Skip to main content

Which risk factor best predicts coronary artery disease using artificial neural network method?



Coronary artery disease (CAD) is recognized as the leading cause of death worldwide. This study analyses CAD risk factors using an artificial neural network (ANN) to predict CAD.


The research data were obtained from a multi-center study, namely the Iran-premature coronary artery disease (I-PAD). The current study used the medical records of 415 patients with CAD hospitalized in Razi Hospital, Birjand, Iran, between May 2016 and June 2019. A total of 43 variables that affect CAD were selected, and the relevant data was extracted. Once the data were cleaned and normalized, they were imported into SPSS (V26) for analysis. The present study used the ANN technique.


The study revealed that 48% of the study population had a history of CAD, including 9.4% with premature CAD and 38.8% with CAD. The variables of age, sex, occupation, smoking, opium use, pesticide exposure, anxiety, sexual activity, and high fasting blood sugar were found to be significantly different among the three groups of CAD, premature CAD, and non-CAD individuals. The neural network achieved success with five hidden fitted layers and an accuracy of 81% in non-CAD diagnosis, 79% in premature diagnosis, and 78% in CAD diagnosis. Anxiety, acceptance, eduction and gender were the four most important factors in the ANN model.


The current study shows that anxiety is a high-prevalence risk factor for CAD in the hospitalized population. There is a need to implement measures to increase awareness about the psychological factors that can be managed in individuals at high risk for future CAD.

Peer Review reports


Cardiovascular diseases (CVDs) are prevalent conditions that affect the heart and vessels, resulting in approximately 31% of all global deaths [1]. CVDs encompass a range of diseases, such as acute coronary syndrome, stroke, heart failure, coronary heart disease, cardiomyopathies, and peripheral vascular diseases [2]. Diabetes, dyslipidemias, obesity, hypertension, low or lack of physical activity, and smoking are cardiovascular-related risk factors [3].

The incidence of CVD may increase due to changes in lifestyles, such as physical inactivity, smoking, drug intake, and an increase in the prevalence of type 2 diabetes mellitus [4]. In addition to the primary risk factors, infection and chronic diseases have recently been considered to be additional risk factors for CVDs [5]. The burden of CVDs in Iran has been projected to rise sharply from 2005 to 2025, primarily due to the country’s aging population. The DALY related to CVDs is predicted to increase by more than two-fold in 2025 compared to 2005 [6]. Identifying and preventing the mentioned risk factors can reduce the prevalence of CVDs. In addition to lifestyle modifications, lipid-lowering drugs, antihypertensives, and antiplatelet and anticoagulant medications are the main prescriptions for preventing and treating CVDs [7].

Data mining, or machine learning, is a process that involves identifying anomalies, patterns, and correlations within large datasets to make predictions that have broad applications in medicine [8]. Various data mining algorithms, such as clustering, decision trees, and neural networks, have been utilized to predict CVDs [9, 10]. Machine learning and data mining methods can be highly beneficial in clinical decision-making because they provide significant decision-making power.

The artificial neural network (ANN) is favored among data mining algorithms due to its simplicity, high speed, and ability to solve complex relationships between variables [11]. The ANN is one of the machine learning techniques with the highest prediction accuracy. This approach has been implemented in numerous studies to determine predictors of Coronary Artery Disease (CAD) [12, 13]. Although, association between coronary heart disease (CHD) risk factors including; age, gender, lipid profle, arterial hypertension, fasting plasma glucose, smoking status, obesity and serum interferon in related CHD were assessed using ANN previously, but the association between demographic characteristic, lifestyle risk factors such as, stress and anxiety as well as clinical risk factors in relation to CAD were not assessed. This study uses the ANN to compare and predict CVD risk factors in a selected hospital.


Design and study population

The present study is a subsection of a multi-center study in Iran, namely the Iran-premature coronary artery disease (I-PAD), which has been previously described in depth [14]. This study included angiographic patients with occlusion of at least one coronary artery ≥ 75% or left main coronary ≥ 50%, as well as control subjects with normal coronary arteries. The participants were selected from Razi Hospital in Birjand, Iran, between 2016 and 2019. The patients undergoing coronary angiography were limited to men and women aged ≤ 60 or ≤ 70 years, respectively. The criteria for premature CAD patients consisted of age ≤ 45 or ≤ 55 years for men and women, respectively. Patients who had previously undergone coronary artery bypass grafting, balloon angioplasty, or percutaneous coronary intervention were excluded from the study. This study determined 500 patients undergoing coronary angiography in Birjand between 2016 and 2019 as the sample size. Patients were searched through the Isfahan angiographic data registration system, and if they met the inclusion criteria, they were contacted by interviewers and invited to participate in the study. Additionally, eligible angiographic patients hospitalized in the ward were selected for questioning. Among the 415 individuals included in the study, 54.9% (228) were female. The mean age of the participants was 54.10 ± 7.68 years. Among the participants, 200 individuals (48.2%) exhibited CVD, characterized by the presence of at least one vascular stenosis exceeding 50% in the left main artery and 70% in other vessels. The remaining 215 participants (51.8%) did not show signs of CAD. Within this group, 39 individuals (9.4%) had premature CAD, while 161 individuals (38.8%) had CAD.

