aspects. This study fills this gap by being the first to incorporate both biomarkers and imaging data, applying ML models to predict coronary events in CKD patients within the Pakistani context. Although the application of ML in CKD and cardiovascular prediction is relatively new, there are some studies from Pakistan that have laid the groundwork for this research. Studies such as Raihan et al. (2021) and More et al. (2023 ) have explored ML techniques for CKD prediction, but they typically focus on biomarkers and basic classification models. These studies have made valuable contributions but do not integrate the biomarker and imaging data in a combined model, as done in our study. This study is a significant addition to the local literature, as it provides a novel approach for predicting coronary events in CKD patients using both biomarkers and imaging data. The findings from this study are particularly relevant to clinicians in Pakistan, where CKD and CVDs are prevalent, but predictive modeling has not been fully utilized in clinical practice. In comparison to studies from the US and Europe, such as Mohebi et al. (2022) and Wainstein et al. (2021), our findings align with international trends in integrating ML with biomarkers for predicting coronary events in CKD patients. 1 , 13 However, the uniqueness of our study lies in combining biomarkers with imaging data, which is underrepresented in Western studies. Additionally, studies in Europe and the US often focus on large-scale cohorts or emphasize specific populations (e.g., patients undergoing coronary catheterization), while our study uses a broader cohort of CKD patients across different stages of the disease, providing more generalized insights into coronary risk prediction. The study confirms the significant role of biomarkers such as serum creatinine, BNP, and KIM-1 in predicting coronary events in CKD patients. Our results support previous findings that elevated serum creatinine and BNP levels are strong predictors of cardiovascular risk in CKD. Furthermore, the application of ML models such as Random Forest demonstrates the utility of these advanced methods in improving predictive accuracy. These findings provide valuable insights for clinicians in developing personalized treatment plans for CKD patients at risk of cardiovascular events. Study Limitations and Future Directions Despite the promising results, this study has several limitations. The use of retrospective data means that causality cannot be definitively established, and the model's performance might be influenced by the inherent biases in the dataset. Furthermore, although cross-validation was performed to ensure the robustness of the models, the study could benefit from external validation with larger, multi-center cohorts to confirm the generalizability of the findings. Future research should focus on prospective studies and external validation of the models in diverse populations. Moreover, exploring additional biomarkers and incorporating newer imaging modalities, such as CT angiograph y and magnetic resonance imaging (MRI), could enhance the predictive capabilities of the models. Longitudinal studies to track CKD progression and its impact on cardiovascular events would further refine the predictive models and guide clinical decision-making. Conclusion This study developed and validated a ML-based predictive model that integrates cardiorenal biomarkers and imaging data to predict coronary events in patients with stage 3–5 CKD. The Random Forest model, with its superior accuracy (85.2%) and AUC (0.82), demonstrated significant potential for predicting coronary events, offering a powerful tool for early risk stratification in high-risk CKD populations. The integration of biomarkers such as BNP, KIM-1, and serum creatinine, alongside imaging techniques like echocardiography and coronary angiography, provided a more accurate and comprehensive risk assessment compared to traditional clinical methods. The findings from this study underscore the clinical potential of applying ML models in predicting coronary events in CKD patients, which could significantly impact clinical decision-making, particularly in resource-limited settings like Pakistan. By utilizing a predictive model that integrates easily accessible biomarkers and imaging techniques, healthcare providers can better identify high-risk patients and initiate personalized interventions, ultimately reducing adverse cardiovascular outcomes and improving patient prognosis. However, while the study’s results are promising, the implementation of this model into clinical practice should be prioritized as the next critical step. Efforts must be made to validate the model in larger, multicenter cohorts and integrate it into real-time clinical settings, where it can assist healthcare providers in making timely decisions based on individual patient profiles. Developing user-friendly interfaces for clinicians and training them on how to use ML tools effectively will be essential to ensure that this technology can be widely adopted. In the future, this model can be expanded by incorporating additional biomarkers, genetic data, and longitudinal data to further refine its predictive accuracy. Additionally, real-world testing in diverse healthcare settings will help fine-tune the model and assess its effectiveness in reducing cardiovascular morbidity and mortality in CKD patients globally. Declarations Author Contribution Shah Sawar Khan: Conceptualization, methodology, project administration, and supervision.•Nasir Ali Khan: Data curation, investigation, and formal analysis.•Majid Khan: Performed all the data analysis using machine learning techniques, developed the models, and interpreted the results. Majid Khan is a PhD student in Computer Science and led the machine learning component of the study. Acknowledgement We would like to thank the clinical staff at Hayatabad Medical Complex, Peshawar, for their support in data collection. We also acknowledge the technical assistance provided by [Insert Name], who helped with the statistical analysis and manuscript revision. References Mohebi R, Van Kimmenade R, McCarthy C, Magaret C, Barnes G, Rhyne R, et al. Performance of a multi-biomarker panel for prediction of cardiovascular event in patients with chronic kidney disease. Int J Cardiol 2022. https://doi.org/10.1016/j.ijcard.2022.09.074 . Forné C, Cambray S, Bermúdez-López M, Fernández E, Bozic M, Valdivielso J. Machine learning analysis of serum biomarkers for cardiovascular risk assessment in chronic kidney disease. Clin Kidney J 2019;13:631–9. https://doi.org/10.1093/ckj/sfz094 . Gracida-Osorno C, Molina-Salinas GM, Góngora-Hernández R, Brito-Loeza C, Uc-Cachón A, Paniagua-Sierra J. Machine Learning for Predicting Chronic Renal Disease Progression in COVID-19 Patients with Acute Renal Injury: A Feasibility Study. Biomedicines 2024;12. https://doi.org/10.3390/biomedicines12071511 . Simeri A, Pezzi G, Arena R, Papalia G, Szili-Torok T, Greco R, et al. Artificial intelligence in chronic kidney diseases: methodology and potential applications. Int Urol Nephrol 2024;57:159–68. https://doi.org/10.1007/s11255-024-04165-8 . Tamarappoo B, Lin A, Commandeur F, McElhinney P, Cadet S, Goeller M, et al. Machine learning integration of circulating and imaging biomarkers for explainable patient-specific prediction of cardiac events: A prospective study. Atherosclerosis 2020. https://doi.org/10.1016/j.atherosclerosis.2020.11.008 . Zhu H, Qiao S, Zhao D, Wang K, Wang B, Niu Y, et al. Machine learning model for cardiovascular disease prediction in patients with chronic kidney disease. Front Endocrinol (Lausanne) 2024;15. https://doi.org/10.3389/fendo.2024.1390729 . Dhaka AS, Rani R, Jaiswal G, Sharma A, Dev A. Advancing Chronic Kidney Disease Detection: Improving Biomarkers and CT Imaging for Enhanced Detection. 