Enhancing CVD Risk Prediction: Integrating ECG Signals with Conventional Models Using AI

Enhancing CVD Risk Prediction: Integrating ECG Signals with Conventional Models Using AI | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Enhancing CVD Risk Prediction: Integrating ECG Signals with Conventional Models Using AI Maryam Mahdavi, Anoshirvan Kazemnejad, Abbas Asosheh, Davood Khalili, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6542485/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 07 Nov, 2025 Read the published version in Scientific Reports → Version 1 posted 14 You are reading this latest preprint version Abstract Introduction: Non-communicable diseases (NCDs), particularly cardiovascular diseases (CVDs), have become the leading cause of mortality worldwide, with Iran exhibiting higher-than-average incidence and mortality rates. Early detection of high-risk individuals is critical, as CVD often progresses silently. Electrocardiogram (ECG) signals, when integrated with machine learning (ML), may enhance risk prediction beyond traditional models. Objective This study aimed to evaluate the predictive performance of ECG signal features for incident CVD using machine learning models in a large population-based cohort from the Tehran Lipid and Glucose Study (TLGS). Methods A total of 4,637 adults aged 40–79 years without prior CVD at baseline (2006–2008) were followed up until 2018. Baseline characteristics, laboratory measurements, and ECG signal features were collected. CVD events were defined as coronary heart disease (CHD) or stroke. A Weibull regression model assessed the association between ECG features and incident CVD, with model performance evaluated using Harrell’s C-index, Net Reclassification Index (NRI), and Integrated Discrimination Improvement (IDI). Results Over a 10-year follow-up, 483 participants (10.4%) developed CVD. The addition of ECG signal features improved risk prediction in women, increasing the Harrell’s C-index from 0.84 to 0.85 and demonstrating significant reclassification improvement (NRI: 55.7%, IDI: 2.8%). However, no meaningful improvement was observed in men. ECG-based modeling outperformed traditional risk scores, particularly for intermediate-risk categories among women. Conclusion Incorporating ECG signal features into ML-based risk models significantly enhanced CVD prediction performance in women, suggesting potential utility for improving individualized preventive strategies. Further research is warranted to refine ECG-based risk stratification tools for broader clinical application. Health sciences/Cardiology/Cardiovascular biology/Cardiovascular diseases Health sciences/Biomarkers/Predictive markers ECG signal NRI and IDI CVD prediction model Background Non-communicable diseases (NCDs) have emerged as a major global public health concern, accounting for nearly 60% of total annual mortality worldwide ( 1 ). Among these, cardiovascular diseases (CVDs) represent a leading contributor, affecting a substantial portion of the global population ( 2 ). In Iran, cardiovascular diseases (CVDs) are the leading cause of death ( 3 ) with age-standardized incidence and mortality rates, as well as disability-adjusted life years (DALYs) attributed to CVDs, surpassing global averages ( 4 ). The increasing prevalence of sedentary lifestyles, rising obesity rates, and declining physical activity levels among Iranians are expected to exacerbate the burden of CVDs in the coming years ( 5 ). Demographic transitions, including an aging population, are likely to further elevate the incidence of these conditions. CVDs are among the primary causes of morbidity and mortality on a global scale. In 2015, they were responsible for approximately 17.9 million deaths and 347.5 million DALYs ( 6 , 7 ). Although mortality rates from CVDs are declining in high-income countries, a significant proportion of deaths still occur in low- and middle-income countries, particularly in the Eastern Mediterranean region ( 7 ). In Iran, there has been a notable shift in mortality patterns from infectious diseases to NCDs over recent decades, with CVDs being highly prevalent ( 8 ). Iran's high prevalence of CVDs presents significant healthcare and economic challenges ( 9 ). Cardiovascular complications such as stroke, myocardial infarction (MI), and coronary artery disease (CAD) can result in both fatal and non-fatal outcomes. While advancements in treatment have contributed to reduced mortality, many individuals continue to suffer from long-term complications, including psychological distress, fatigue, sleep disturbances, dyspnea, and reduced quality of life ( 10 ). In severe cases, CVDs may progress to heart failure, recurrent MI, or sudden cardiac death ( 11 ). Cardiovascular disease (CVD), a multifactorial and chronic condition stemming from various heart and vascular system disorders, remains the primary cause of premature mortality and long-term disability globally. Although pharmacological and surgical interventions are routinely employed to manage CVD, they do not offer a permanent cure and often have a considerable impact on patients' quality of life. Consequently, contemporary approaches to CVD management place significant emphasis on prevention. Emerging evidence suggests that up to 80% of premature deaths attributed to CVD may be preventable through timely interventions ( 12 ). Since CVD typically progresses slowly and may remain asymptomatic for extended periods, it is frequently diagnosed in its advanced stages, when treatment becomes more complex. Thus, early detection of individuals at high risk is essential for implementing effective preventive strategies ( 13 ). In line with this, recent clinical guidelines increasingly advocate for the use of CVD risk prediction models to identify individuals who may benefit from early, targeted preventive measures. Notably, the majority of these models have been developed based on conventional statistical methodologies ( 14 ). The artificial intelligence (AI) methodology involves replicating and emulating human cognitive functions, particularly in learning, reasoning, and problem-solving. AI encompasses a range of specialized fields, such as machine learning, evolutionary computing, natural language processing, robotics, and other related disciplines. In the context of medical research, machine learning (ML) is a widely adopted technique for automating data analysis through modeling. ML algorithms are designed to learn from datasets, identify patterns, and make informed decisions. ML is particularly effective at handling and interpreting complex data types, such as those found in biomedical and healthcare sectors, where traditional statistical methods may not be sufficient ( 15 ). Advancements in computer science, coupled with the demand for precision medicine, have led to an increase in multidimensional data from various sources. This necessitates the development of advanced tools and models capable of processing, understanding, and analyzing this vast and intricate data. These tools must also accurately forecast outcomes and predict risks. The most effective predictive model, which delivers optimal performance, is determined by several key factors. These include the specific objectives and purposes for which the models are developed, their ability to generalize across different datasets, their robustness in handling diverse conditions, and their capacity to produce consistent and reproducible results when applied in real clinical settings ( 16 ). In this study, we aimed to quantify the association between ECG signals and incident CVD in a population-based cohort called the Tehran and Lipid Glucose Study (TLGS) during more than a decade of follow-up. This study explores whether integrating ECG signal data into machine learning can improve cardiovascular disease prediction. Materials and Methods Participants The Tehran Lipid and Glucose Study (TLGS), launched in 1999, is a population-based prospective cohort study conducted in Tehran's District 13, aimed at examining risk factors associated with non-communicable diseases (NCDs). The study design has been described in previous publications (17). In brief, the first phase (1999–2001) was cross-sectional, enrolling 15,005 individuals aged ≥3 years through a multistage random sampling method. The study continued as a longitudinal follow-up, with data from 8,071 participants aged 40 to 79 years in the third phase used for this analysis. The sixth phase (2015–2018) follow-up included cardiovascular disease (CVD) events such as coronary heart disease (CHD) or stroke, along with follow-up duration. The study was conducted in multiple phases: Phase one (1999–2001) and phase two (2002–2005) followed a multistage, random cluster sampling method. Follow-ups occurred at approximately 3.5-year intervals, continuing through phases three to seven (2006–2020), with an average 73% participation rate per phase. For the current analysis, 5,479 adults aged 40–79 years who participated in the third examination cycle were initially considered. We entered the participants in phase III of TLGS for data analysis because ECG signals were gathered in this phase onwards. After excluding individuals based on specific criteria, history of CVD (n=772), lost to follow-up for CVD events (n=15), and missing ECG data (n=2207). The final study population included 4,637 adults without any established CVD (CHD or stroke). Participants had no previous record of coronary artery disease (including angina, myocardial infarction, coronary artery bypass grafting, or percutaneous coronary intervention), cerebrovascular conditions (such as stroke or transient ischemic attack), or peripheral arterial disease (such as claudication) at the third examination cycle, which served as the baseline, with follow-up extending until March 2018. This large-scale, population-based study continues to be a valuable resource for monitoring risk factors associated with chronic diseases and cardiovascular conditions. Measurements Eligible participants underwent initial interviews to collect socio-demographic and medical data. All measurements were conducted by trained staff following standardized study protocols. Further details on the measurement methods can be found in previous studies [19]. Participants were seated, and their systolic blood pressure (SBP) and diastolic blood pressure (DBP) were measured twice by a general physician after they had rested for 15 minutes. A standard mercury sphygmomanometer was used consistently to measure blood pressure, with the first reading determining the maximum inflation level and the average of two readings recorded. For laboratory assessments, participants were directed to fast for 12 to 14 hours before blood sampling. Samples were immediately transported to the TLGS laboratory for analysis using Selectra autoanalyzers (Vital Scientific, Spankeren, Netherlands). Fasting plasma glucose (FPG) was measured using an enzymatic colorimetric method based on glucose oxidation. Lipid profile assessments were performed using Pars Azmoon (Tehran, Iran) commercial kits. Total cholesterol (TC) was measured through enzymatic colorimetric tests utilizing cholesterol esterase, cholesterol oxidase, and glycerol phosphate oxidase. High-density lipoprotein cholesterol (HDL-C) was measured using phosphotungstic acid precipitation. Laboratory tests were conducted only when internal quality control values fell within the acceptable range. For more information on the measurement methods, please refer to other studies (18). Definitions of outcome Participants were followed up annually for cardiovascular disease (CVD) events by trained nurses through telephone interviews. If an event was reported, a trained physician collected additional data through home visits or hospital record reviews. The final diagnosis was confirmed by the Cohort Outcome Panel (19). Coronary heart disease (CHD) was defined as the occurrence of myocardial infarction (MI), probable MI, unstable angina pectoris, or angiography-confirmed coronary artery disease (CAD). Stroke was classified as definite or possible stroke according to the World Health Organization (WHO) definition (20). In this study, cardiovascular disease (CVD) was characterized as the incidence of either stroke or coronary heart disease (CHD), with the time to the initial occurrence of either event recorded as the time-to-event. ECG signal The electrocardiogram (ECG) records variations in electrical potentials across chest surface electrodes, reflecting cardiac activity. Each heartbeat appears as a series of deviations from the ECG baseline, corresponding to the heart's electrical activity that drives muscle contraction (21). Key ECG components include the P wave, QRS complex, and T wave. Features were extracted for 12 leads in ECG signals across multiple domains, including time (22), frequency (22), time–frequency (23), peak-R and morphological distances (24), and Heart Rate Variability (HRV) (25, 26) (Table 1). Analysis Method Normally distributed continuous variables were presented as mean and standard deviation (SD), and categorical baseline characteristics were described as frequency (%). The independent T-test was