The study received approval from the Ethics Committee of Isfahan University of Medical Sciences (IR.MUI.REC.1396.2.055).

Data collection

The data collected pertained to age, sex, religion, ethnicity, education, occupation, economic status of the family, history of drug use, smoking, and alcohol consumption, as well as personal and family history of CVD and medications. Behavioral patterns (anxiety, depression, and coping with stress) were collected using the Hospital Anxiety and Depression Scale, sleep quality using the Pittsburgh Sleep Quality Index, physical activity using the International Physical Activity Questionnaire, and dietary patterns using the Food Frequency Questionnaire. Then, according to standard protocols, height, waist circumference, hip circumference, neck circumference, and thighs were measured for each patient. The methodology paper of this study provides a thorough description of the study’s methodology [14]. Each group’s sample size was calculated using a = 0.05, a power of 0.8, and an odds ratio of 1.30. Individuals were recruited through convenience sampling in hospitals across the country [14].

The following instruments were used in the current study: Physical Activity Questionnaire (20 items), Stress Coping Questionnaire (23 items), Depression Questionnaire (17 items), Environmental Questionnaire, Sensitivity Questionnaire, Oral and Dental Questionnaire, Sleep Quality Questionnaire (17 items), Sexual Activity Questionnaire (10 items), Drug Use Questionnaire, Cigarettes and Alcohol Questionnaire, Biography Questionnaire that includes personal or familial history of the disease, Medications Questionnaire, Family Economic Status Questionnaire (12 items), and Nutrition Questionnaire (112 items). Moreover, the Physical Examination Questionnaire was administered, which includes patients’ vital signs (pulse, systolic blood pressure (SBP), and diastolic blood pressure (DBP). These parameters were measured twice, 15 min apart [14].

General tests were taken and recorded after 12 h of fasting. They comprised total cholesterol (Chol), triglyceride (TG), high-density lipoprotein cholesterol (HDL-c), low-density lipoprotein cholesterol (LDLc), and fasting blood sugar (FBS).

Diabetes Mellitus (DM) defined FBS ≥ 126 mg/dl or history of diabetes, and DLP defined as history of DLP or one of lipid profile was abnormal. Lipid profile including TC, TG, LDL, HDL and FBS standardized according to the defined values. Based on the standards, the normal and abnormal values of the laboratory test were considered as follows: TG ≥ 150 mg/dl, Chol ≥ 240 mg/dl, LDL ≥ 100 mg/dl and HDL < 40 mg/dl in males and < 50 mg/dl for females considered abnormal.

Statistical analysis

Inspired by the function of the human brain, the ANN refers to a family of models with a large parametric space and a flexible structure. A neural network, in fact, is a parallel distribution processor that aims to store experimental knowledge and make it usable [15]. Any neural network that includes a hidden layer has the potential to be a reliable predictor, provided that there is an adequate number of hidden neurons [16]. However, there is no theoretical consensus on finding the optimal number of hidden neurons and functions between layers [17]. One type of ANN is the feed-forward network, in which the path always moves forward and does not return to the neurons of the previous layer.

According to Akella and Akella (2021), the neural network (NN) is the best method for predicting and diagnosing CAD. The reported accuracy of the NN is 93%, with a specificity of 0.93 and a sensitivity of 0.89 [18]. The steps of the ANN in our study include data preparation, model creation, assessment, and interpretation. During the data preprocessing step, several preprocessing approaches were utilized on the dataset to ensure the integrity and suitability of the data for analysis using ANN. One task involved in the data cleaning process was identifying outliers and data inconsistencies to maintain the integrity of the dataset [19].

In order to address issues related to the different magnitudes of variables, a standardization process was implemented during the scaling stage. This process aimed to establish a uniform scale for continuous variables so as to ensure consistency. The variance inflation factor was used to identify the presence of multicollinearity among variables, with a threshold of 5 being regarded as an indicator of multicollinearity [20].

The dataset was randomized to create three distinct subgroups. The subsets consisted of a training set, which comprised 70% of the data and was utilized for model training, a testing set that accounted for 20% of the data, and a validation set that constituted 10% of the data [21]. The architectural design incorporated the input layer, which was responsible for including the predictor variables relevant to the study. The optimal number of layers and neurons was determined through a methodical approach that involved conducting experiments and refining hyperparameters. The output layer was implemented using sigmoid activation functions.