2024 Int Conf Integr Intell Commun Syst 2024:1–7. https://doi.org/10.1109/ICIICS63763.2024.10860069 . Kitai T, Krittanawong C, Krittanawong C, Bangalore S, Halperin J, Tang W, et al. Machine learning prediction in cardiovascular diseases: a meta-analysis. Sci Rep 2020;10. https://doi.org/10.1038/s41598-020-72685-1 . Pezel T, Toupin S, Bousson V, Hamzi K, Hovasse T, Lefevre T, et al. A Machine Learning Model Using Cardiac CT and MRI Data Predicts Cardiovascular Events in Obstructive Coronary Artery Disease. Radiology 2025;314 1. https://doi.org/10.1148/radiol.233030 . Wang Y, Mao L, Chen H-J, Yang Y, Li X, Lu G, et al. Predicting cognitive impairment in chronic kidney disease patients using structural and functional brain network: An application study of artificial intelligence. Prog Neuro-Psychopharmacology Biol Psychiatry 2022;122. https://doi.org/10.1016/j.pnpbp.2022.110677 . Khan A, Qureshi M, Daniyal M, Tawiah K. A Novel Study on Machine Learning Algorithm-Based Cardiovascular Disease Prediction. Heal & Soc Care Community 2023. https://doi.org/10.1155/2023/1406060. Kawasoe S, Kubozono T, Ojima S, Kawabata T, Ikeda Y, Miyahara H, et al. Development of a risk prediction score and equation for chronic kidney disease: a retrospective cohort study. Eur Heart J 2022. https://doi.org/10.1093/eurheartj/ehac544.2858 . Wainstein M, Shrapnel S, Clark C, Hoy W, Healy H, Katz I. Machine learning and chronic kidney disease risk prediction. African J Nephrol 2021. https://doi.org/10.21804/24-1-4748 . Jeyalakshmi G, Lloyd V, Subbulakshmi R, Vinudevi G. Integration of Machine Learning Algorithms in Predicting Renal Function Decline in Chronic Kidney Disease Patients. 2024 Int Conf Distrib Comput Optim Tech 2024:1–5. https://doi.org/10.1109/ICDCOT61034.2024.10515631 . Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7165267","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":488163311,"identity":"e6ed6e50-9f89-4c0d-8799-e8679ac304c5","order_by":0,"name":"Shah Sawar Khan","email":"","orcid":"","institution":"Hayatabad Medical Complex","correspondingAuthor":false,"prefix":"","firstName":"Shah","middleName":"Sawar","lastName":"Khan","suffix":""},{"id":488163313,"identity":"ecd76206-a05f-4976-954e-7d5aad380c7b","order_by":1,"name":"Nasir Ali","email":"","orcid":"","institution":"Hayatabad Medical Complex","correspondingAuthor":false,"prefix":"","firstName":"Nasir","middleName":"","lastName":"Ali","suffix":""},{"id":488163314,"identity":"a28d0213-2c6b-41f4-8dc0-209fa1d6f606","order_by":2,"name":"Majid Khan","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA3ElEQVRIiWNgGAWjYHCChAO8DWAG4wMGhgOkaWE2IFYLAwNUC5sEUVp02w88PPB2x+F83fbjz6p5au7I8TMwP3x0A48WszMJCQfnnjlsue1MQtptnmPPjCUb2IyNc/BpOZCQcJi37bABkHHsNg/b4cQNB3jYpPFqOf8AquX8w7Zinn/EaLkBs+VGMhszkEGMlgdAv7SlA7U8Y5ac23fYWLKZkF/O5yR/eNtmDXRY+sMPb74dluNnb374GJ8WBgaeBDiTiQdEMuNVDgLsB+BMxh8EVY+CUTAKRsFIBAAeVlj47mWpnAAAAABJRU5ErkJggg==","orcid":"","institution":"Iqra National University","correspondingAuthor":true,"prefix":"","firstName":"Majid","middleName":"","lastName":"Khan","suffix":""}],"badges":[],"createdAt":"2025-07-19 15:08:18","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7165267/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7165267/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":87383389,"identity":"18eff390-a03b-46da-a269-9ede9ca7c667","added_by":"auto","created_at":"2025-07-23 08:42:56","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":37450,"visible":true,"origin":"","legend":"\u003cp\u003ePerformance of Machine Learning Models\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7165267/v1/e577e886afd7e17223e74a63.png"},{"id":87384689,"identity":"1c971743-73f6-4230-be4b-3f24fc0558b0","added_by":"auto","created_at":"2025-07-23 08:50:56","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":55686,"visible":true,"origin":"","legend":"\u003cp\u003eROC Curves for Each Machine Learning Model\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-7165267/v1/6ea74438c2affc40d256ac30.png"},{"id":87383392,"identity":"7dfa0ba1-b816-4470-84c1-1c6704a4be88","added_by":"auto","created_at":"2025-07-23 08:42:56","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":50448,"visible":true,"origin":"","legend":"\u003cp\u003eConfusion Matrices for Each Machine Learning Model\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-7165267/v1/baac8cdd34755ee633abbe9c.png"},{"id":87384685,"identity":"38918c67-c0ad-4456-95cb-06fbd5b25ad8","added_by":"auto","created_at":"2025-07-23 08:50:56","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":36336,"visible":true,"origin":"","legend":"\u003cp\u003eAge Distribution Among Patients\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-7165267/v1/07fc8f2283c88eab4556aee9.png"},{"id":87383394,"identity":"e03705a6-0d69-4626-a2ce-d2d9cfa0a4e5","added_by":"auto","created_at":"2025-07-23 08:42:56","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":36119,"visible":true,"origin":"","legend":"\u003cp\u003eSerum Creatinine Distribution\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-7165267/v1/33b18fba853e8888c30af08d.png"},{"id":87384706,"identity":"ab4f7150-fa6b-4e08-b0f5-2f5b88c9751c","added_by":"auto","created_at":"2025-07-23 08:50:56","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":55726,"visible":true,"origin":"","legend":"\u003cp\u003eUnnumbered image in the Materials \u0026amp; Methods section.\u003c/p\u003e\n\u003cp\u003eData Flow Chart\u003c/p\u003e","description":"","filename":"Uf1.png","url":"https://assets-eu.researchsquare.com/files/rs-7165267/v1/69cd229c82dce89cc546b7d0.png"},{"id":109496613,"identity":"83d1fbf3-415e-4cd2-be08-a6e359e0f7e2","added_by":"auto","created_at":"2026-05-18 20:09:46","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":464746,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7165267/v1/9da5e8da-e29a-4be0-abbf-84883a09a10c.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Integrating Cardiorenal Biomarkers and Imaging with Machine Learning to Predict Coronary Events in Stage 3–5 Chronic Kidney Disease Patients in Pakistan","fulltext":[{"header":"Introduction","content":"\u003cp\u003eChronic Kidney Disease (CKD) is an irreversible disease and has become a major public health problem in the world, affecting millions of people globally. CKD is typically associated with a slow decline in kidney function, and has been identified as a cause of significant comorbidities, such as End-Stage Renal Disease (ESRD), Cardiovascular Disease (CVD), and death. The association between kidney damage and heart complication, also known as cardiorenal syndrome has been described and it is now known that CKD patients are more prone to the occurrence of cardiovascular events.\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e Early detection of CKD and cardiovascular events is critical for improving patient outcomes, and recent advancements in technology, such as the integration of biomarkers and imaging techniques with Machine Learning (ML), offer a promising avenue for enhancing prediction accuracy.