applied to analyze continuous variables, while the chi-squared test was used for categorical variables. These tests were used to compare the baseline characteristics of study participants between those with and without CVD. The event date was defined as the date of the CVD incident. Those who met the following criteria were censored: leaving the residential area, deaths not related to CVD, loss of follow-up, or end of follow-up. The univariate survival regression model explored the relationship between potential features and CVD incidence. The univariate model’s significance threshold for feature selection was set at an alpha level of 0.2. These features were then entered into the forward stepwise Cox regression model, and important features for CVD incidence were selected in both men and women. We used the Weibull regression model for the analysis of CVD outcome. The Baseline Weibull regression model was developed based on the traditional risk markers (age, smoking, systolic blood pressure, use of anti-hypertensive drugs, total and HDL cholesterol, smoking, and diabetes). An improved Weibull regression model was developed by adding the features of the ECG signal to the baseline and basic Weibull models. In survival analysis, Harrell’s C statistic is used to measure discrimination performance. The model's performance was evaluated on the data set using Harrell's concordance index (C-index), specifically focusing on its ability to rank subjects by risk. It calculated the probability that for each randomly selected tie of two people who did and did not develop the event of interest, a person who developed an event of interest, at a certain specific time has a higher risk score than a randomly selected person who did not develop an event during the same, specific follow-up interval (27). The 95% confidence intervals for Harrell’s C statistics of different models were estimated with bootstrap resampling. The discrimination of the model was determined by comparing the Net Reclassification Index (cNRI), and Integrated Discrimination Improvement Index (IDI) between models. Absolute and relative IDI and cut-point-based and cut-point-free NRI were used as measures of predictive ability added to the baseline survival-based regression model by paraclinical parameters (28). Cut-points for NRI were considered as low risk: <5%, borderline risk: 5%-7.5%, intermediate risk: 7.5%-20%, and high risk: ≥20% based on the 10-year ASCVD-PCE score classification. All analyses were performed using Python, with a two-tailed p-value < 0.05 deemed statistically significant. Result A total of 2094 men and 2543 women were followed up. During a 10-year follow-up, out of 4637 non-CVD participants in the TLGS, 483 (10.4%) developed CVD. Table 2 presents the descriptive statistics of the study participants, with analyses conducted separately for male and female subjects. A comparison of the baseline characteristics between participants developing CVD and those not developing CVD is illustrated in Table 2. At baseline, subjects with CVD had higher age, systolic and diastolic blood pressure, FPG and 2hPG levels, cholesterol, and LDL in both men and women than subjects without CVD. Mean values of the age (57.76 vs. 47 years) were significantly higher in participants developing CVD than those not developing CVD in men. Mean values of the age (58.67 vs. 45.69 years) were significantly higher in participants developing CVD than those not developing CVD in women. No difference was observed between participants developing CVD and those not developing CVD in the mean HDL in men. There was no significant difference in smokers between participants developing CVD and those not developing CVD, in both men and women. In this study, specific electrocardiogram (ECG) features were selected for analysis, with different sets chosen for male and female subjects. For male subjects, three features were selected: the maximum amplitude in lead aVR, which is the highest amplitude recorded in that lead; the root mean square in lead I, a measure of the signal magnitude calculated as the square root of the mean of squared values; and the minimum amplitude in lead V6, the lowest amplitude recorded in that lead. For female subjects, a broader set of thirteen features was utilized. These include the spectral centroid from the Fast Fourier Transform (FFT) in lead aVF, representing the center of mass of the frequency spectrum; kurtosis in lead aVR, a statistical measure of the signal distribution; the mean of wavelet coefficients at level 1 in lead aVR, which is the average of the wavelet decomposition coefficients at the first level; the spectral centroid from FFT in lead I; the standard deviation in lead II, indicating the variability of the signal amplitude; the dominant frequency from FFT in lead II, identifying the most prominent frequency component; the minimum amplitude in lead V3; the standard deviation of the PR interval in lead V3, reflecting the variability of the time from the P wave's start to the QRS complex's start; the spectral centroid from FFT in lead V4; the maximum R-peak amplitude in lead V4, the highest amplitude of the QRS complex's peak; kurtosis in lead V5; the mean QT interval in lead V5, the average duration from the Q wave's start to the T wave's end; and the energy of wavelet coefficients at level 1 in lead V6, representing the energy content of the wavelet decomposition coefficients at the first level. Harrell’s C index of discrimination can provide helpful information on the predictive performance of a predictive model. As shown in Table 3, Harrell’s C index for models with and without ECG signal features in men were 0.77 (CI: 0.75–0.79) and 0.77 (95% CI: 0.74–0.79), respectively. There was a slight difference in the goodness of fit as indicated by AIC between these two risk algorithms (AIC: 3896 vs. 3899) in men. Also, the Harrell’s C index for models with and without ECG signal features in women was 0.85 (CI: 0.83–0.88) and 0.84 (95% CI: 0.81–0.86), respectively. There was a significant difference in the goodness of fit as indicated by AIC between these two risk algorithms (AIC: 2375 vs. 2395) in women. Examining the clinical relevance of a new risk biomarker corresponds to studying the predictive power of a currently available predictive model augmented by new biomarker(s). Generally, the addition of ECG signal features to the Framingham model in women significantly improved risk classification as indicated by cut point-free NRI of 55.7% (95% CIs: 46.5–65.0%), absolute IDI of 2.8% (95% CIs: 1.0–4.6%) but the addition of ECG signal features to the Framingham model in men significantly didn't improve risk classification. Table 4 presents the improvement in the reclassification of individuals between risk categories after complementing FRS with ECG signal features. In each of the four FRS categories (i.e., 0–4.9%, 5–7.4%, 7.5–19.9%, and ≥20%), 18.5%, 9.1%, 10.3%, and -16.1% of women were correctly reclassified, respectively. In each of the four FRS categories (i.e., 0–4.9%, 5–7.4%, 7.5–19.9%, and ≥20%), 0.0%, 5.9%, 1.7%, and -3.0% of men were correctly reclassified, respectively. Discussion This study represents the first evaluation of electrocardiogram (ECG) signal features within the Tehran Lipid and Glucose Study (TLGS) for predicting cardiovascular disease (CVD) risk. Our findings demonstrate that ECG signal features significantly enhance the predictive capacity for CVD in women, both statistically and clinically. Prior research by Mahdavi et al. evaluated the American Heart Association (AHA) risk score classification in TLGS participants, noting its effectiveness in identifying high-risk individuals but limited ability to accurately distinguish those with severe cardiovascular outcomes ( 29 ). This suggests a need for refined risk stratification approaches to improve predictive accuracy. Khalili et al. previously highlighted the predictive utility of abnormal resting ECGs compared to Rose Questionnaire angina in estimating 10-year coronary heart disease (CHD) risk in an urban Iranian population. Their study categorized participants into four groups based on Rose Angina and ECG ischemia status, finding that adding abnormal ECG findings to angina did not significantly increase CHD event risk prediction. However, their analysis relied on the Minnesota Coding (MC) system, a standardized ECG classification method widely used in epidemiological studies, rather than raw ECG signal data ( 30 ). The MC system, introduced in 1960 and expanded in 1983 to include serial comparisons, involves complex measurement protocols that make visual coding time-consuming and prone to errors. In contrast, the current study leverages ECG signal feature extraction to minimize measurement and coding errors, offering a more robust approach to risk prediction ( 26 ). Most population-based studies employ the MC system for ECG coding, which can be performed manually or through automated methods. Both approaches, however, are susceptible to errors, and neither manual coding by a single individual nor automated techniques can be considered fully reliable ( 27 ). Hadaegh et al. demonstrated the added value of ECG abnormalities beyond the Framingham Risk Score (FRS) for CHD risk stratification in Middle Eastern women. Their findings indicated that incorporating ECG abnormalities—specifically ST depression or T-wave changes—into the FRS did not significantly improve C-statistics but enhanced predictive performance by 20.8% (95% CI 5.0–38.9) using the cut-point-free Net Reclassification Improvement (NRI). Notably, among women, ECG abnormalities were independently associated with increased CHD risk only in the intermediate risk group. Like Khalili et al., their study utilized MC for ECG analysis ( 28 ). The reliance on manual ECG pattern recognition and MC in population-based and clinical studies is increasingly being questioned, as these methods are labor-intensive and error-prone. Automated ECG signal processing, as applied in the current study, offers a promising alternative. By extracting signal features directly from ECG data, this approach reduces errors associated with visual coding and enhances the precision of CVD risk prediction. This study underscores the potential of ECG signal analysis to refine risk stratification and improve outcomes in epidemiological research, particularly for women in intermediate risk groups. Conclusions In this study, the Minnesota Coding (MC) system and select ECG features were not employed; instead, comprehensive ECG signal data were analyzed. This approach significantly enhanced the statistical and clinical prediction of cardiovascular disease (CVD) risk in women, but no such effect was observed in men. Declarations This study was performed according to the ethical principles of the Helsinki Declaration, and all procedures involving human subjects were approved by the National Research Council of the Islamic Republic of Iran (IR.SBMU.ENDOCRINE.REC.1398.103), the Human Research Review Committee of the Endocrine Research Center, Shahid Beheshti University, Tehran, Iran. Written informed consent was obtained from all subjects. Ethics approval and consent to participate: The data for this research were obtained from the Tehran Lipid and Glucose Study (TLGS), conducted by the Endocrine Research Center at Shahid Beheshti University of Medical Sciences. The authors would like to express their gratitude to everyone involved in the design and data collection of the TLGS, as well as to the study participants. This project was approved by the Ethics Committee of Tarbiat Modares University under the code IR.MODARES.REC.1403.100 Consent for publication: Not applicable. Availability of data and materials: The datasets used and analyzed in the current study are available from the corresponding author upon reasonable request. Competing interest: The authors declare that they have no competing interests, whether financial or non-financial, that could influence the integrity of this research. Funding: None Author's contributions: 'MM': Data collection, literature review, and manuscript preparation. 'DK': Study design, revising the manuscript, and final approval of the manuscript. 'MM', 'AT', and 'KH': Data analysis and data interpretation. 'AA' and 'AK': Study design, manuscript preparation, revising the manuscript, and the final approval of the manuscript. All authors reviewed and approved the final draft. Acknowledgment: The data for this research were obtained from the Tehran Lipid and Glucose Study (TLGS), conducted by the Endocrine Research Center at Shahid Beheshti University of Medical Sciences. The authors would like to express their gratitude to everyone involved in the design and data collection of the TLGS, as well as to the study participants. This project was approved by the Ethics Committee of Tarbiat Modares University under the code IR.MODARES.REC.1403.100. References Taheri Soodejani M. Non-communicable diseases in the world over the past century: a secondary data analysis. Frontiers in Public Health. 2024;12:1436236. Roth GA, Mensah GA, Johnson CO, Addolorato G, Ammirati E, Baddour LM, et al. Global burden of cardiovascular diseases and risk factors, 1990–2019: update from the GBD 2019 study. Journal of the American college of cardiology. 