The development of the model’s progression:

The training protocol for the ANN model included the following phases:

The training dataset was processed using a feed-forward network [22] in order to generate predictions. The convergence criteria for training involved determining when to stop training based on specific parameters. These parameters included reaching a maximum number of epochs or achieving adequate performance on the validation set. A minimum threshold of 0.001 was established for the relative change in the training error ratio. The criterion mentioned above suggests that it is advisable to stop the training process when the accuracy level exceeds a specific threshold at a particular training stage, also known as early quitting. Additionally, examining concealed layers involved analyzing a wide range of hidden levels, spanning from 1 to 50 [23].

The evaluation of the model:

The performance of the model was evaluated across several classes using Receiver Operating Characteristic (ROC) curves. The model’s performance was assessed by quantifying the area under the ROC curve, commonly referred to as the AUC.

The data analysis in this study was conducted using the SPSS 26 program. The mean and standard deviation were used to describe quantitative data, while counts and percentages were used to represent qualitative factors. An independent t-test was used to evaluate the mean differences of quantitative variables between the premature CAD and non-CAD groups. An analysis of variance (ANOVA) was used to assess the differences in means between premature CAD, early CAD, and non-CAD groups. The chi-square test was employed to analyze qualitative factors across study groups. The exact Fisher’s test was used when the expected frequency in more than 20% of the cells was less than 5. Logistic regression was applied to predict the odds ratio of study variables for the occurrence of premature CAD. The assumption of linearity in logistic regression was tested using the Box-Tidwell test [24]. Additionally, ordinal logistic regression was used to model the status of premature CAD at three levels: premature, early CAD, and non-CAD.


The demographic characteristics, medical history, and lifestyle of the premature CAD and non-CAD groups are presented in Table 1. Gender, education, occupation, smoking, opium use, diabetes mellitus, high blood sugar, high LDL, and history of diabetes mellitus differed significantly between the two groups (p < 0.05 to p < 0.001). Lifestyle risk factors, such as acceptance, exposure to poisons, anxiety, and seasonal allergies and sex activity showed significant differences between patients with premature CAD and non-CAD patients (p < 0.01 and p < 0.001, respectively). The two groups had no statistically significant difference for the other variables (Table 1). The analysis was conducted using the simple logistic regression method.

Table 1 Crude CAD odds ratio and demographic, lifestyle, and clinical risk factors by logistic regression in patients without CAD and premature CAD

The underlying factors were also compared between the three groups of non-CAD, early CAD, and premature CAD patients. The variables of age, sex (female), occupation, smoking, opium use, exposure to poisons, anxiety, seasonal allergy, sexual activity, and high fasting blood sugar were significantly different in the three groups (p < 0.05 to p < 0.001) (Table 2). The analysis was conducted using the simple ordinal logistic regression method.

Table 2 Crude CAD odds ratio and demographic, lifestyle, and clinical risk factors by ordinal logistic regression in patients without CAD, early CAD, and premature CAD

The neural network method was used to investigate the simultaneous effect of input variables and eliminate any possible alignment effects between them. The efficiency of the models was calculated by considering the number of hidden layers and hidden nodes, and the results of the ROC curve analysis were presented. The neural network was fitted with seven hidden layers for the premature CAD/non-CAD output variable. The final model achieved a diagnostic accuracy of 98% in distinguishing between premature CAD and non-CAD patients (Fig. 1a).

Fig. 1
figure 1

The model with 98% accuracy in diagnosing premature CAD and non-CAD patients (A), 81% accuracy in non-CAD, 79% in Early CAD, and 78% in premature CAD diagnosis (B)

Furthermore, the output variable proposed in the model (non-CAD, early CAD, premature CAD) was assessed using the same significant input variables listed in Table 2. The neural network achieved success with five hidden fitted layers and an accuracy of 81% in non-CAD diagnosis, 79% in early diagnosis, and 78% in premature CAD diagnosis (Fig. 1b).

Figure 2 displays the obtained ranking and level of importance of the risk factors. The most significant factors contributing to premature CAD were anxiety (100%), high blood pressure (99.4%), acceptance (80.5%), avoidance (76.9%), and seasonal allergies (75.3%). Low level of education (68.1%) and occupation (62%) were ranked next in terms of importance. Opium (30.7%) and smoking use (25.8%) were identified in the network with less than 50% prevalence (Fig. 2a).

Fig. 2
figure 2

The importance of factors in the occurrence of premature CAD. A: Comparison between non-CAD / premature CAD groups. B: Comparison between non-CAD / Early CAD / premature CAD groups

According to the neural network model for output variables (non-CAD / early CAD / premature CAD), the most important factors identified were anxiety (100%), acceptance (75.3%), education (67.3%), gender (51.9%), and high blood pressure (50.2%). The importance rank of opium was 38.3% (Fig. 2b).