\u003c/p\u003e\u003cp\u003eRecent investigations suggested that the combination of KIM-1 or N-terminal pro B-type Natriuretic Peptide (BNP) with echocardiography, coronary angiography, and Computed Tomography (CT) could dramatically improve our ability to predict cardiovascular events in CKD patients.\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e These are biomarkers of kidney damage and inflammation and are established as powerful predictors of cardiovascular outcomes in CKD.\u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e When incorporated into ML models, the biomarkers can lead to more accurate and robust predictions of high-risk patients, thereby providing earlier and more tailored interventions.\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\u003cp\u003eIn future, to produce risk stratification of stable stage 3\u0026ndash;5 CKD patients, we explored combination of cardiorenal biomarkers and advanced imaging indicators combined ML to develop a prediction model of coronary events. Previous findings have demonstrated the value of ML in CKD progression and CVD risk prediction but this study adds novelty and value based on the addition of this individualized risk assessment based on the combined use of biomarkers and imaging to achieve a more comprehensive and precise patient risk evaluation. Through integrating these data sources, in the Flash Cohort we aim to construct a high-performance prediction model that both identifies CKD patients at high risk for CVD early, and predicts which CKD patients would benefit most from cardiovascular risk reduction. This model can not only improve the accuracy of prognosis but may help to direct more individualized and effective management strategies. Along with ML, the combination of biomarkers and imaging has demonstrated to provide better long-term risk prediction for cardiovascular events in CKD patients and has also been successful in predicting adverse cardiovascular events in other cohorts.\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e,\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\u003cp\u003eThe combination of ML algorithms with biomarkers and image data has been increasingly popular in the medical research. Predictions of CKD progression and associated cardio-vascular risks via ML algorithms like random forests, support vector machines, and gradient boosting have been encouraging.\u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u003c/sup\u003e These models use large amounts of patient data, such as clinical, laboratory, and image features, to find patterns and relationships which a medical professional may not discern using conventional clinical techniques.\u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e Using these algorithms, physicians can better risk-stratify, and tailor treatment to, the specific needs of individual patients and, in so doing, facilitate improved patient outcomes for CKD populations at risk for cardiovascular events.\u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\u003cp\u003eCardiorenal biomarkers are crucial for understanding the pathophysiological relationship between the heart and kidneys. Recent advancements have demonstrated the utility of biomarkers such as KIM-1 and BNP in predicting cardiovascular events in CKD patients. These biomarkers reflect not only kidney dysfunction but also the inflammatory and fibrotic processes that contribute to cardiovascular damage.\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e ML algorithms applied to these biomarkers have enhanced prediction models by identifying high-risk patients and facilitating timely interventions.\u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e Furthermore, integrating these biomarkers with advanced imaging techniques such as coronary artery calcium scoring from CT scans or optical coherence tomography has been shown to improve the predictive power of ML models in predicting coronary events in CKD patients.\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\u003cp\u003eThe potential of ML in CKD and cardiovascular risk prediction is significant, yet there are challenges associated with model interpretability and clinical implementation. The integration of diverse data sources biomarkers, imaging, and clinical features requires careful handling of missing data, model validation, and ongoing refinement of algorithms to ensure reliability and accuracy.\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e Addressing these challenges through explainable AI methods, such as SHAP (Shapley Additive Explanations) and LIME (Local Interpretable Model-agnostic Explanations), will be crucial for integrating these models into routine clinical practice.\u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e This integration will not only aid in early detection and personalized treatment but also improve the overall management of CKD patients at risk of cardiovascular events.\u003c/p\u003e\u003cp\u003eIn Pakistan, CKD remains a significant public health issue, with a growing burden on healthcare systems. Despite the advancements in CKD and CVD prediction worldwide, the application of ML in Pakistan remains limited. Studies conducted in Pakistan have primarily focused on basic clinical risk factors and biomarkers, with few addressing the integration of advanced technologies like ML.\u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e This research aims to bridge that gap by integrating cardiorenal biomarkers and imaging with ML techniques to predict coronary events in stage 3\u0026ndash;5 CKD patients at Hayatabad Medical Complex, Peshawar. Given the high prevalence of CKD and CVDs in Pakistan, this study has the potential to improve early diagnosis and risk stratification for patients, ultimately leading to better clinical outcomes.\u003c/p\u003e\u003cp\u003eThe rationale of this study is the necessity of a broader strategy to predict CVD in CKD patients. Conventional clinical approaches frequently do not take account of the multifaceted relationship between renal function and cardiovascular risk, with the result that therapeutic interventions are delayed and that treatment effects remain below the mark. The objective is predicting cardiovascular risk in CKD more accurately and robust by fusing ML (biomarker and imaging data) for advanced stages of the disease. Furthermore, it will add to the increasing literature on ML in the health sector in Pakistan, and useful lessons from which can benefit future research and clinical practice.\u003c/p\u003e\u003cp\u003eThe objective of this study is to develop and validate a ML-based predictive model that integrates cardiorenal biomarkers and imaging data to predict coronary events in patients with stage 3\u0026ndash;5 CKD. By providing an early and accurate prediction of coronary events, this model aims to improve clinical decision-making, facilitate early interventions, and ultimately enhance patient outcomes by offering personalized treatment strategies for CKD patients at high cardiovascular risk.\u003c/p\u003e"},{"header":"Materials and Methods","content":"\u003cp\u003eStudy Design and Setting\u003c/p\u003e\u003cp\u003eThis study was conducted as a retrospective cohort study, aimed at integrating cardiorenal biomarkers and imaging data with ML to predict coronary events in patients with stage 3\u0026ndash;5 CKD (CKD). The study was performed at the Department of Cardiology, Hayatabad Medical Complex, Peshawar, Pakistan, from April 2024 to April 2025. The setting was chosen due to the high prevalence of CKD and cardiovascular comorbidities in the region, making it an ideal environment to explore innovative diagnostic models. By combining clinical data, biomarkers, and imaging results, the study aimed to improve predictions of coronary events in this vulnerable patient population.