2020;76(25):2982-3021. Aminorroaya A, Yoosefi M, Rezaei N, Shabani M, Mohammadi E, Fattahi N, et al. Global, regional, and national quality of care of ischaemic heart disease from 1990 to 2017: a systematic analysis for the Global Burden of Disease Study 2017. European journal of preventive cardiology. 2022;29(2):371-9. Organization WH. Noncommunicable diseases country profiles 2018. 2018. Rahmani A, Sayehmiri K, Asadollahi K, Sarokhani D, Islami F, Sarokhani M. Investigation of the prevalence of obesity in Iran: a systematic review and meta-analysis study. Acta Medica Iranica. 2015:596-607. Uthman OA. Global, regional, and national disability‐adjusted life years (DALYs) for 315 diseases and injuries and healthy life expectancy (HALE) for 195 countries and territories, 1990‐2015: a systematic analysis for the Global Burden of Diseases, Injuries, and Risk Factors (GBD) 2015 Study. Lancet. 2016;388(10053):1603-58. Roth GA, Johnson C, Abajobir A, Abd-Allah F, Abera SF, Abyu G, et al. Global, regional, and national burden of cardiovascular diseases for 10 causes, 1990 to 2015. Journal of the American college of cardiology. 2017;70(1):1-25. Danaei G, Farzadfar F, Kelishadi R, Rashidian A, Rouhani OM, Ahmadnia S, et al. Iran in transition. The Lancet. 2019;393(10184):1984-2005. Mensah GA, Brown DW. An overview of cardiovascular disease burden in the United States. Health affairs. 2007;26(1):38-48. R Jabir N, Nasir Siddiqui A, Kandy Firoz C, Md Ashraf G, Kashif Zaidi S, Shahnawaz Khan M, et al. Current updates on therapeutic advances in the management of cardiovascular diseases. Current pharmaceutical design. 2016;22(5):566-71. Guidry UC, Evans JC, Larson MG, Wilson PW, Murabito JM, Levy D. Temporal trends in event rates after Q-wave myocardial infarction: the Framingham Heart Study. Circulation. 1999;100(20):2054-9. Piepoli MF, Hoes AW, Agewall S, Albus C, Brotons C, Catapano AL, et al. Guidelines: Editor's choice: 2016 European Guidelines on cardiovascular disease prevention in clinical practice: The Sixth Joint Task Force of the European Society of Cardiology and Other Societies on Cardiovascular Disease Prevention in Clinical Practice (constituted by representatives of 10 societies and by invited experts) Developed with the special contribution of the European Association for Cardiovascular Prevention & Rehabilitation (EACPR). European heart journal. 2016;37(29):2315. Liu S, Li Y, Zeng X, Wang H, Yin P, Wang L, et al. Burden of cardiovascular diseases in China, 1990-2016: findings from the 2016 global burden of disease study. JAMA cardiology. 2019;4(4):342-52. Goff Jr DC, Lloyd-Jones DM, Bennett G, Coady S, D’agostino RB, Gibbons R, et al. 2013 ACC/AHA guideline on the assessment of cardiovascular risk: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. Circulation. 2014;129(25_suppl_2):S49-S73. Mansoor H, Elgendy IY, Segal R, Bavry AA, Bian J. Risk prediction model for in-hospital mortality in women with ST-elevation myocardial infarction: a machine learning approach. Heart & Lung. 2017;46(6):405-11. Faizal ASM, Thevarajah TM, Khor SM, Chang S-W. A review of risk prediction models in cardiovascular disease: conventional approach vs. artificial intelligent approach. Computer methods and programs in biomedicine. 2021;207:106190. Azizi F, Madjid M, Rahmani M, Emami H, Mirmiran P, Hadjipour R. Tehran Lipid and Glucose Study (TLGS): rationale and design. Iranian journal of endocrinology and metabolism. 2000;2(2):77-86. Azizi F, Ghanbarian A, Momenan AA, Hadaegh F, Mirmiran P, Hedayati M, et al. Prevention of non-communicable disease in a population in nutrition transition: Tehran Lipid and Glucose Study phase II. Trials. 2009;10:1-15. Saatchi M, Mansournia MA, Khalili D, Daroudi R, Yazdani K. Estimation of generalized impact fraction and population attributable fraction of hypertension based on JNC-IV and 2017 ACC/AHA guidelines for cardiovascular diseases using parametric G-formula: Tehran Lipid and Glucose Study (TLGS). Risk Management and Healthcare Policy. 2020:1015-28. Organization WH. Cerebrovascular disorders: a clinical and research classification. Cerebrovascular disorders: a clinical and research classification1978. Afsar FA, Arif M, Yang J. Detection of ST segment deviation episodes in ECG using KLT with an ensemble neural classifier. Physiological measurement. 2008;29(7):747. Singh AK, Krishnan S. ECG signal feature extraction trends in methods and applications. BioMedical Engineering OnLine. 2023;22(1):22. Addison PS. Wavelet transforms and the ECG: a review. Physiological measurement. 2005;26(5):R155. Srinivasulu A, Sriraam N. Signal processing framework for the detection of ventricular ectopic beat episodes. Journal of Medical Signals & Sensors. 2023;13(3):239-51. Electrophysiology TFotESoCtNASoP. Heart rate variability: standards of measurement, physiological interpretation, and clinical use. Circulation. 1996;93(5):1043-65. Shaffer F, Ginsberg JP. An overview of heart rate variability metrics and norms. Frontiers in public health. 2017;5:258. Harrell Jr FE, Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Statistics in medicine. 1996;15(4):361-87. Pencina MJ, D'Agostino Sr RB, D'Agostino Jr RB, Vasan RS. Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Statistics in medicine. 2008;27(2):157-72. Mahdavi M, Kazemnejad A, Asosheh A, Khalili D. Cardiovascular risk patterns through AI-enhanced clustering of longitudinal health data. Journal of Diabetes & Metabolic Disorders. 2025;24(1):1-10. Prineas RJ, Crow RS, Zhang Z-M. The Minnesota code manual of electrocardiographic findings: Springer Science & Business Media; 2009. Tables Table 1: ECG Signal Analysis Features Features Domain-frequency These features use the Fast Fourier Transform (FFT) to analyze the signal in the frequency domain, including: The amplitude of the ECG signal over time The average and midpoint of the signal values The amount of signal dispersion around the mean The minimum and maximum signal amplitudes and the distance between them. The amount of asymmetry in the distribution of amplitudes The peak intensity of the distribution of the number of outliers The root mean square value describes the signal's overall power and indicates the signal's overall strength. Domain-time A general statistical representation of the amplitude of the ECG signal over time, including: The total energy of the signal across all frequencies Identifying the frequency that has the most energy The center of gravity is the frequency spectrum associated with the heart rate. It determines whether the energy is concentrated at low or high frequencies. These features are useful for detecting arrhythmias or heart rhythm abnormalities. Wavelet Wavelet decomposition decomposes the ECG signal into different levels (up to level 3) to extract time-frequency information from the signal, including: The energy of the wavelets at each level, which is useful for identifying sharp changes such as the QRS complex. Average wavelet coefficients at each level, which are very useful for identifying local changes in the signal structure and play an important role in detecting ectopic beats or ischemia. Peak-R and morphological distances These features are extracted based on R-peaks (peak points of the heart rate), including: The maximum amplitude of one of the peak-R waves The average duration of the QRS complex indicates the duration of ventricular activation. The time interval from the beginning of the P wave to the peak of the R wave The time interval from the beginning of the Q wave to the end of the T wave is very important for diagnosing electrical disorders of the heart. The fluctuation of each of these intervals Heart Rate Variability (HRV) HRV features measure the time variation between successive R-peaks and are indicative of autonomic nervous system activity including: The average of the R-R intervals, which is the inverse of the heart rate. The fluctuation of these intervals. The root mean square of the differences between consecutive R-R intervals and is suitable for assessing the variability of the heart rate over short intervals. The percentage of R-R intervals whose difference is greater than 50 milliseconds and indicates the activity of the parasympathetic branch of the nervous system. Table 2: Baseline characteristics of Participants Women Men Variables Non-CVD (n=2353) CVD (n=190) P-value Non-CVD (n=1801) CVD (n=293) P-value Age (year) 45.96±11.26 58.67±9.94 <0.001 47.00±12.66 57.76±11.69 <0.001 SBP (mmHg) 112.59±17.79 130.90±20.95 <0.001 117.77±16.84 127.81±20.06 <0.001 DBP (mmHg) 73.09±10.18 78.53±11.61 <0.001 76.05±10.11 79.15±11.49 <0.001 TC (mg/dl) 196.94±39.41 218.26±41.85 <0.001 190.92±36.46 200.95±38.05 <0.001 HDL-C (mg/dl) 45.00±10.52 42.03±9.49 <0.001 37.64±8.53 37.94±8.50 0.578 LDL-C (mg/dl) 121.69±32.83 136.87±35.63 <0.001 119.67±31.55 127.35±34.40 <0.001 FPG (mg/dl) 95.84±28.89 121.58±52.27 <0.001 96.95±28.31 107.72±41.08 <0.001 2-h PG (mg/dl) 110.92±44.60 134.98±56.52 <0.001 107.98±52.80 120.41±58.55 0.001 DM 260(11.3) 67(37.2) <0.001 165(9.5) 57(20.2) <0.001 DM medication 139(5.9) 47(24.7) <0.001 72(4.0) 29(9.9) <0.001 HTN 335(14.5) 83(45.1) <0.001 285(16.1) 92(31.9) <0.001 HTN medication 111(4.7) 29(15.3) <0.001 46(2.6) 20(6.8) <0.001 Current Smoker 65(2.8) 6(3.3) 0.708 449(25.4) 76(26.4) 0.732 SBP, systolic blood pressure; DBP, diastolic blood pressure; TC, Total Cholesterol; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; FPG, fasting plasma glucose; 2-hPG, 2-hour post-challenge plasma glucose; DM, diabetes melitus; HTN, Hypertension. Data are given as the mean±SD for continuous variables and data are given as the n(%) for categorical variables Table3: Predictive performances of the basic Framingham’s “general CVD risk” algorithm vs. enhanced model Basic model enhanced model Women Harrell’s C index (95% CIs) 0.84(0.81,0.86) 0.85(0.83,0.88) Akaike information criterion 2395 2375 Added predictive values Absolute IDI (95% CIs) 0.0280(0.0100,0.0460) P-value=0.002 Relative IDI (95% CIs) 0.2139(-0.0421,0.4699) P-value=0.102 Cutpoint-based NRI (95% CIs) 0.1541(0.0416,0.2665) P-value=0.007 Cutpoint-free NRI (95% CIs) 0.5575(0.4653,0.6496) P-value<0.001 Men Harrell’s C index (95% CIs) 0.77(0.74,0.79) 0.77(0.75,0.79) Akaike information criterion 3899 3896 Added predictive values Absolute IDI (95% CIs) 0.0030(-0.0022,0.0081) P-value=0.261 Relative IDI (95% CIs) 0.0242(-0.0501,0.0986) P-value=0.523 Cutpoint-based NRI (95% CIs) 0.0154(-0.0601,0.0908) P-value=0.690 Cutpoint-free NRI (95% CIs) 0.1138(-0.0322,0.2598) P-value=0.127 Akaike information criterion (AIC) was used as a measure of model fit. The lower is the AIC the better will be the model fitness. Difference in AIC > 10 was considered significant Table4: Reclassification table comparing risk strata for models incorporating CVD risk factors with and without ECG signal features Model with ECG signal features Reclassified Net correctly reclassified % <%5 %5 - %7.5 %7.5 - %20 ≥%20 Increased risk Decreased risk Women Event 0 - 5% 22 2 3 0 5 0 18.5 5 - 7.5% 4 2 5 0 5 4 9.1 7.5 - 20% 4 9 44 21 21 13 10.3 ≥ 20% 0 0 9 47 0 9 -16.1 Non-evnet 0 - 5% 1464 70 16 1 87 0 5.6 5 - 7.5% 82 79 53 1 54 82 -13.0 7.5 - 20% 34 78 210 31 31 112 -22.9 ≥ 20% 1 1 46 90 0 48 -34.8 Men Event 0 - 5% 7 0 0 0 0 0 0.0 5 - 7.5% 2 12 3 0 3 2 5.9 7.5 - 20% 0 3 110 5 5 3 1.7 ≥ 20% 0 0 4 130 0 4 -3.0 Non-evnet 0 - 5% 402 41 0 0 41 0 9.3 5 - 7.5% 57 283 41 0 41 57 -4.2 7.5 - 20% 1 41 530 21 21 42 -3.5 ≥ 20% 0 0 22 267 0 22 -7.6 ACC/AHA, American College of Cardiology/American Heart Association Additional Declarations No competing interests reported. Cite Share Download PDF Status: Published Journal Publication published 07 Nov, 2025 Read the published version in Scientific Reports → Version 1 posted Editorial decision: Revision requested 03 Sep, 2025 Reviewers agreed at journal 03 Aug, 2025 Reviews received at journal 25 Jul, 2025 Reviews received at journal 20 Jul, 2025 Reviewers agreed at journal 14 Jul, 2025 Reviews received at journal 13 Jul, 2025 Reviewers agreed at journal 10 Jul, 2025 Reviewers agreed at journal 09 Jul, 2025 Reviewers agreed at journal 29 Jun, 2025 Reviewers invited by journal 23 Jun, 2025 Editor assigned by journal 09 Jun, 2025 Editor invited by journal 11 May, 2025 Submission checks completed at journal 08 May, 2025 First submitted to journal 27 Apr, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6542485","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":475587388,"identity":"0a665831-2801-48ef-aa60-55f2b43f1cbd","order_by":0,"name":"Maryam Mahdavi","email":"","orcid":"","institution":"Tarbiat Modares University","correspondingAuthor":false,"prefix":"","firstName":"Maryam","middleName":"","lastName":"Mahdavi","suffix":""},{"id":475587389,"identity":"ec2f1cac-5227-499c-b258-2b438ee53e99","order_by":1,"name":"Anoshirvan Kazemnejad","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA7UlEQVRIiWNgGAWjYBACCRiDH5nDkECMFskGkrUYHEDWgg9Ith9/+OgGg12e8bXDB2/83GNjz8B++AHDwz24tUjzJCQb5zAkF5vdTku27HmWltjAk2bAkPAMtxY5hoRj0jkMzInbbueYSfAcOAz0RQ7QLwfwaOF/2P47h6E+cfPsHDPJPwf+2zPwv8GvRVoimY05h+Fw4gbpHDNpngMHGBskCNgiOeMZM9BhxxNnAP1iLXMgObFN4pnBAXxaJM6nP/ycw1Cd2D87+eDNNwfs7Pn5kx8+/IFHCxgw/kPisAExIQ2jYBSMglEwCggAAN46TksG6DLfAAAAAElFTkSuQmCC","orcid":"","institution":"Tarbiat