The results of the current study indicate that the important risk factors affecting the incidence of CAD are gender, age, education, occupation, smoking, opium use, pesticide exposure, history of dyslipidemia, history of diabetes, high blood sugar, acceptance, anxiety, seasonal allergy and sexual activity using the simple logistic regression method.

The incidence of the disease in the current study was found to be lower in women than in men. Additionally, it was observed that age and weight could potentially increase the risk of CAD. Previous research has also demonstrated that certain risk factors, including age, smoking, and sex (male), exhibit positive correlations with CAD [11, 25]. Furthermore, the risk of CAD was found to be higher among employed patients than those who were unemployed. The results of a meta-analysis of 13 European cohort studies conducted between 1985 and 2006 revealed an increased risk of CVDs associated with increased workload and job strain [26]. Therefore, the type of job, including its time and duration, can be a risk factor for CVDs.

The results indicate that individuals with lower levels of education have a higher incidence of CAD than those with higher education levels. Similarly, previous studies have revealed that individuals with lower education are at a greater risk [27, 28], which supports the results of the present study. Individuals with low socioeconomic status and education tend to exhibit a higher prevalence of drug abuse, smoking, diabetes, hypertension, and hyperlipidemia as risk factors for CVDs (CVDs).

Based on the results, the prevalence of smoking and opium use is higher in patients with CAD compared to those without CAD. Alongside this, it has been reported that family history, smoking, and co-morbidities can increase the risk of CAD in patients [29]. Besides, there is a stronger correlation between CAD in men and the use of opium and smoking. In addition, higher prevalence of opium use in males results in different patterns of premature CAD [30] and coronary artery bypass grafting in patients [31]. Indeed, the results of our previous study indicated that opium consumption reduced the ejection fraction, which is a key variable for heart failure [32].

The results of a basic study indicated that the administration of opium decreased the serum levels of AST, ALT, total protein, total cholesterol, and TG while also resulting in an increase in urea and creatinine in diabetic rats [33]. The results of the PERSIAN cohort study revealed a positive correlation between opium consumption and lower levels of cholesterol and LDL. Additionally, a decrease in HDL levels was observed among opium abusers [34]. Further, several clinical studies have suggested that opium consumption does not have a significant effect on cholesterol, TG, LDL, or HDL-C [35, 36]. On the other hand, controversial evidence regarding opium use has been reported, including its potential effects on lowering lipid profile and FBS, reduction in left ventricular ejection fraction (LVEF), and enhancement of free radicals through the activation of lipid peroxidation [37]. The results of the aforementioned studies indicate that smoking and opium use are risk factors for CAD in patients.

Dyslipidemia and diabetes mellitus also have a higher prevalence in patients with premature CAD compared to non-CAD patients in this study. Metabolic syndrome has been linked to an elevated risk of CVD, according to reports. Atherogenic dyslipidemia is a significant modifiable risk factor in patients with CVD [38]. According to reports, type 2 diabetes is associated with a two to four times higher risk of CVD events and significantly raises mortality rates [39]. These results additionally support the findings of the current study.

The current study found that the prevalence of poison exposure as a lifestyle risk factor is greater among patients with CAD and premature CAD than among non-CAD patients. Organophosphorus pesticides (OPs) and household insecticides are widely used worldwide. The results of a cohort study revealed that patients who were acutely exposed to OPs had a higher incidence of arrhythmia, CAD, and CHF compared to patients who were not poisoned by OPs [40]. The adverse effects on heart rhythms may be caused by the suppression of the acetylcholinesterase (AChE) enzyme and the induction of oxidative stress in OPs. Exposure to pyrethroids, a type of insecticide, has been found to have a negative association with CVD and coronary heart disease in adults in the United States [41]. Furthermore, environmental exposure may play a crucial role in the development and severity of CVD. Likewise, air pollution and heavy metals can potentially exacerbate diseases by initiating or increasing pathophysiological processes. These processes include disruptions to carbohydrate and lipid metabolism, as well as impairments to vascular function that are commonly associated with CVD [42]. According to the findings of the studies above, air pollution, cigarette smoking, and opium use may increase the risk of CAD by increasing inflammation and oxidative stress.

Acceptance of stress has a higher prevalence, whereas anxiety has a lower prevalence in patients with CAD than in premature and non-CAD populations. Acceptance can be a valuable tool for alleviating stress.

It has been theorized that acceptance training is an essential element of mindfulness meditations to enhance health outcomes, affective reactivity, and stress levels [43]. Previous studies have indicated that certain psychiatric factors, such as chronic anxiety and daily stressors, have a negative impact on cardiovascular health [44, 45]. The results of a meta-analysis have also indicated that stress is implicated in the prognosis of CVD [46]. Short-term emotional stress has the potential to serve as a catalyst for cardiac events. Similarly, long-term stress, such as work-related stress and social isolation, has been found to be linked to CVD in individuals [45].