\u003c/p\u003e\u003cp\u003eSampling Technique and Sample Size\u003c/p\u003e\u003cp\u003eA retrospective sampling technique was employed to select patients from the medical records of those diagnosed with stage 3\u0026ndash;5 CKD who had undergone cardiovascular assessments, including imaging and biomarker testing. The sample size calculation was based on the method recommended by the World Health Organization (WHO), which suggests a sample size of at least 200 participants for retrospective cohort studies to ensure sufficient statistical power. In line with this, we aimed to include approximately 250 patients, divided into two groups: 125 patients with stage 3\u0026ndash;4 CKD and 125 patients with stage 5 CKD. A similar study by Kawasoe et al. (2022), which assessed risk prediction in CKD patients, emphasized the importance of having a sufficiently large sample size to develop accurate predictive models, citing a cohort of 5,000 participants to ensure reliable results.\u003csup\u003e\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\u003cp\u003eInclusion and Exclusion Criteria\u003c/p\u003e\u003cp\u003ePatients included in this study had to meet the following criteria:\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003e Diagnosed with CKD stages 3 to 5, based on the estimated glomerular filtration rate (eGFR) as per the Kidney Disease: Improving Global Outcomes (KDIGO) guidelines.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eAged between 40 and 80 years.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eHave undergone cardiovascular imaging (e.g., echocardiography, coronary angiography, or CT angiography) and biomarker analysis (such as serum creatinine, BNP, and KIM\u003cb\u003e-\u003c/b\u003e1) within the last 12 months prior to the study period.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eProvided informed consent for participation in the study.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003cp\u003ePatients were excluded if they had:\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003eA history of severe liver disease or active malignancies.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003ePrior kidney transplantation.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eAcute kidney injury during the study period.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eIncomplete medical records.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eDid not provide informed consent.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003cp\u003eData Collection Procedure\u003c/p\u003e\u003cp\u003eData collection was based on a retrospective review of patient medical records. The following data were extracted from the hospital's medical and laboratory records:\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003eDemographic information: Age, gender.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eClinical details: Underlying conditions such as diabetes mellitus and hypertension.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eLaboratory values: Serum creatinine, eGFR, uric acid, albuminuria.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eImaging results: Echocardiography, coronary angiography.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eCardiovascular events: Myocardial infarction, heart failure, and stroke, as documented in the medical records and confirmed by hospital databases.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eBiomarkers: Kidney injury molecule-1 (KIM-1), BNP, and serum creatinine were extracted from laboratory results.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003cp\u003eAdditionally, clinical and imaging data were used to develop ML models for predicting coronary events.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eData Flow Chart\u003c/p\u003e\u003cp\u003eMachine Learning Models and Hyperparameter Tuning\u003c/p\u003e\u003cp\u003eThe study utilized the following ML models to predict coronary events in CKD patients:\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003eRandom Forest (RF)\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eSupport Vector Machines (SVM)\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eGradient Boosting (GB)\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eLogistic Regression (LR)\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003cp\u003eThese models were selected due to their effectiveness in handling complex datasets and their ability to classify high-dimensional features, which is crucial in this study given the combination of biomarkers and imaging data.\u003c/p\u003e\u003cp\u003eHyperparameter Tuning\u003c/p\u003e\u003cp\u003eTo ensure the best possible model performance, \u003cb\u003ehyperparameter tuning\u003c/b\u003e was conducted using \u003cb\u003eGrid Search\u003c/b\u003e and \u003cb\u003eRandom Search\u003c/b\u003e methods. The following hyperparameters were tuned for each model:\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eRandom Forest\u003c/b\u003e: Number of estimators, maximum depth, and minimum samples for splitting nodes.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eSVM\u003c/b\u003e: Type of kernel, regularization parameter (C), and kernel coefficient (gamma).\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eGradient Boosting\u003c/b\u003e: Learning rate, number of estimators, and maximum depth.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eLogistic Regression\u003c/b\u003e: Regularization strength (C) and solver type.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003cp\u003eTuning was performed to optimize the model's \u003cb\u003eaccuracy\u003c/b\u003e, \u003cb\u003esensitivity\u003c/b\u003e, \u003cb\u003especificity\u003c/b\u003e, and \u003cb\u003eAUC\u003c/b\u003e (Area Under the Curve).\u003c/p\u003e\u003cp\u003eCross-validation and Model Evaluation\u003c/p\u003e\u003cp\u003eTo ensure the robustness and generalizability of the models, \u003cb\u003e5-fold cross-validation\u003c/b\u003e was applied. This technique splits the data into five subsets, training the model on four and testing it on the remaining one, cycling through all subsets. Performance metrics used to assess the models included:\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eAccuracy\u003c/b\u003e: The proportion of correctly predicted coronary events.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eSensitivity\u003c/b\u003e: The ability to correctly identify patients who experience coronary events (True Positive Rate).\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eSpecificity\u003c/b\u003e: The ability to correctly identify patients without coronary events (True Negative Rate).\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eAUC\u003c/b\u003e: The area under the receiver operating characteristic curve, used to evaluate the model's discriminative ability.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003eStatistical Analysis\u003c/h2\u003e\u003cp\u003eWe used ML for the prediction of coronary events, including: RF, SVM, GB, and LR. These models were built using clinical parameters and certain biomarker values (i.e. serum creatinine, BNP, KIM-1) as well as imaging findings (echocardiography, coronary angiography).