Modares University","correspondingAuthor":true,"prefix":"","firstName":"Anoshirvan","middleName":"","lastName":"Kazemnejad","suffix":""},{"id":475587390,"identity":"bae81584-4e3a-4e1f-bda6-192087941031","order_by":2,"name":"Abbas Asosheh","email":"","orcid":"","institution":"Tarbiat Modares University","correspondingAuthor":false,"prefix":"","firstName":"Abbas","middleName":"","lastName":"Asosheh","suffix":""},{"id":475587391,"identity":"50f0d35a-2c69-4814-b2be-cc8a4117bd08","order_by":3,"name":"Davood Khalili","email":"","orcid":"","institution":"Shahid Beheshti University of Medical Sciences","correspondingAuthor":false,"prefix":"","firstName":"Davood","middleName":"","lastName":"Khalili","suffix":""},{"id":475587392,"identity":"7488601f-3701-4f62-9ffd-bf62549c0eee","order_by":4,"name":"Kamyab Hosseinpour","email":"","orcid":"","institution":"Sharif University of Technology","correspondingAuthor":false,"prefix":"","firstName":"Kamyab","middleName":"","lastName":"Hosseinpour","suffix":""},{"id":475587395,"identity":"08ebb4f3-c18f-4809-a852-ce222f6cff3b","order_by":5,"name":"Ahmadreza Tajari","email":"","orcid":"","institution":"Institute for Biological Information Processes","correspondingAuthor":false,"prefix":"","firstName":"Ahmadreza","middleName":"","lastName":"Tajari","suffix":""}],"badges":[],"createdAt":"2025-04-27 21:38:11","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6542485/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6542485/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41598-025-26471-6","type":"published","date":"2025-11-07T15:57:20+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":95564021,"identity":"6ea96d75-bf48-49c4-a987-33831b6894d8","added_by":"auto","created_at":"2025-11-10 16:06:34","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":677257,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6542485/v1/36b7fd59-8308-4f50-83e8-85b9566dc11c.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Enhancing CVD Risk Prediction: Integrating ECG Signals with Conventional Models Using AI","fulltext":[{"header":"Background","content":"\u003cp\u003eNon-communicable diseases (NCDs) have emerged as a major global public health concern, accounting for nearly 60% of total annual mortality worldwide (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e). Among these, cardiovascular diseases (CVDs) represent a leading contributor, affecting a substantial portion of the global population (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e). In Iran, cardiovascular diseases (CVDs) are the leading cause of death (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e) with age-standardized incidence and mortality rates, as well as disability-adjusted life years (DALYs) attributed to CVDs, surpassing global averages (\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e). The increasing prevalence of sedentary lifestyles, rising obesity rates, and declining physical activity levels among Iranians are expected to exacerbate the burden of CVDs in the coming years (\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e). Demographic transitions, including an aging population, are likely to further elevate the incidence of these conditions.\u003c/p\u003e \u003cp\u003eCVDs are among the primary causes of morbidity and mortality on a global scale. In 2015, they were responsible for approximately 17.9\u0026nbsp;million deaths and 347.5\u0026nbsp;million DALYs (\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e). Although mortality rates from CVDs are declining in high-income countries, a significant proportion of deaths still occur in low- and middle-income countries, particularly in the Eastern Mediterranean region (\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e). In Iran, there has been a notable shift in mortality patterns from infectious diseases to NCDs over recent decades, with CVDs being highly prevalent (\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e). Iran's high prevalence of CVDs presents significant healthcare and economic challenges (\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e). Cardiovascular complications such as stroke, myocardial infarction (MI), and coronary artery disease (CAD) can result in both fatal and non-fatal outcomes. While advancements in treatment have contributed to reduced mortality, many individuals continue to suffer from long-term complications, including psychological distress, fatigue, sleep disturbances, dyspnea, and reduced quality of life (\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e). In severe cases, CVDs may progress to heart failure, recurrent MI, or sudden cardiac death (\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eCardiovascular disease (CVD), a multifactorial and chronic condition stemming from various heart and vascular system disorders, remains the primary cause of premature mortality and long-term disability globally. Although pharmacological and surgical interventions are routinely employed to manage CVD, they do not offer a permanent cure and often have a considerable impact on patients' quality of life. Consequently, contemporary approaches to CVD management place significant emphasis on prevention. Emerging evidence suggests that up to 80% of premature deaths attributed to CVD may be preventable through timely interventions (\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e). Since CVD typically progresses slowly and may remain asymptomatic for extended periods, it is frequently diagnosed in its advanced stages, when treatment becomes more complex. Thus, early detection of individuals at high risk is essential for implementing effective preventive strategies (\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e). In line with this, recent clinical guidelines increasingly advocate for the use of CVD risk prediction models to identify individuals who may benefit from early, targeted preventive measures. Notably, the majority of these models have been developed based on conventional statistical methodologies (\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe artificial intelligence (AI) methodology involves replicating and emulating human cognitive functions, particularly in learning, reasoning, and problem-solving. AI encompasses a range of specialized fields, such as machine learning, evolutionary computing, natural language processing, robotics, and other related disciplines. In the context of medical research, machine learning (ML) is a widely adopted technique for automating data analysis through modeling. ML algorithms are designed to learn from datasets, identify patterns, and make informed decisions. ML is particularly effective at handling and interpreting complex data types, such as those found in biomedical and healthcare sectors, where traditional statistical methods may not be sufficient (\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e). Advancements in computer science, coupled with the demand for precision medicine, have led to an increase in multidimensional data from various sources. This necessitates the development of advanced tools and models capable of processing, understanding, and analyzing this vast and intricate data. These tools must also accurately forecast outcomes and predict risks. The most effective predictive model, which delivers optimal performance, is determined by several key factors. These include the specific objectives and purposes for which the models are developed, their ability to generalize across different datasets, their robustness in handling diverse conditions, and their capacity to produce consistent and reproducible results when applied in real clinical settings (\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e). In this study, we aimed to quantify the association between ECG signals and incident CVD in a population-based cohort called the Tehran and Lipid Glucose Study (TLGS) during more than a decade of follow-up. This study explores whether integrating ECG signal data into machine learning can improve cardiovascular disease prediction.\u003c/p\u003e"},{"header":"Materials and Methods","content":"\u003cp\u003e\u003cem\u003eParticipants\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe Tehran Lipid and Glucose Study (TLGS), launched in 1999, is a population-based prospective cohort study conducted in Tehran\u0026apos;s District 13, aimed at examining risk factors associated with non-communicable diseases (NCDs). The study design has been described in previous publications (17). In brief, the first phase (1999\u0026ndash;2001) was cross-sectional, enrolling 15,005 individuals aged \u0026ge;3 years through a multistage random sampling method. The study continued as a longitudinal follow-up, with data from 8,071 participants aged 40 to 79 years in the third phase used for this analysis. The sixth phase (2015\u0026ndash;2018) follow-up included cardiovascular disease (CVD) events such as coronary heart disease (CHD) or stroke, along with follow-up duration. The study was conducted in multiple phases:\u003c/p\u003e\n\u003cul type=\"disc\"\u003e\n \u003cli\u003ePhase one (1999\u0026ndash;2001) and phase two (2002\u0026ndash;2005) followed a multistage, random cluster sampling method.\u003c/li\u003e\n \u003cli\u003eFollow-ups occurred at approximately 3.5-year intervals, continuing through phases three to seven (2006\u0026ndash;2020), with an average 73% participation rate per phase.\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eFor the current analysis, 5,479 adults aged 40\u0026ndash;79 years who participated in the third examination cycle were initially considered. We entered the participants in phase III of TLGS for data analysis because ECG signals were gathered in this phase onwards. After excluding individuals based on specific criteria, history of CVD (n=772), lost to follow-up for CVD events (n=15), and missing\u0026nbsp;ECG\u0026nbsp;data (n=2207). The final study population included 4,637 adults without any established CVD (CHD or stroke). Participants had no previous record of coronary artery disease (including angina, myocardial infarction, coronary artery bypass grafting, or percutaneous coronary intervention), cerebrovascular conditions (such as stroke or transient ischemic attack), or peripheral arterial disease (such as claudication) at the third examination cycle, which served as the baseline, with follow-up extending until March 2018. This large-scale, population-based study continues to be a valuable resource for monitoring risk factors associated with chronic diseases and cardiovascular conditions.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eMeasurements\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eEligible participants underwent initial interviews to collect socio-demographic and medical data. All measurements were conducted by trained staff following standardized study protocols. Further details on the measurement methods can be found in previous studies [19]. Participants were seated, and their systolic blood pressure (SBP) and diastolic blood pressure (DBP) were measured twice by a general physician after they had rested for 15 minutes. A standard mercury sphygmomanometer was used consistently to measure blood pressure, with the first reading determining the maximum inflation level and the average of two readings recorded. For laboratory assessments, participants were directed to fast for 12 to 14 hours before blood sampling. Samples were immediately transported to the TLGS laboratory for analysis using Selectra autoanalyzers (Vital Scientific, Spankeren, Netherlands). Fasting plasma glucose (FPG) was measured using an enzymatic colorimetric method based on glucose oxidation. Lipid profile assessments were performed using Pars Azmoon (Tehran, Iran) commercial kits. Total cholesterol (TC) was measured through enzymatic colorimetric tests utilizing cholesterol esterase, cholesterol oxidase, and glycerol phosphate oxidase. High-density lipoprotein cholesterol (HDL-C) was measured using phosphotungstic acid precipitation. Laboratory tests were conducted only when internal quality control values fell within the acceptable range. For more information on the measurement methods, please refer to other studies (18).\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eDefinitions\u003c/em\u003e of outcome\u003c/p\u003e\n\u003cp\u003eParticipants were followed up annually for cardiovascular disease (CVD) events by trained nurses through telephone interviews. If an event was reported, a trained physician collected additional data through home visits or hospital record reviews. The final diagnosis was confirmed by the Cohort Outcome Panel (19). Coronary heart disease (CHD) was defined as the occurrence of myocardial infarction (MI), probable MI, unstable angina pectoris, or angiography-confirmed coronary artery disease (CAD). Stroke was classified as definite or possible stroke according to the World Health Organization (WHO) definition (20). In this study, cardiovascular disease (CVD) was characterized as the incidence of either stroke or coronary heart disease (CHD), with the time to the initial occurrence of either event recorded as the time-to-event.