The results of the current study showed that the anxiety score was higher in non-CAD patients compared to CAD patients. This may be attributed to their “constructive worrying” and their early visits to physicians. The findings of a study also indicated that patients with generalized anxiety disorder have a capacity for constructive worrying and are more likely to seek help in response to less severe symptoms [47]. Furthermore, stress that is not “overwhelming” or “adequate acute stress” may enhance performance and can be beneficial in certain cases [46].

The results of the current study also showed that seasonal allergy was higher, but sexual activity was lower in non-CAD patients compared to premature CAD patients. Also, the prevalence of female gender and Anxiety in non-CAD patients was higher than premature CAD patients.

It has been reported that female patients with allergic rhinitis show significant higher levels of sensitivity to irritants and airway hyperresponsiveness than males. Furthermore, the female allergic subjects tended to have higher concentrations of substance P before and after non-specific challenges and difference between post allergen challenge was highly significant in female patients [48]. These data indicated that difference in seasonal allergy between premature CAD and non-CAD patients may be gender- related of patients.

In the other hands, seasonal allergies are a risk factor for psychiatric disorders, and association between seasonal allergies and eating disorders, substance use and mood disorders was also reported [49]. Sexual health is related to general health in both genders. Changes in lifestyle including, obesity, smoking and psychosocial factors (stress, depression and anxiety) can contribute and amplify the sexual dysfunction [50]. Weight loss is also associated with an improvement of function and quality of life [51]. According to the results of mentioned studies the significant difference in seasonal allergy and sexual activity might be attributed to the gender of patients.

In our study, the ANN predictors appear reasonable, as most of them are reportedly associated with CAD. In the current study, the most highly ranked and important risk factors for premature CAD and non-CAD patients include anxiety, acceptance, education, gender, and SBP. Low education level, mental stress, diabetes mellitus, hypertension, and obesity, as well as exposure to occupational and environmental risks, are the most significant risk factors for CAD [52]. The epidemiologic studies also suggest that individuals with psychological stress are at an increased cardiovascular risk [53].

In addition, Deep Neural Network techniques were used to create a model for predicting the risk of CVD in type 2 diabetes mellitus patients. The top five predictors in the CVD risk prediction model were BMI, anxiety, depression, total cholesterol, and SBP [54]. Additionally, according to the ANN model, depression, anxiety, and BMI are identified as the three most significant predictive factors for heart attacks. The predictors selected by the ANN in our study are consistent with previous reports [55]. The current study has limitations in terms of design and analysis, specifically related to issues such as reverse causality, incidence-prevalence bias, and unmeasured confounding.


This article introduced an ANN-based model of CAD risk prediction. The proposed model enhances CAD risk assessment and decision support for appropriate treatment. While gender, education, smoking, opium use, history of dyslipidemia, and diabetes have been identified as potential risk factors for CAD, it is not clear if they can be accurately predicted using ANN for CAD risk assessment. There is no association found between a history of high dyslipidemia, diabetes mellitus, and fasting blood sugar in relation to CAD. However, it has been determined that anxiety, acceptance, and gender are the most significant factors contributing to CAD. Although more progress has been made in understanding the contribution of psychological disorders to CAD, further clinical studies are needed to elucidate these mechanisms.

Data availability

All data generated or analyzed during this study are included in the article.





Artificial neural network


Coronary artery bypass grafting


Coronary artery disease




Cardiovascular diseases


Deep Neural Network


Fasting blood sugar


Food Frequency Questionnaire


Generalized anxiety disorder


Anxiety and Depression Scale


High-density lipoprotein cholesterol


Iran-premature coronary artery disease


International Physical Activity Questionnaire


Low-density lipoprotein cholesterol


Left main


Neural network


Percutaneous coronary intervention


Sleep quality using the Pittsburgh Sleep Quality Index


Systolic Blood Pressure


Type 2 diabetes mellitus




  1. EJ, MJ, SE, M, SR, R, et al. Heart disease and stroke statistics—2017 update: a report from the American Heart Association. Circulation. 2017;135(10):e146–e603.

  2. Shaito A, Thuan DTB, Phu HT, Nguyen THD, Hasan H, Halabi S, et al. Herbal medicine for cardiovascular diseases: efficacy, mechanisms, and safety. Front Pharmacol. 2020;11:422.

    CAS  PubMed  PubMed Central  Google Scholar 

  3. Patel SA, Winkel M, Ali MK, Narayan KV, Mehta NK. Cardiovascular mortality associated with 5 leading risk factors: national and state preventable fractions estimated from survey data. Annal Intern med. 2015;163(4):245–53.