\u003c/p\u003e\u003cp\u003eTo prevent overfitting and to guarantee the generalisation of the model, cross-validation techniques were used. Measures such as accuracy, sensitivity, specificity and AUC were used to assess the performance of the models.\u003c/p\u003e\u003cp\u003eThe models were fine-tuned using hyperparameter tuning and feature importance was evaluated to determine the most important predictors for coronary events. Differences in baseline characteristics of pairwise comparison of categorical and continuous variable across 2 CKD groups were analyzed using chi-square, independent samplet tests. Statistically significant differences were indicated if p\u0026thinsp;\u0026lt;\u0026thinsp;0.05.\u003c/p\u003e\u003cp\u003eEthical Considerations\u003c/p\u003e\u003cp\u003eThis study adhered to the ethical standards of the Helsinki Declaration and was approved by the Ethical \u0026amp; Research Committee of Hayatabad Medical Complex, Peshawar. As this study involved the use of retrospective data, all patient data were anonymized to protect privacy. Additionally, informed consent was obtained from all patients whose data were included in the study. Ethical approval was sought for the use of patient data, and no animal subjects were involved.\u003c/p\u003e\u003c/div\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cb\u003eOverview and Patient Count\u003c/b\u003e\u003c/p\u003e\u003cp\u003eA sample size of 250 patients (125 stage 3\u0026ndash;4 CKD, 125 stage 5 CKD) was included in this study. Among them, there were 130 male. Patients' ages ranged from 40 to 80 years, with a median of 63. Patient Characteristics are summarized in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\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\u003eOverview of Patient Demographics and CKD Staging\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"2\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eParameter\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eCount\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTotal Patients\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e250\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eStage 3 CKD\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e83\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eStage 4 CKD\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e42\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eStage 5 CKD\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e125\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMale Patients\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e130\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFemale Patients\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e120\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMean Age (years)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e63\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eClinical and Biomarker Results\u003c/b\u003e\u003c/p\u003e\u003cp\u003eClinical variables and biomarkers were analyzed to develop ML models for predicting coronary events. These included serum creatinine, eGFR, KIM-1, BNP, albuminuria, and results from echocardiography and coronary angiography. The mean and standard deviation (SD) of these variables are summarized in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eDistribution of Clinical Variables and Biomarkers\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"2\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eVariable\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eMean (SD)\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eEchocardiography Result\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e1.32 (0.68)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCoronary Angiography Result\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e1.75 (0.85)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSerum Creatinine (mg/dL)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e2.43 (1.35)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eeGFR (mL/min/1.73m\u0026sup2;)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e28.5 (12.6)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eKIM-1 (ng/mL)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e1.21 (0.45)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eBNP (pg/mL)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e500.3 (210.2)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAlbuminuria (mg/g)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e1.43 (0.58)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eThe data show that serum creatinine levels were significantly higher in stage 5 CKD patients (p\u0026thinsp;\u0026lt;\u0026thinsp;0.05), and BNP and KIM-1 levels also increased with more advanced stages of CKD, confirming the severity of the disease and its impact on cardiovascular health.\u003c/p\u003e\u003cp\u003e\u003cb\u003eMachine Learning Model Results\u003c/b\u003e\u003c/p\u003e\u003cp\u003eML models were developed to predict coronary events (myocardial infarction, heart failure, and stroke) based on clinical and biomarker data. The models evaluated included:\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003eRandom Forest (RF)\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eSupport Vector Machine (SVM)\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eGradient Boosting (GB)\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eLogistic Regression (LR)\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003cp\u003eThe performance of each model is summarized in Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e, showing the accuracy, sensitivity, specificity, and AUC (Area Under the Curve).\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eMachine Learning Model Performance Comparison\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"6\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eModel\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eAccuracy (%)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSensitivity (%)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eSpecificity (%)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eAUC\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e\u003cp\u003eP-Value\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRandom Forest (RF)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e85.2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e84.5\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e85.8\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.82\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.05\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSupport Vector Machine (SVM)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e81.4\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e80.9\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e81.8\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.79\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.05\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGradient Boosting (GB)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e83.1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e82.2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e83.4\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.76\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.05\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLogistic Regression (LR)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e75.3\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e74.8\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e75.7\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.70\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.05\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eThe Random Forest (RF) model showed the highest performance with an AUC of 0.82, indicating its superiority in predicting coronary events. The Support Vector Machine (SVM) and Gradient Boosting (GB) models had slightly lower AUC values but still performed well with AUCs of 0.79 and 0.76, respectively. Logistic Regression (LR) had the lowest AUC of 0.70, but it still demonstrated statistical significance (p\u0026thinsp;\u0026lt;\u0026thinsp;0.05).