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eECG signal\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe electrocardiogram (ECG) records variations in electrical potentials across chest surface electrodes, reflecting cardiac activity. Each heartbeat appears as a series of deviations from the ECG baseline, corresponding to the heart\u0026apos;s electrical activity that drives muscle contraction (21). Key ECG components include the P wave, QRS complex, and T wave. Features were extracted for 12 leads in ECG signals across multiple domains, including time (22), frequency (22), time\u0026ndash;frequency (23), peak-R and morphological distances (24), and Heart Rate Variability (HRV)\u0026nbsp;(25, 26)\u0026nbsp;(Table\u0026nbsp;1).\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eAnalysis Method\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eNormally distributed continuous variables were presented as mean and standard deviation (SD), and categorical baseline characteristics were described as frequency (%).\u0026nbsp;The independent T-test was applied to analyze continuous variables, while the chi-squared test was used for categorical variables. These tests were used to compare the baseline characteristics of study participants between those with and without CVD. The event date was defined as the date of the CVD incident. Those who met the following criteria were censored: leaving the residential area, deaths not related to CVD, loss of follow-up, or end of follow-up.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe univariate survival regression model explored the relationship between potential features and CVD incidence. The univariate model\u0026rsquo;s significance threshold for feature selection was set at an alpha level of 0.2. These features were then entered into the forward stepwise Cox regression model, and important features for CVD incidence were selected in both men and women.\u003c/p\u003e\n\u003cp\u003eWe used the Weibull regression model for the analysis of CVD outcome. The Baseline Weibull regression model was developed based on the traditional risk markers (age, smoking, systolic blood pressure, use of anti-hypertensive drugs, total and HDL cholesterol, smoking, and diabetes). An improved Weibull regression model was developed by adding the features of the ECG signal to the baseline and basic Weibull models. In survival analysis, Harrell\u0026rsquo;s C statistic is used to measure discrimination performance. The model\u0026apos;s performance was evaluated on the data set using Harrell\u0026apos;s concordance index (C-index), specifically focusing on its ability to rank subjects by risk. It calculated the probability that for each randomly selected tie of two people who did and did not develop the event of interest, a person who developed an event of interest, at a certain specific time has a higher risk score than a randomly selected person who did not develop an event during the same, specific follow-up interval (27). The 95% confidence intervals for Harrell\u0026rsquo;s \u003cem\u003eC\u0026nbsp;\u003c/em\u003estatistics of different models were estimated with bootstrap resampling.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe discrimination of the model was determined by comparing the Net Reclassification Index (cNRI), and Integrated Discrimination Improvement Index (IDI) between models. Absolute and relative IDI and cut-point-based and cut-point-free NRI were used as measures of predictive ability added to the baseline survival-based regression model by paraclinical parameters (28). Cut-points for NRI were considered as low risk: \u0026lt;5%, borderline risk: 5%-7.5%, intermediate risk: 7.5%-20%, and high risk: \u0026ge;20% based on the 10-year ASCVD-PCE score classification. All analyses were performed using Python, with a two-tailed p-value \u0026lt; 0.05 deemed statistically significant.\u003c/p\u003e"},{"header":"Result","content":"\u003cp\u003eA total of 2094 men and 2543 women were followed up. During a 10-year follow-up, out of 4637 non-CVD participants in the TLGS, 483 (10.4%) developed CVD. Table 2 presents the descriptive statistics of the study participants, with analyses conducted separately for male and female subjects. A comparison of the baseline characteristics between participants developing CVD and those not developing CVD is illustrated in Table 2. At baseline, subjects with CVD had higher age, systolic and diastolic blood pressure, FPG and 2hPG levels, cholesterol, and LDL in both men and women than subjects without CVD. Mean values of the age (57.76 vs. 47 years) were significantly higher in participants developing CVD than those not developing CVD in men. Mean values of the age (58.67 vs. 45.69 years) were significantly higher in participants developing CVD than those not developing CVD in women. No difference was observed between participants developing CVD and those not developing CVD in the mean HDL in men. There was no significant difference in smokers between participants developing CVD and those not developing CVD, in both men and women.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIn this study, specific electrocardiogram (ECG) features were selected for analysis, with different sets chosen for male and female subjects. For male subjects, three features were selected: the maximum amplitude in lead aVR, which is the highest amplitude recorded in that lead; the root mean square in lead I, a measure of the signal magnitude calculated as the square root of the mean of squared values; and the minimum amplitude in lead V6, the lowest amplitude recorded in that lead. For female subjects, a broader set of thirteen features was utilized. These include the spectral centroid from the Fast Fourier Transform (FFT) in lead aVF, representing the center of mass of the frequency spectrum; kurtosis in lead aVR, a statistical measure of the signal distribution; the mean of wavelet coefficients at level 1 in lead aVR, which is the average of the wavelet decomposition coefficients at the first level; the spectral centroid from FFT in lead I; the standard deviation in lead II, indicating the variability of the signal amplitude; the dominant frequency from FFT in lead II, identifying the most prominent frequency component; the minimum amplitude in lead V3; the standard deviation of the PR interval in lead V3, reflecting the variability of the time from the P wave\u0026apos;s start to the QRS complex\u0026apos;s start; the spectral centroid from FFT in lead V4; the maximum R-peak amplitude in lead V4, the highest amplitude of the QRS complex\u0026apos;s peak; kurtosis in lead V5; the mean QT interval in lead V5, the average duration from the Q wave\u0026apos;s start to the T wave\u0026apos;s end; and the energy of wavelet coefficients at level 1 in lead V6, representing the energy content of the wavelet decomposition coefficients at the first level.\u003c/p\u003e\n\u003cp\u003eHarrell\u0026rsquo;s C index of discrimination can provide helpful information on the predictive performance of a predictive model. As shown in Table 3, Harrell\u0026rsquo;s C index for models with and without ECG signal features in men were 0.77 (CI: 0.75\u0026ndash;0.79) and 0.77 (95% CI: 0.74\u0026ndash;0.79), respectively. There was a slight difference in the goodness of fit as indicated by AIC between these two risk algorithms (AIC: 3896 vs. 3899) in men. Also, the Harrell\u0026rsquo;s C index for models with and without ECG signal features in women was 0.85 (CI: 0.83\u0026ndash;0.88) and 0.84 (95% CI: 0.81\u0026ndash;0.86), respectively. There was a significant difference in the goodness of fit as indicated by AIC between these two risk algorithms (AIC: 2375 vs. 2395) in women. Examining the clinical relevance of a new risk biomarker corresponds to studying the predictive power of a currently available predictive model augmented by new biomarker(s). Generally, the addition of ECG signal features to the Framingham model in women significantly improved risk classification as indicated by cut point-free NRI of 55.7% (95% CIs: 46.5\u0026ndash;65.0%), absolute IDI of 2.8% (95% CIs: 1.0\u0026ndash;4.6%) but the addition of ECG signal features to the Framingham model in men significantly didn\u0026apos;t improve risk classification.\u003c/p\u003e\n\u003cp\u003eTable 4 presents the improvement in the reclassification of individuals between risk categories after complementing FRS with ECG signal features. In each of the four FRS categories (i.e., 0\u0026ndash;4.9%, 5\u0026ndash;7.4%, 7.5\u0026ndash;19.9%, and \u0026ge;20%), 18.5%, 9.1%, 10.3%, and -16.1% of women were correctly reclassified, respectively. In each of the four FRS categories (i.e., 0\u0026ndash;4.9%, 5\u0026ndash;7.4%, 7.5\u0026ndash;19.9%, and \u0026ge;20%), 0.0%, 5.9%, 1.7%, and -3.0% of men were correctly reclassified, respectively.\u0026nbsp;\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis study represents the first evaluation of electrocardiogram (ECG) signal features within the Tehran Lipid and Glucose Study (TLGS) for predicting cardiovascular disease (CVD) risk. Our findings demonstrate that ECG signal features significantly enhance the predictive capacity for CVD in women, both statistically and clinically. Prior research by Mahdavi et al. evaluated the American Heart Association (AHA) risk score classification in TLGS participants, noting its effectiveness in identifying high-risk individuals but limited ability to accurately distinguish those with severe cardiovascular outcomes (\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e). This suggests a need for refined risk stratification approaches to improve predictive accuracy.\u003c/p\u003e \u003cp\u003eKhalili et al. previously highlighted the predictive utility of abnormal resting ECGs compared to Rose Questionnaire angina in estimating 10-year coronary heart disease (CHD) risk in an urban Iranian population. Their study categorized participants into four groups based on Rose Angina and ECG ischemia status, finding that adding abnormal ECG findings to angina did not significantly increase CHD event risk prediction. However, their analysis relied on the Minnesota Coding (MC) system, a standardized ECG classification method widely used in epidemiological studies, rather than raw ECG signal data (\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e). The MC system, introduced in 1960 and expanded in 1983 to include serial comparisons, involves complex measurement protocols that make visual coding time-consuming and prone to errors. In contrast, the current study leverages ECG signal feature extraction to minimize measurement and coding errors, offering a more robust approach to risk prediction (\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eMost population-based studies employ the MC system for ECG coding, which can be performed manually or through automated methods. Both approaches, however, are susceptible to errors, and neither manual coding by a single individual nor automated techniques can be considered fully reliable (\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e). Hadaegh et al. demonstrated the added value of ECG abnormalities beyond the Framingham Risk Score (FRS) for CHD risk stratification in Middle Eastern women. Their findings indicated that incorporating ECG abnormalities\u0026mdash;specifically ST depression or T-wave changes\u0026mdash;into the FRS did not significantly improve C-statistics but enhanced predictive performance by 20.8% (95% CI 5.0\u0026ndash;38.9) using the cut-point-free Net Reclassification Improvement (NRI). Notably, among women, ECG abnormalities were independently associated with increased CHD risk only in the intermediate risk group. Like Khalili et al., their study utilized MC for ECG analysis (\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe reliance on manual ECG pattern recognition and MC in population-based and clinical studies is increasingly being questioned, as these methods are labor-intensive and error-prone. Automated ECG signal processing, as applied in the current study, offers a promising alternative. By extracting signal features directly from ECG data, this approach reduces errors associated with visual coding and enhances the precision of CVD risk prediction. This study underscores the potential of ECG signal analysis to refine risk stratification and improve outcomes in epidemiological research, particularly for women in intermediate risk groups.\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eIn this study, the Minnesota Coding (MC) system and select ECG features were not employed; instead, comprehensive ECG signal data were analyzed. This approach significantly enhanced the statistical and clinical prediction of cardiovascular disease (CVD) risk in women, but no such effect was observed in men.