    Google Scholar 

  4. Deepa R, Arvind K, Mohan V. Diabetes and risk factors for coronary artery disease. Curr sci. 2002:1497–505.

  5. Bergh C, Fall K, Udumyan R, Sjöqvist H, Fröbert O, Montgomery S. Severe infections and subsequent delayed cardiovascular disease. Eur j Prevent Cardiol. 2017;24(18):1958–66.

    Google Scholar 

  6. Sadeghi M, Haghdoost AA, Bahrampour A, Dehghani M. Modeling the burden of cardiovascular diseases in Iran from 2005 to 2025: the impact of demographic changes. Iran j Public Health. 2017;46(4):506.

    PubMed  PubMed Central  Google Scholar 

  7. Flora GD, Nayak MK. A brief review of cardiovascular diseases, associated risk factors and current treatment regimes. Curr Pharm Design. 2019;25(38):4063–84.

    CAS  Google Scholar 

  8. Bellazzi R, Ferrazzi F, Sacchi L. Predictive data mining in clinical medicine: a focus on selected methods and applications. Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery. 2011;1(5):416–30.

    Google Scholar 

  9. Desai SD, Giraddi S, Narayankar P, Pudakalakatti NR, Sulegaon S. Back-propagation neural network versus logistic regression in heart disease classification. Adv comput communicat technol: Springer; 2019. p. 133– 44.

  10. Kim J, Lee J, Lee Y. Data-mining-based coronary heart disease risk prediction model using fuzzy logic and decision tree. Healthc Inf res. 2015;21(3):167–74.

    Google Scholar 

  11. Ayatollahi H, Gholamhosseini L, Salehi M. Predicting coronary artery disease: a comparison between two data mining algorithms. BMC Public Health. 2019;19(1):1–9.

    Google Scholar 

  12. Cheng X, Han W, Liang Y, Lin X, Luo J, Zhong W et al. Risk prediction of coronary artery stenosis in patients with coronary heart disease based on logistic regression and artificial neural network. Computat Mathematical Methods Med. 2022;2022.

  13. Atkov OY, Gorokhova SG, Sboev AG, Generozov EV, Muraseyeva EV, Moroshkina SY, et al. Coronary heart disease diagnosis by artificial neural networks including genetic polymorphisms and clinical parameters. J Cardiol. 2012;59(2):190–4.

    PubMed  Google Scholar 

  14. Zarepur E, Mohammadifard N, Mansourian M, Roohafza H, Sadeghi M, Khosravi A, et al. Rationale, design, and preliminary results of the Iran-premature coronary artery disease study (I-PAD): a multi-center case-control study of different Iranian ethnicities. ARYA Atherosclerosis. 2020;16(6):295.

    PubMed  PubMed Central  Google Scholar 

  15. Cybenko G. Approximation by superpositions of a sigmoidal function. Math Control Signals Syst. 1989;2(4):303–14.

    MathSciNet  Google Scholar 

  16. Wen Z, Liao W, Chen S. Production of cellulase by Trichoderma reesei from dairy manure. Bioresour Technol. 2005;96(4):491–9.

    CAS  PubMed  Google Scholar 

  17. Kupusinac A, Stokić E, Doroslovački R. Predicting body fat percentage based on gender, age and BMI by using artificial neural networks. Comput Methods Programs Biomed. 2014;113(2):610–9.

    PubMed  Google Scholar 

  18. Akella A, Akella S. Machine learning algorithms for predicting coronary artery disease: efforts toward an open source solution. Future Sci OA. 2021;7(6):FSO698.

    CAS  PubMed  PubMed Central  Google Scholar 

  19. Hinton GE, Salakhutdinov RR. Reducing the dimensionality of data with neural networks. Science. 2006;313(5786):504–7.

    ADS  MathSciNet  CAS  PubMed  Google Scholar 

  20. Ringle CM, Wende S, Becker J-M. SmartPLS 3. SmartPLS GmbH, Boenningstedt. J Service Sci Manag. 2015;10(3):32–49.

    Google Scholar 

  21. Nguyen QH, Ly H-B, Ho LS, Al-Ansari N, Le HV, Tran VQ, et al. Influence of data splitting on performance of machine learning models in prediction of shear strength of soil. Math Probl Eng. 2021;2021:1–15.

    Google Scholar 

  22. Zhao Y-Y, Boyd J, Hrudey SE, Li X-F. Characterization of new nitrosamines in drinking water using liquid chromatography tandem mass spectrometry. Env sci Technol. 2006;40(24):7636–41.

    CAS  Google Scholar 

  23. Caruana R, Lawrence S, Giles C. Overfitting in neural nets: Backpropagation, conjugate gradient, and early stopping. Adv Neural Inf Process Syst. 2000;13.