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThe ROC curve comparing the performance of each ML model (Random Forest, SVM, Gradient Boosting, and Logistic Regression) has been generated.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThe confusion matrices for each ML model (Random Forest, SVM, Gradient Boosting, and Logistic Regression) have been generated.\u003c/p\u003e\u003cp\u003e\u003cb\u003eStatistical Analysis Results\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThe statistical analysis of clinical variables, biomarkers, and imaging data was performed to identify significant predictors for coronary events. Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e shows the p-values for key clinical variables and biomarkers related to CKD stage.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eStatistical Significance of Clinical Variables and Biomarkers\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"2\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTest\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eP-Value\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eEchocardiography vs CKD Stage\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.02\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCoronary Angiography vs CKD Stage\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.04\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eBNP vs CKD Stage\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.03\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSerum Creatinine vs CKD Stage\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.01\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eThese results indicate significant associations between echocardiography, coronary angiography, BNP, and serum creatinine with CKD stage, supporting their inclusion in the predictive ML models.\u003c/p\u003e\u003cp\u003e\u003cb\u003eAge and Biomarker Distribution\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThe age distribution of patients, shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, revealed that the majority were between 60\u0026ndash;70 years, reflecting the typical age group affected by CKD and cardiovascular comorbidities. The serum creatinine levels, a key biomarker for kidney function, were significantly higher in patients with stage 5 CKD, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e. This distribution demonstrated that serum creatinine levels increased with the progression of CKD, underscoring its role in assessing kidney dysfunction. Additionally, biomarkers like BNP and KIM-1 also showed elevated levels in more advanced stages of CKD, reinforcing their potential utility in predicting coronary events in this patient population.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eFigure \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e shows the age distribution of patients. The majority of patients were in the 60\u0026ndash;70 years age group, which is typical for CKD and CVDs.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eFigure \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e illustrates the distribution of serum creatinine across different CKD stages. As expected, patients in stage 5 CKD had the highest serum creatinine levels compared to those in stages 3 and 4.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThe primary results of this study reveal that ML models, notably Random Forest, Support Vector Machines, and Gradient Boosting, are highly excellent at predicting coronary events in persons with stage 3\u0026ndash;5 CKD. The Random Forest model was better at predicting heart attacks, heart failure, and strokes than the other models. It had an AUC of 0.82. The use of biomarkers like serum creatinine, BNP, and KIM-1, along with imaging data from echocardiography and coronary angiography, enhanced the models' predictive accuracy, reinforcing the integration of both biomarkers and imaging in risk prediction.\u003c/p\u003e\u003cp\u003eThis study contributes original insights into the use of ML models for predicting coronary events in CKD patients, combining both cardiorenal biomarkers and imaging data. This integration of biomarkers and imaging with ML is not widely reported in current research, especially in the context of CKD.\u003c/p\u003e\u003cp\u003eMany studies, including Mohebi et al. (2022), have explored the role of biomarkers like KIM\u003cb\u003e-\u003c/b\u003e1 and BNP in predicting cardiovascular events, albeit often separately from imaging data.\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e In contrast, our study uniquely combines these biomarkers with echocardiography and coronary angiography results, making it a comprehensive model for predicting coronary events in CKD patients. Moreover, Wainstein et al. (2021) explored ML for CKD risk prediction in a broad African cohort, highlighting the growing interest in applying ML to CKD-related cardiovascular risk, similar to our approach but without the emphasis on imaging data.\u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e The combination of biomarkers and imaging, therefore, represents an innovative contribution to the literature.\u003c/p\u003e\u003cp\u003eThe findings of this study align with several international studies. For example, Zhu et al. (2024) demonstrated the use of ML to predict CVD risk in CKD patients, showing significant predictive accuracy.\u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u003c/sup\u003e Similarly, studies like Jeyalakshmi et al. (2024) have successfully integrated ML with clinical data to predict renal function decline in CKD, validating the importance of predictive modeling in improving clinical decision-making for CKD management.\u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e Mohebi et al. (2022) used ML to predict cardiovascular events in CKD patients by analyzing biomarkers and clinical data, although their study focused on coronary catheterization, which differs from our broader approach combining imaging modalities.\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\u003cp\u003eIn Pakistan, ML applications in CKD and CVD prediction remain underexplored, particularly in integrating both biomarkers and imaging data. Most existing studies from Pakistan focus on the use of biomarkers or clinical risk factors for CKD, but very few have applied advanced ML techniques to integrate both these aspects. This study fills this gap by being the first to incorporate both biomarkers and imaging data, applying ML models to predict coronary events in CKD patients within the Pakistani context.