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003eThis study was performed according to the ethical principles of the Helsinki Declaration, and all procedures involving human subjects were approved by the National Research Council of the Islamic Republic of Iran (IR.SBMU.ENDOCRINE.REC.1398.103), the Human Research Review Committee of the Endocrine Research Center, Shahid Beheshti University, Tehran, Iran. Written informed consent was obtained from all subjects.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate:\u0026nbsp;\u003c/strong\u003eThe data for this research were obtained from the Tehran Lipid and Glucose Study (TLGS), conducted by the Endocrine Research Center at Shahid Beheshti University of Medical Sciences. The authors would like to express their gratitude to everyone involved in the design and data collection of the TLGS, as well as to the study participants. This project was approved by the Ethics Committee of Tarbiat Modares University under the code IR.MODARES.REC.1403.100\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication:\u003c/strong\u003e Not applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials:\u003c/strong\u003e The datasets used and analyzed in the current study are available from the corresponding author upon reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interest:\u003c/strong\u003e The authors declare that they have no competing interests, whether financial or non-financial, that could influence the integrity of this research.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding:\u003c/strong\u003e None\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor\u0026apos;s contributions:\u003c/strong\u003e \u0026apos;MM\u0026apos;: Data collection, literature review, and manuscript preparation. \u0026apos;DK\u0026apos;: Study design, revising the manuscript, and final approval of the manuscript. \u0026apos;MM\u0026apos;, \u0026apos;AT\u0026apos;, and \u0026apos;KH\u0026apos;: Data analysis and data interpretation. \u0026apos;AA\u0026apos; and \u0026apos;AK\u0026apos;: Study design, manuscript preparation, revising the manuscript, and the final approval of the manuscript. All authors reviewed and approved the final draft.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgment:\u0026nbsp;\u003c/strong\u003eThe data for this research were obtained from the Tehran Lipid and Glucose Study (TLGS), conducted by the Endocrine Research Center at Shahid Beheshti University of Medical Sciences. The authors would like to express their gratitude to everyone involved in the design and data collection of the TLGS, as well as to the study participants. This project was approved by the Ethics Committee of Tarbiat Modares University under the code IR.MODARES.REC.1403.100.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eTaheri Soodejani M. Non-communicable diseases in the world over the past century: a secondary data analysis. Frontiers in Public Health. 2024;12:1436236.\u003c/li\u003e\n\u003cli\u003eRoth GA, Mensah GA, Johnson CO, Addolorato G, Ammirati E, Baddour LM, et al. Global burden of cardiovascular diseases and risk factors, 1990\u0026ndash;2019: update from the GBD 2019 study. Journal of the American college of cardiology. 2020;76(25):2982-3021.\u003c/li\u003e\n\u003cli\u003eAminorroaya A, Yoosefi M, Rezaei N, Shabani M, Mohammadi E, Fattahi N, et al. Global, regional, and national quality of care of ischaemic heart disease from 1990 to 2017: a systematic analysis for the Global Burden of Disease Study 2017. European journal of preventive cardiology. 2022;29(2):371-9.\u003c/li\u003e\n\u003cli\u003eOrganization WH. Noncommunicable diseases country profiles 2018. 2018.\u003c/li\u003e\n\u003cli\u003eRahmani A, Sayehmiri K, Asadollahi K, Sarokhani D, Islami F, Sarokhani M. Investigation of the prevalence of obesity in Iran: a systematic review and meta-analysis study. Acta Medica Iranica. 2015:596-607.\u003c/li\u003e\n\u003cli\u003eUthman OA. Global, regional, and national disability‐adjusted life years (DALYs) for 315 diseases and injuries and healthy life expectancy (HALE) for 195 countries and territories, 1990‐2015: a systematic analysis for the Global Burden of Diseases, Injuries, and Risk Factors (GBD) 2015 Study. Lancet. 2016;388(10053):1603-58.\u003c/li\u003e\n\u003cli\u003eRoth GA, Johnson C, Abajobir A, Abd-Allah F, Abera SF, Abyu G, et al. Global, regional, and national burden of cardiovascular diseases for 10 causes, 1990 to 2015. Journal of the American college of cardiology. 2017;70(1):1-25.\u003c/li\u003e\n\u003cli\u003eDanaei G, Farzadfar F, Kelishadi R, Rashidian A, Rouhani OM, Ahmadnia S, et al. Iran in transition. The Lancet. 2019;393(10184):1984-2005.\u003c/li\u003e\n\u003cli\u003eMensah GA, Brown DW. An overview of cardiovascular disease burden in the United States. Health affairs. 2007;26(1):38-48.\u003c/li\u003e\n\u003cli\u003eR Jabir N, Nasir Siddiqui A, Kandy Firoz C, Md Ashraf G, Kashif Zaidi S, Shahnawaz Khan M, et al. Current updates on therapeutic advances in the management of cardiovascular diseases. Current pharmaceutical design. 2016;22(5):566-71.\u003c/li\u003e\n\u003cli\u003eGuidry UC, Evans JC, Larson MG, Wilson PW, Murabito JM, Levy D. Temporal trends in event rates after Q-wave myocardial infarction: the Framingham Heart Study. Circulation. 1999;100(20):2054-9.\u003c/li\u003e\n\u003cli\u003ePiepoli MF, Hoes AW, Agewall S, Albus C, Brotons C, Catapano AL, et al. Guidelines: Editor\u0026apos;s choice: 2016 European Guidelines on cardiovascular disease prevention in clinical practice: The Sixth Joint Task Force of the European Society of Cardiology and Other Societies on Cardiovascular Disease Prevention in Clinical Practice (constituted by representatives of 10 societies and by invited experts) Developed with the special contribution of the European Association for Cardiovascular Prevention \u0026amp; Rehabilitation (EACPR). European heart journal. 2016;37(29):2315.\u003c/li\u003e\n\u003cli\u003eLiu S, Li Y, Zeng X, Wang H, Yin P, Wang L, et al. Burden of cardiovascular diseases in China, 1990-2016: findings from the 2016 global burden of disease study. JAMA cardiology. 2019;4(4):342-52.\u003c/li\u003e\n\u003cli\u003eGoff Jr DC, Lloyd-Jones DM, Bennett G, Coady S, D\u0026rsquo;agostino RB, Gibbons R, et al. 2013 ACC/AHA guideline on the assessment of cardiovascular risk: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. Circulation. 2014;129(25_suppl_2):S49-S73.\u003c/li\u003e\n\u003cli\u003eMansoor H, Elgendy IY, Segal R, Bavry AA, Bian J. Risk prediction model for in-hospital mortality in women with ST-elevation myocardial infarction: a machine learning approach. Heart \u0026amp; Lung. 2017;46(6):405-11.\u003c/li\u003e\n\u003cli\u003eFaizal ASM, Thevarajah TM, Khor SM, Chang S-W. A review of risk prediction models in cardiovascular disease: conventional approach vs. artificial intelligent approach. Computer methods and programs in biomedicine. 2021;207:106190.\u003c/li\u003e\n\u003cli\u003eAzizi F, Madjid M, Rahmani M, Emami H, Mirmiran P, Hadjipour R. Tehran Lipid and Glucose Study (TLGS): rationale and design. Iranian journal of endocrinology and metabolism. 2000;2(2):77-86.\u003c/li\u003e\n\u003cli\u003eAzizi F, Ghanbarian A, Momenan AA, Hadaegh F, Mirmiran P, Hedayati M, et al. Prevention of non-communicable disease in a population in nutrition transition: Tehran Lipid and Glucose Study phase II. Trials. 2009;10:1-15.\u003c/li\u003e\n\u003cli\u003eSaatchi M, Mansournia MA, Khalili D, Daroudi R, Yazdani K. Estimation of generalized impact fraction and population attributable fraction of hypertension based on JNC-IV and 2017 ACC/AHA guidelines for cardiovascular diseases using parametric G-formula: Tehran Lipid and Glucose Study (TLGS). Risk Management and Healthcare Policy. 2020:1015-28.\u003c/li\u003e\n\u003cli\u003eOrganization WH. Cerebrovascular disorders: a clinical and research classification. Cerebrovascular disorders: a clinical and research classification1978.\u003c/li\u003e\n\u003cli\u003eAfsar FA, Arif M, Yang J. Detection of ST segment deviation episodes in ECG using KLT with an ensemble neural classifier. Physiological measurement. 2008;29(7):747.\u003c/li\u003e\n\u003cli\u003eSingh AK, Krishnan S. ECG signal feature extraction trends in methods and applications. BioMedical Engineering OnLine. 2023;22(1):22.\u003c/li\u003e\n\u003cli\u003eAddison PS. Wavelet transforms and the ECG: a review. Physiological measurement. 2005;26(5):R155.\u003c/li\u003e\n\u003cli\u003eSrinivasulu A, Sriraam N. Signal processing framework for the detection of ventricular ectopic beat episodes. Journal of Medical Signals \u0026amp; Sensors. 2023;13(3):239-51.\u003c/li\u003e\n\u003cli\u003eElectrophysiology TFotESoCtNASoP. Heart rate variability: standards of measurement, physiological interpretation, and clinical use. Circulation. 1996;93(5):1043-65.\u003c/li\u003e\n\u003cli\u003eShaffer F, Ginsberg JP. An overview of heart rate variability metrics and norms. Frontiers in public health. 2017;5:258.\u003c/li\u003e\n\u003cli\u003eHarrell Jr FE, Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Statistics in medicine. 1996;15(4):361-87.\u003c/li\u003e\n\u003cli\u003ePencina MJ, D\u0026apos;Agostino Sr RB, D\u0026apos;Agostino Jr RB, Vasan RS. Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Statistics in medicine. 2008;27(2):157-72.\u003c/li\u003e\n\u003cli\u003eMahdavi M, Kazemnejad A, Asosheh A, Khalili D. Cardiovascular risk patterns through AI-enhanced clustering of longitudinal health data. Journal of Diabetes \u0026amp; Metabolic Disorders. 2025;24(1):1-10.\u003c/li\u003e\n\u003cli\u003ePrineas RJ, Crow RS, Zhang Z-M. The Minnesota code manual of electrocardiographic findings: Springer Science \u0026amp; Business Media; 2009.\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Tables","content":"\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" align=\"\" width=\"747\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 100%;\"\u003e\n \u003cp\u003eTable 1: ECG Signal Analysis Features\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26.6399%;\"\u003e\n \u003cp\u003eFeatures\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73.3601%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26.6399%;\"\u003e\n \u003cp\u003eDomain-frequency\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73.3601%;\"\u003e\n \u003cp\u003eThese features use the Fast Fourier Transform (FFT) to analyze the signal in the frequency domain, including:\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003eThe amplitude of the ECG signal over time\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eThe average and midpoint of the signal values \u0026nbsp;\u003c/li\u003e\n \u003cli\u003eThe amount of signal dispersion around the mean\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eThe minimum and maximum signal amplitudes and the distance between them.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eThe amount of asymmetry in the distribution of amplitudes\u003c/li\u003e\n \u003cli\u003eThe peak intensity of the distribution of the number of outliers\u003c/li\u003e\n \u003cli\u003eThe root mean square value describes the signal\u0026apos;s overall power and indicates the signal\u0026apos;s overall strength.\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26.6399%;\"\u003e\n \u003cp\u003eDomain-time\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73.3601%;\"\u003e\n \u003cp\u003eA general statistical representation of the amplitude of the ECG signal over time, including:\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003eThe total energy of the signal across all frequencies\u003c/li\u003e\n \u003cli\u003eIdentifying the frequency that has the most energy\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eThe center of gravity is the frequency spectrum associated with the heart rate. It determines whether the energy is concentrated at low or high frequencies. These features are useful for detecting arrhythmias or heart rhythm abnormalities.\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26.6399%;\"\u003e\n \u003cp\u003eWavelet\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73.3601%;\"\u003e\n \u003cp\u003eWavelet decomposition decomposes the ECG signal into different levels (up to level 3) to extract time-frequency information from the signal, including:\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003eThe energy of the wavelets at each level, which is useful for identifying sharp changes such as the QRS complex.\u003c/li\u003e\n \u003cli\u003eAverage wavelet coefficients at each level, which are very useful for identifying local changes in the signal structure and play an important role in detecting ectopic beats or ischemia.\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26.6399%;\"\u003e\n \u003cp\u003ePeak-R and morphological distances\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73.3601%;\"\u003e\n \u003cp\u003eThese features are extracted based on R-peaks (peak points of the heart rate), including:\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003eThe maximum amplitude of one of the peak-R waves\u003c/li\u003e\n \u003cli\u003eThe average duration of the QRS complex indicates the duration of ventricular activation. The time interval from the beginning of the P wave to the peak of the R wave\u003c/li\u003e\n \u003cli\u003eThe time interval from the beginning of the Q wave to the end of the T wave is very important for diagnosing electrical disorders of the heart.