  24. Agresti A. Categorical data analysis. Hoboken. NJ: wiley; 2002.

    Google Scholar 

  25. Yiu KH, de Graaf FR, Schuijf JD, van Werkhoven JM, Marsan NA, Veltman CE, et al. Age-and gender-specific differences in the prognostic value of CT coronary angiography. Heart. 2012;98(3):232–7.

    PubMed  Google Scholar 

  26. Kivimäki M, Nyberg ST, Batty GD, Fransson EI, Heikkilä K, Alfredsson L, et al. Job strain as a risk factor for coronary heart disease: a collaborative meta-analysis of individual participant data. Lancet. 2012;380(9852):1491–7.

    PubMed  PubMed Central  Google Scholar 

  27. Kilander L, Berglund L, Boberg M, Vessby B, Lithell H. Education, lifestyle factors and mortality from cardiovascular disease and cancer. A 25-year follow-up of Swedish 50-year-old men. Int j Epidemiol. 2001;30(5):1119–26.

    CAS  PubMed  Google Scholar 

  28. Bruthans J, Mayer O Jr, De Bacquer D, De Smedt D, Reiner Z, Kotseva K, et al. Educational level and risk profile and risk control in patients with coronary heart disease. Eur j Prevent Cardiol. 2016;23(8):881–90.

    Google Scholar 

  29. Andria N, Nassar A, Kusniec F, Ghanim D, Qarawani D, Kachel E, et al. Ethnicity of symptomatic coronary artery disease referred for coronary angiography in the Galilee: prevalence, risk factors, and a case for screening and modification. Isr Med Associat J: IMAJ. 2018;20(3):182–5.

    Google Scholar 

  30. Sadeghian S, Graili P, Salarifar M, Karimi AA, Darvish S, Abbasi SH. Opium consumption in men and diabetes mellitus in women are the most important risk factors of premature coronary artery disease in Iran. Int j Cardiol. 2010;141(1):116–8.

    PubMed  Google Scholar 

  31. Aghadavoudi O, Eizadi-Mood N, Najarzadegan MR. Comparing cardiovascular factors in opium abusers and non-users candidate for coronary artery bypass graft surgery. Adv Biomed res. 2015;4.

  32. Moezi SA, Azdaki N, Kazemi T, Partovi N, Hanafi Bojd N, Mashreghi Moghaddam HR et al. The effects of opium uses on syntax score of angiography patients with coronary artery disease (CAD). Toxin Rev. 2021:1–7.

  33. Ahmed HAM, Ahmed SM, El Gawish E, Alanwar AM, Ibrahem M. Effects of Opium Addiction on some biochemical parameters in Diabetic rats. Int J Biochem Res Rev. 2016;10(3):1.

    Google Scholar 

  34. Kazemi M, Bazyar M, Naghizadeh MM, Dehghan A, Rahimabadi MS, Chijan MR, et al. Lipid profile dysregulation in opium users based on Fasa PERSIAN cohort study results. Sci rep. 2021;11(1):1–9.

    Google Scholar 

  35. Fatemi SS, Hasanzadeh M, Arghami A, Sargolzaee MR. Lipid profile comparison between opium addicts and non-addicts. J Tehran Univ Heart Center. 2008;3(3):169–72.

    CAS  Google Scholar 

  36. Asgari S, Naderi G, Soghrati M, Ahmadi P, Shah RJ. A study of plasma lipid peroxidation, lipids and blood sugar level in opium addicts compared with control group. 2005.

  37. Hedayati-Moghadam M, Moezi SA, Kazemi T, Sami A, Akram M, Zainab R et al. The effects of Papaver somniferum (Opium Poppy) on health, its controversies and consensus evidence. Toxin Reviews. 2021:1–14.

  38. Cannon CP. Mixed dyslipidemia, metabolic syndrome, diabetes mellitus, and cardiovascular disease: clinical implications. Am J Cardiol. 2008;102(12):5L–9L.

    PubMed  Google Scholar 

  39. Huxley R, Barzi F, Woodward M. Excess risk of fatal coronary heart disease associated with diabetes in men and women: meta-analysis of 37 prospective cohort studies. BMJ. 2006;332(7533):73–8.

    PubMed  PubMed Central  Google Scholar 

  40. Hung D-Z, Yang H-J, Li Y-F, Lin C-L, Chang S-Y, Sung F-C, et al. The long-term effects of organophosphates poisoning as a risk factor of CVDs: a nationwide population-based cohort study. PLoS ONE. 2015;10(9):e0137632.

    PubMed  PubMed Central  Google Scholar 

  41. Xue Q, Pan A, Wen Y, Huang Y, Chen D, Yang C-X, et al. Association between pyrethroid exposure and cardiovascular disease: a national population-based cross-sectional study in the US. Env Int. 2021;153:106545.

    CAS  Google Scholar 

  42. Cosselman KE, Navas-Acien A, Kaufman JD. Environmental factors in cardiovascular disease. Nat Rev Cardiol. 2015;12(11):627–42.