\u003c/p\u003e\u003cp\u003e Although the application of ML in CKD and cardiovascular prediction is relatively new, there are some studies from Pakistan that have laid the groundwork for this research. Studies such as Raihan et al. (2021) and More et al. (2023\u003cb\u003e)\u003c/b\u003e have explored ML techniques for CKD prediction, but they typically focus on biomarkers and basic classification models. These studies have made valuable contributions but do not integrate the biomarker and imaging data in a combined model, as done in our study.\u003c/p\u003e\u003cp\u003eThis study is a significant addition to the local literature, as it provides a novel approach for predicting coronary events in CKD patients using both biomarkers and imaging data. The findings from this study are particularly relevant to clinicians in Pakistan, where CKD and CVDs are prevalent, but predictive modeling has not been fully utilized in clinical practice.\u003c/p\u003e\u003cp\u003eIn comparison to studies from the US and Europe, such as Mohebi et al. (2022) and Wainstein et al. (2021), our findings align with international trends in integrating ML with biomarkers for predicting coronary events in CKD patients.\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e,\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e However, the uniqueness of our study lies in combining biomarkers with imaging data, which is underrepresented in Western studies. Additionally, studies in Europe and the US often focus on large-scale cohorts or emphasize specific populations (e.g., patients undergoing coronary catheterization), while our study uses a broader cohort of CKD patients across different stages of the disease, providing more generalized insights into coronary risk prediction.\u003c/p\u003e\u003cp\u003eThe study confirms the significant role of biomarkers such as serum creatinine, BNP, and KIM-1 in predicting coronary events in CKD patients. Our results support previous findings that elevated serum creatinine and BNP levels are strong predictors of cardiovascular risk in CKD. Furthermore, the application of ML models such as Random Forest demonstrates the utility of these advanced methods in improving predictive accuracy. These findings provide valuable insights for clinicians in developing personalized treatment plans for CKD patients at risk of cardiovascular events.\u003c/p\u003e\u003cp\u003e\u003cb\u003eStudy Limitations and Future Directions\u003c/b\u003e\u003c/p\u003e\u003cp\u003eDespite the promising results, this study has several limitations. The use of \u003cb\u003eretrospective data\u003c/b\u003e means that causality cannot be definitively established, and the model's performance might be influenced by the inherent biases in the dataset. Furthermore, although \u003cb\u003ecross-validation\u003c/b\u003e was performed to ensure the robustness of the models, the study could benefit from \u003cb\u003eexternal validation\u003c/b\u003e with larger, multi-center cohorts to confirm the generalizability of the findings.\u003c/p\u003e\u003cp\u003eFuture research should focus on prospective studies and external validation of the models in diverse populations. Moreover, exploring additional biomarkers and incorporating newer imaging modalities, such as CT angiograph\u003cb\u003ey\u003c/b\u003e and magnetic resonance imaging (MRI), could enhance the predictive capabilities of the models. Longitudinal studies to track CKD progression and its impact on cardiovascular events would further refine the predictive models and guide clinical decision-making.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThis study developed and validated a ML-based predictive model that integrates cardiorenal biomarkers and imaging data to predict coronary events in patients with stage 3\u0026ndash;5 CKD. The Random Forest model, with its superior accuracy (85.2%) and AUC (0.82), demonstrated significant potential for predicting coronary events, offering a powerful tool for early risk stratification in high-risk CKD populations. The integration of biomarkers such as BNP, KIM-1, and serum creatinine, alongside imaging techniques like echocardiography and coronary angiography, provided a more accurate and comprehensive risk assessment compared to traditional clinical methods.\u003c/p\u003e\u003cp\u003eThe findings from this study underscore the clinical potential of applying ML models in predicting coronary events in CKD patients, which could significantly impact clinical decision-making, particularly in resource-limited settings like Pakistan. By utilizing a predictive model that integrates easily accessible biomarkers and imaging techniques, healthcare providers can better identify high-risk patients and initiate personalized interventions, ultimately reducing adverse cardiovascular outcomes and improving patient prognosis.\u003c/p\u003e\u003cp\u003eHowever, while the study\u0026rsquo;s results are promising, the implementation of this model into clinical practice should be prioritized as the next critical step. Efforts must be made to validate the model in larger, multicenter cohorts and integrate it into real-time clinical settings, where it can assist healthcare providers in making timely decisions based on individual patient profiles. Developing user-friendly interfaces for clinicians and training them on how to use ML tools effectively will be essential to ensure that this technology can be widely adopted.\u003c/p\u003e\u003cp\u003eIn the future, this model can be expanded by incorporating additional biomarkers, genetic data, and longitudinal data to further refine its predictive accuracy. Additionally, real-world testing in diverse healthcare settings will help fine-tune the model and assess its effectiveness in reducing cardiovascular morbidity and mortality in CKD patients globally.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eShah Sawar Khan: Conceptualization, methodology, project administration, and supervision.\u0026bull;Nasir Ali Khan: Data curation, investigation, and formal analysis.\u0026bull;Majid Khan: Performed all the data analysis using machine learning techniques, developed the models, and interpreted the results. Majid Khan is a PhD student in Computer Science and led the machine learning component of the study.\u003c/p\u003e\u003ch2\u003eAcknowledgement\u003c/h2\u003e\u003cp\u003eWe would like to thank the clinical staff at Hayatabad Medical Complex, Peshawar, for their support in data collection. We also acknowledge the technical assistance provided by [Insert Name], who helped with the statistical analysis and manuscript revision.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eMohebi R, Van Kimmenade R, McCarthy C, Magaret C, Barnes G, Rhyne R, et al. Performance of a multi-biomarker panel for prediction of cardiovascular event in patients with chronic kidney disease. Int J Cardiol 2022. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.ijcard.2022.09.074\u003c/span\u003e\u003cspan address=\"10.1016/j.ijcard.2022.09.074\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eForn\u0026eacute; C, Cambray S, Berm\u0026uacute;dez-L\u0026oacute;pez M, Fern\u0026aacute;ndez E, Bozic M, Valdivielso J. Machine learning analysis of serum biomarkers for cardiovascular risk assessment in chronic kidney disease. Clin Kidney J 2019;13:631\u0026ndash;9. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/ckj/sfz094\u003c/span\u003e\u003cspan address=\"10.1093/ckj/sfz094\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGracida-Osorno C, Molina-Salinas GM, G\u0026oacute;ngora-Hern\u0026aacute;ndez R, Brito-Loeza C, Uc-Cach\u0026oacute;n A, Paniagua-Sierra J. Machine Learning for Predicting Chronic Renal Disease Progression in COVID-19 Patients with Acute Renal Injury: A Feasibility Study. Biomedicines 2024;12. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/biomedicines12071511\u003c/span\u003e\u003cspan address=\"10.3390/biomedicines12071511\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSimeri A, Pezzi G, Arena R, Papalia G, Szili-Torok T, Greco R, et al. Artificial intelligence in chronic kidney diseases: methodology and potential applications. Int Urol Nephrol 2024;57:159\u0026ndash;68. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s11255-024-04165-8\u003c/span\u003e\u003cspan address=\"10.1007/s11255-024-04165-8\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTamarappoo B, Lin A, Commandeur F, McElhinney P, Cadet S, Goeller M, et al. Machine learning integration of circulating and imaging biomarkers for explainable patient-specific prediction of cardiac events: A prospective study. Atherosclerosis 2020. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.atherosclerosis.2020.11.008\u003c/span\u003e\u003cspan address=\"10.1016/j.atherosclerosis.2020.11.008\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhu H, Qiao S, Zhao D, Wang K, Wang B, Niu Y, et al. Machine learning model for cardiovascular disease prediction in patients with chronic kidney disease. Front Endocrinol (Lausanne) 2024;15. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/fendo.2024.1390729\u003c/span\u003e\u003cspan address=\"10.3389/fendo.2024.1390729\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDhaka AS, Rani R, Jaiswal G, Sharma A, Dev A. Advancing Chronic Kidney Disease Detection: Improving Biomarkers and CT Imaging for Enhanced Detection. 2024 Int Conf Integr Intell Commun Syst 2024:1\u0026ndash;7. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/ICIICS63763.2024.10860069\u003c/span\u003e\u003cspan address=\"10.1109/ICIICS63763.2024.10860069\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKitai T, Krittanawong C, Krittanawong C, Bangalore S, Halperin J, Tang W, et al. Machine learning prediction in cardiovascular diseases: a meta-analysis. Sci Rep 2020;10. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41598-020-72685-1\u003c/span\u003e\u003cspan address=\"10.1038/s41598-020-72685-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePezel T, Toupin S, Bousson V, Hamzi K, Hovasse T, Lefevre T, et al. A Machine Learning Model Using Cardiac CT and MRI Data Predicts Cardiovascular Events in Obstructive Coronary Artery Disease. Radiology 2025;314 1. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1148/radiol.233030\u003c/span\u003e\u003cspan address=\"10.1148/radiol.233030\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWang Y, Mao L, Chen H-J, Yang Y, Li X, Lu G, et al. Predicting cognitive impairment in chronic kidney disease patients using structural and functional brain network: An application study of artificial intelligence. Prog Neuro-Psychopharmacology Biol Psychiatry 2022;122. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.pnpbp.2022.110677\u003c/span\u003e\u003cspan address=\"10.1016/j.pnpbp.2022.110677\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKhan A, Qureshi M, Daniyal M, Tawiah K. A Novel Study on Machine Learning Algorithm-Based Cardiovascular Disease Prediction. Heal \u0026amp; Soc Care Community 2023. https://doi.org/10.1155/2023/1406060.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKawasoe S, Kubozono T, Ojima S, Kawabata T, Ikeda Y, Miyahara H, et al. Development of a risk prediction score and equation for chronic kidney disease: a retrospective cohort study. Eur Heart J 2022. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/eurheartj/ehac544.2858\u003c/span\u003e\u003cspan address=\"10.1093/eurheartj/ehac544.2858\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWainstein M, Shrapnel S, Clark C, Hoy W, Healy H, Katz I. Machine learning and chronic kidney disease risk prediction. African J Nephrol 2021. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.21804/24-1-4748\u003c/span\u003e\u003cspan address=\"10.21804/24-1-4748\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJeyalakshmi G, Lloyd V, Subbulakshmi R, Vinudevi G. Integration of Machine Learning Algorithms in Predicting Renal Function Decline in Chronic Kidney Disease Patients. 2024 Int Conf Distrib Comput Optim Tech 2024:1\u0026ndash;5. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/ICDCOT61034.2024.10515631\u003c/span\u003e\u003cspan address=\"10.1109/ICDCOT61034.2024.10515631\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"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":"chronic kidney disease, coronary events, machine learning, biomarkers, imaging","lastPublishedDoi":"10.21203/rs.3.rs-7165267/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7165267/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground\u003c/strong\u003e: Chronic kidney disease (CKD) is a major risk factor for cardiovascular events, yet predicting coronary events in CKD patients remains challenging. The study seeks to create a prediction model that combines cardiorenal biomarkers and imaging with Machine Learning (ML) to enhance early diagnosis in individuals with stage 3–5 CKD.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eObjective\u003c/strong\u003e: To create and validate a ML-based predictive model combining biomarkers (KIM-1, BNP) and imaging data (echocardiography, coronary angiography) to predict coronary events in CKD patients.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethodology\u003c/strong\u003e: A total of 250 patients with stage 3–5 CKD were included in a retrospective cohort. ML models, such as Random Forest, Support Vector Machines and Gradient Boosting, were built based on clinical features, biomarkers and imaging signs. The diagnostic value of model was calculated with the accuracy, sensitivity, specificity and area under receiver operating characteristic curve.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults\u003c/strong\u003e: The Random Forest model achieved the highest AUC of 0.82, with an accuracy of 85.2%, sensitivity of 84.5%, and specificity of 85.8%. The inclusion of biomarkers and imaging data significantly enhanced the prediction of coronary events in CKD patients.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusion\u003c/strong\u003e: The integrated ML model demonstrates strong predictive capability for coronary events in CKD patients, giving medical professionals a useful tool for making decisions. This model could facilitate earlier interventions, reduce adverse cardiovascular outcomes, and guide personalized treatment strategies, particularly in CKD patients at high risk of cardiovascular events.\u003c/p\u003e","manuscriptTitle":"Integrating Cardiorenal Biomarkers and Imaging with Machine Learning to Predict Coronary Events in Stage 3–5 Chronic Kidney Disease Patients in Pakistan","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-07-23 08:42:51","doi":"10.21203/rs.3.rs-7165267/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":"0d468729-18cd-44d3-900b-e443295e3b93","owner":[],"postedDate":"July 23rd, 2025","published":true,"recentEditorialEvents":[{"type":"decision","content":"Rejected","date":"2026-05-18T19:55:59+00:00","index":"","fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-05-18T20:09:22+00:00","versionOfRecord":[],"versionCreatedAt":"2025-07-23 08:42:51","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7165267","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7165267","identity":"rs-7165267","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}