\u003c/li\u003e\n \u003cli\u003eThe fluctuation of each of these intervals\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26.6399%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73.3601%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 26.6399%;\"\u003e\n \u003cp\u003eHeart Rate Variability (HRV)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73.3601%;\"\u003e\n \u003cp\u003eHRV features measure the time variation between successive R-peaks and are indicative of autonomic nervous system activity including:\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003eThe average of the R-R intervals, which is the inverse of the heart rate.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eThe fluctuation of these intervals.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eThe root mean square of the differences between consecutive R-R intervals and is suitable for assessing the variability of the heart rate over short intervals.\u003c/li\u003e\n \u003cli\u003eThe percentage of R-R intervals whose difference is greater than 50 milliseconds and indicates the activity of the parasympathetic branch of the nervous system.\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" align=\"\" width=\"699\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 38.9799%;\"\u003e\n \u003cp\u003eTable 2: Baseline characteristics of Participants\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 7.0354%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 15.7821%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" valign=\"top\" style=\"width: 23.1978%;\"\u003e\n \u003cp\u003eWomen\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" valign=\"top\" style=\"width: 25.0993%;\"\u003e\n \u003cp\u003eMen\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 15.7821%;\"\u003e\n \u003cp\u003eVariables\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 8.7467%;\"\u003e\n \u003cp\u003eNon-CVD\u003c/p\u003e\n \u003cp\u003e(n=2353)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003eCVD\u003c/p\u003e\n \u003cp\u003e(n=190)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 5.4192%;\"\u003e\n \u003cp\u003eP-value\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003eNon-CVD\u003c/p\u003e\n \u003cp\u003e(n=1801)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003eCVD\u003c/p\u003e\n \u003cp\u003e(n=293)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 7.0354%;\"\u003e\n \u003cp\u003eP-value\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 15.7821%;\"\u003e\n \u003cp\u003eAge (year)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 8.7467%;\"\u003e\n \u003cp\u003e45.96\u0026plusmn;11.26\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e58.67\u0026plusmn;9.94\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 5.4192%;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e47.00\u0026plusmn;12.66\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e57.76\u0026plusmn;11.69\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 7.0354%;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 15.7821%;\"\u003e\n \u003cp\u003eSBP (mmHg)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 8.7467%;\"\u003e\n \u003cp\u003e112.59\u0026plusmn;17.79\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e130.90\u0026plusmn;20.95\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 5.4192%;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e117.77\u0026plusmn;16.84\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e127.81\u0026plusmn;20.06\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 7.0354%;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 15.7821%;\"\u003e\n \u003cp\u003eDBP (mmHg)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 8.7467%;\"\u003e\n \u003cp\u003e73.09\u0026plusmn;10.18\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e78.53\u0026plusmn;11.61\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 5.4192%;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e76.05\u0026plusmn;10.11\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e79.15\u0026plusmn;11.49\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 7.0354%;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 15.7821%;\"\u003e\n \u003cp\u003eTC (mg/dl)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 8.7467%;\"\u003e\n \u003cp\u003e196.94\u0026plusmn;39.41\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e218.26\u0026plusmn;41.85\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 5.4192%;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e190.92\u0026plusmn;36.46\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e200.95\u0026plusmn;38.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 7.0354%;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 15.7821%;\"\u003e\n \u003cp\u003eHDL-C (mg/dl)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 8.7467%;\"\u003e\n \u003cp\u003e45.00\u0026plusmn;10.52\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e42.03\u0026plusmn;9.49\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 5.4192%;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e37.64\u0026plusmn;8.53\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e37.94\u0026plusmn;8.50\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 7.0354%;\"\u003e\n \u003cp\u003e0.578\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 15.7821%;\"\u003e\n \u003cp\u003eLDL-C (mg/dl)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 8.7467%;\"\u003e\n \u003cp\u003e121.69\u0026plusmn;32.83\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e136.87\u0026plusmn;35.63\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 5.4192%;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e119.67\u0026plusmn;31.55\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e127.35\u0026plusmn;34.40\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 7.0354%;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 15.7821%;\"\u003e\n \u003cp\u003eFPG (mg/dl)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 8.7467%;\"\u003e\n \u003cp\u003e95.84\u0026plusmn;28.89\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e121.58\u0026plusmn;52.27\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 5.4192%;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e96.95\u0026plusmn;28.31\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e107.72\u0026plusmn;41.08\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 7.0354%;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 15.7821%;\"\u003e\n \u003cp\u003e2-h PG (mg/dl)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 8.7467%;\"\u003e\n \u003cp\u003e110.92\u0026plusmn;44.60\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e134.98\u0026plusmn;56.52\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 5.4192%;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e107.98\u0026plusmn;52.80\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e120.41\u0026plusmn;58.55\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 7.0354%;\"\u003e\n \u003cp\u003e0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 15.7821%;\"\u003e\n \u003cp\u003eDM\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 8.7467%;\"\u003e\n \u003cp\u003e260(11.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e67(37.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 5.4192%;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e165(9.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e57(20.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 7.0354%;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 15.7821%;\"\u003e\n \u003cp\u003eDM medication\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 8.7467%;\"\u003e\n \u003cp\u003e139(5.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e47(24.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 5.4192%;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e72(4.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e29(9.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 7.0354%;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 15.7821%;\"\u003e\n \u003cp\u003eHTN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 8.7467%;\"\u003e\n \u003cp\u003e335(14.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e83(45.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 5.4192%;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e285(16.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e92(31.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 7.0354%;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 15.7821%;\"\u003e\n \u003cp\u003eHTN medication\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 8.7467%;\"\u003e\n \u003cp\u003e111(4.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e29(15.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 5.4192%;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e46(2.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e20(6.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 7.0354%;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 15.7821%;\"\u003e\n \u003cp\u003eCurrent Smoker\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 8.7467%;\"\u003e\n \u003cp\u003e65(2.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e6(3.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 5.4192%;\"\u003e\n \u003cp\u003e0.708\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e449(25.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9.0319%;\"\u003e\n \u003cp\u003e76(26.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 7.0354%;\"\u003e\n \u003cp\u003e0.732\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"7\" valign=\"top\" style=\"width: 65.8856%;\"\u003e\n \u003cp\u003eSBP, systolic blood pressure; DBP, diastolic blood pressure; TC, Total Cholesterol; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; FPG, fasting plasma glucose; 2-hPG,\u0026nbsp;2-hour post-challenge plasma glucose; DM, diabetes melitus; HTN, Hypertension.\u003c/p\u003e\n \u003cp\u003eData are given as the mean\u0026plusmn;SD \u0026nbsp; \u0026nbsp; for continuous variables and data are given as the n(%) for categorical variables\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" align=\"\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"3\" valign=\"top\" style=\"width: 624px;\"\u003e\n \u003cp\u003eTable3: Predictive performances of the basic Framingham\u0026rsquo;s \u0026ldquo;general CVD risk\u0026rdquo; algorithm vs. enhanced model\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 207px;\"\u003e\n \u003cp\u003eBasic model\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eenhanced model\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eWomen\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 207px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eHarrell\u0026rsquo;s C index (95% CIs)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 207px;\"\u003e\n \u003cp\u003e0.84(0.81,0.86)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003e0.85(0.83,0.88)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eAkaike information criterion\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 207px;\"\u003e\n \u003cp\u003e2395\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003e2375\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eAdded predictive values\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 207px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eAbsolute IDI (95% CIs)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 207px;\"\u003e\n \u003cp\u003e0.0280(0.0100,0.0460)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eP-value=0.002\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eRelative IDI (95% CIs)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 207px;\"\u003e\n \u003cp\u003e0.2139(-0.0421,0.4699)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eP-value=0.102\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eCutpoint-based NRI (95% CIs)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 207px;\"\u003e\n \u003cp\u003e0.1541(0.0416,0.2665)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eP-value=0.007\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eCutpoint-free NRI (95% CIs)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 207px;\"\u003e\n \u003cp\u003e0.5575(0.4653,0.6496)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eP-value\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMen\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 207px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eHarrell\u0026rsquo;s C index (95% CIs)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 207px;\"\u003e\n \u003cp\u003e0.77(0.74,0.79)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003e0.77(0.75,0.79)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eAkaike information criterion\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 207px;\"\u003e\n \u003cp\u003e3899\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003e3896\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eAdded predictive values\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 207px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eAbsolute IDI (95% CIs)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 207px;\"\u003e\n \u003cp\u003e0.0030(-0.0022,0.0081)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eP-value=0.261\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eRelative IDI (95% CIs)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 207px;\"\u003e\n \u003cp\u003e0.0242(-0.0501,0.0986)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eP-value=0.523\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eCutpoint-based NRI (95% CIs)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 207px;\"\u003e\n \u003cp\u003e0.0154(-0.0601,0.0908)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eP-value=0.690\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eCutpoint-free NRI (95% CIs)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 207px;\"\u003e\n \u003cp\u003e0.1138(-0.0322,0.2598)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eP-value=0.127\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"3\" valign=\"top\" style=\"width: 624px;\"\u003e\n \u003cp\u003eAkaike information criterion (AIC) was used as a measure of model fit. The lower is the AIC the better will be the model fitness. Difference in AIC \u0026gt; 10 was considered significant\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" align=\"\" width=\"671\" class=\"fr-table-selection-hover\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"9\" valign=\"top\" style=\"width: 671px;\"\u003e\n \u003cp\u003eTable4: Reclassification table comparing risk strata for models incorporating CVD risk factors with and without ECG signal features\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 277px;\"\u003e\n \u003cp\u003eModel with \u0026nbsp;ECG signal features\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 189px;\"\u003e\n \u003cp\u003eReclassified\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eNet correctly reclassified %\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 45px;\"\u003e\n \u003cp\u003e\u0026lt;%5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e%5 - %7.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e%7.5 - %20\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e\u0026ge;%20\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003eIncreased risk\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eDecreased risk\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eWomen\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 45px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003eEvent\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 45px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e0 - 5%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 45px;\"\u003e\n \u003cp\u003e22\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003e18.5\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e5 - 7.5%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 45px;\"\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003e9.1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e7.5 - 20%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 45px;\"\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e44\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e13\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003e10.3\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e\u0026ge; 20%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 45px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e47\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003e-16.1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003eNon-evnet\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 45px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e0 - 5%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 45px;\"\u003e\n \u003cp\u003e1464\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e70\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e16\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e87\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003e5.6\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e5 - 7.5%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 45px;\"\u003e\n \u003cp\u003e82\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e79\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e53\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e54\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e82\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003e-13.0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e7.5 - 20%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 45px;\"\u003e\n \u003cp\u003e34\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e78\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e210\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e31\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e31\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e112\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003e-22.9\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e\u0026ge; 20%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 45px;\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e46\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e90\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e48\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003e-34.8\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMen\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 45px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003eEvent\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 45px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e0 - 5%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 45px;\"\u003e\n \u003cp\u003e7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003e0.0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e5 - 7.5%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 45px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e12\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003e5.9\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e7.5 - 20%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 45px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e110\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003e1.7\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e\u0026ge; 20%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 45px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e130\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003e-3.0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003eNon-evnet\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 45px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e0 - 5%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 45px;\"\u003e\n \u003cp\u003e402\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e41\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e41\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003e9.3\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e5 - 7.5%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 45px;\"\u003e\n \u003cp\u003e57\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e283\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e41\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e41\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e57\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003e-4.2\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e7.5 - 20%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 45px;\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e41\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e530\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e42\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003e-3.5\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e\u0026ge; 20%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 45px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e22\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e267\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e22\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003e-7.6\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"9\" valign=\"top\" style=\"width: 671px;\"\u003e\n \u003cp\u003eACC/AHA, American College of Cardiology/American Heart Association\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"ECG signal, NRI and IDI, CVD, prediction model","lastPublishedDoi":"10.21203/rs.3.rs-6542485/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6542485/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eIntroduction:\u003c/h2\u003e \u003cp\u003eNon-communicable diseases (NCDs), particularly cardiovascular diseases (CVDs), have become the leading cause of mortality worldwide, with Iran exhibiting higher-than-average incidence and mortality rates. Early detection of high-risk individuals is critical, as CVD often progresses silently. Electrocardiogram (ECG) signals, when integrated with machine learning (ML), may enhance risk prediction beyond traditional models.\u003c/p\u003e\u003ch2\u003eObjective\u003c/h2\u003e \u003cp\u003eThis study aimed to evaluate the predictive performance of ECG signal features for incident CVD using machine learning models in a large population-based cohort from the Tehran Lipid and Glucose Study (TLGS).\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eA total of 4,637 adults aged 40\u0026ndash;79 years without prior CVD at baseline (2006\u0026ndash;2008) were followed up until 2018. Baseline characteristics, laboratory measurements, and ECG signal features were collected. CVD events were defined as coronary heart disease (CHD) or stroke. A Weibull regression model assessed the association between ECG features and incident CVD, with model performance evaluated using Harrell\u0026rsquo;s C-index, Net Reclassification Index (NRI), and Integrated Discrimination Improvement (IDI).\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eOver a 10-year follow-up, 483 participants (10.4%) developed CVD. The addition of ECG signal features improved risk prediction in women, increasing the Harrell\u0026rsquo;s C-index from 0.84 to 0.85 and demonstrating significant reclassification improvement (NRI: 55.7%, IDI: 2.8%). However, no meaningful improvement was observed in men. ECG-based modeling outperformed traditional risk scores, particularly for intermediate-risk categories among women.\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e \u003cp\u003eIncorporating ECG signal features into ML-based risk models significantly enhanced CVD prediction performance in women, suggesting potential utility for improving individualized preventive strategies. Further research is warranted to refine ECG-based risk stratification tools for broader clinical application.\u003c/p\u003e","manuscriptTitle":"Enhancing CVD Risk Prediction: Integrating ECG Signals with Conventional Models Using AI","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-06-26 08:20:41","doi":"10.21203/rs.3.rs-6542485/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-09-03T04:05:10+00:00","index":"","fulltext":""},{"type":"reviewerAgreed","content":"171490957037726679771549767193181180218","date":"2025-08-03T12:52:25+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-07-25T12:38:00+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-07-20T23:32:52+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"140197207819964418712034905912733818412","date":"2025-07-14T11:26:04+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-07-13T07:10:51+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"285875803302778790352622743306646474767","date":"2025-07-10T06:36:47+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"83794735921916276899621314440025669397","date":"2025-07-10T03:17:07+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"201264561318986694098468363538864942549","date":"2025-06-30T00:59:04+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-06-24T01:00:20+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-06-09T10:34:15+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-05-12T03:49:16+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-05-08T07:35:43+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2025-04-27T21:26:07+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"18f14100-5832-45c8-ba1e-0183e0c1d1ea","owner":[],"postedDate":"June 26th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":50500261,"name":"Health sciences/Cardiology/Cardiovascular biology/Cardiovascular diseases"},{"id":50500262,"name":"Health sciences/Biomarkers/Predictive markers"}],"tags":[],"updatedAt":"2025-11-10T16:00:26+00:00","versionOfRecord":{"articleIdentity":"rs-6542485","link":"https://doi.org/10.1038/s41598-025-26471-6","journal":{"identity":"scientific-reports","isVorOnly":false,"title":"Scientific Reports"},"publishedOn":"2025-11-07 15:57:20","publishedOnDateReadable":"November 7th, 2025"},"versionCreatedAt":"2025-06-26 08:20:41","video":"","vorDoi":"10.1038/s41598-025-26471-6","vorDoiUrl":"https://doi.org/10.1038/s41598-025-26471-6","workflowStages":[]},"version":"v1","identity":"rs-6542485","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6542485","identity":"rs-6542485","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}