    CAS  PubMed  Google Scholar 

  43. Lindsay EK, Creswell JD. Mechanisms of mindfulness training: Monitor and Acceptance Theory (MAT). Clin Psychol rev. 2017;51:48–59.

    PubMed  Google Scholar 

  44. Steptoe A, Kivimäki M. Stress and cardiovascular disease: an update on current knowledge. Ann Rev Public Health. 2013;34:337–54.

    Google Scholar 

  45. Steptoe A, Kivimäki M. Stress and cardiovascular disease. Nat Reviews Cardiol. 2012;9(6):360–70.

    CAS  Google Scholar 

  46. Esch T, Stefano GB, Fricchione GL, Benson H. Stress in cardiovascular diseases. Signature. 2002;8(5):101.

    Google Scholar 

  47. Parker G, Hyett M, Hadzi-Pavlovic D, Brotchie H, Walsh W. GAD is good? Generalized anxiety disorder predicts a superior five-year outcome following an acute coronary syndrome. Psychiatry Res. 2011;188(3):383–9.

    PubMed  Google Scholar 

  48. Tomljenovic D, Baudoin T, Megla ZB, Geber G, Scadding G, Kalogjera L. Females have stronger neurogenic response than males after non-specific nasal challenge in patients with seasonal allergic rhinitis. Med Hypotheses. 2018;116:114–8.

    PubMed  Google Scholar 

  49. Oh H, Koyanagi A, DeVylder JE, Stickley A. Seasonal allergies and psychiatric disorders in the United States. Int j env res Public Health. 2018;15(9):1965.

    Google Scholar 

  50. Mollaioli D, Ciocca G, Limoncin E, Di Sante S, Gravina GL, Carosa E, et al. Lifestyles and sexuality in men and women: the gender perspective in sexual medicine. Reproduct Biol Endocrinol. 2020;18:1–11.

    Google Scholar 

  51. Abdelsamea GA, Amr M, Tolba A, Elboraie HO, Soliman A, Al-Amir Hassan B, et al. Impact of weight loss on sexual and psychological functions and quality of life in females with sexual dysfunction: a forgotten avenue. Front Psychol. 2023;14:1090256.

    PubMed  PubMed Central  Google Scholar 

  52. Li H, Luo M, Zheng J, Luo J, Zeng R, Feng N et al. An artificial neural network prediction model of congenital heart disease based on risk factors: a hospital-based case-control study. Medicine. 2017;96(6).

  53. Dimsdale JE. Psychological stress and cardiovascular disease. J Am Coll Cardiol. 2008;51(13):1237–46.

    PubMed  PubMed Central  Google Scholar 

  54. Chu H, Chen L, Yang X, Qiu X, Qiao Z, Song X et al. Roles of anxiety and depression in predicting cardiovascular disease among patients with type 2 diabetes mellitus: a machine learning approach. Front Psychol. 2021:1189.

  55. Sattaru NC, Baker MR, Umrao D, Pandey UK, Tiwari M, Chakravarthi MK, editors. Heart Attack Anxiety Disorder using Machine Learning and Artificial Neural Networks (ANN) Approaches. 2022 2nd International Conference on Advance Computing and Innovative Technologies in Engineering (ICACITE); 2022: IEEE.

Download references


We thank the research council of Birjand University of Medical Sciences and the Iranian Network of Cardiovascular Research (INCVR).


None applicable.

Author information

Authors and Affiliations



S.A.M. and N.A. contributed to designing the study and collecting the data. T.K., N.P., Y.M., and M.N.M. assisted in collecting the data. F.S. and S.K.B. assisted in analyzing the data. E.Z., H.A., and F.N. contributed to the conceptualization, methodology, and development of the manuscript. N.M. and N.S. contributed to the conceptualization and critical review of the manuscript. M.R.K. was responsible for the study design and also prepared and revised the manuscript. The final version of the manuscript was approved by all of the authors.

Corresponding author

Correspondence to Mohammad Reza Khazdair.

Ethics declarations

Ethics approval and consent to participate

The Declaration of Helsinki was considered in this study, and written informed consent was obtained from all patients. The informed consent from patients with No formal education was obtained from their family (their wife/husband, daughter or sons). They read, signed the consent forms, and stamped them with their finger. The study protocol was approved by the Ethics Committee of Isfahan University of Medical Sciences (IR.MUI.REC.1396.2.055).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

The original version of this article was revised: The typo in the affiliation number ‘1’ has been corrected.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit The Creative Commons Public Domain Dedication waiver ( applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Azdaki, N., Salmani, F., Kazemi, T. et al. Which risk factor best predicts coronary artery disease using artificial neural network method?. BMC Med Inform Decis Mak 24, 52 (2024).

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: