Decision Support System to Risk Stratification in the Acute Coronary Syndrome Using Fuzzy Logic

Acute coronary syndrome (ACS) is a set of symptoms and signs which define a range of conditions related with the unexpected reduced blood flow to the heart. In ACS, the heart muscles cannot function properly due to the decrease of blood flow. Myocardial infarction (MI) is a condition which comes under the umbrella of acute coronary syndrome. ,e aim of risk stratification (RS) in ACS is to recognize patients at high risk of ischemic events. Yet, no investigative study is available to identify the patients at high risk. ,erefore, to facilitate this process, it would be ideal to have a reliable and trustworthy method by the help of which the doctors can make early and easy decisions for the patient and for detecting the related disease. ,is research used the features of GRACE Score to RS in the ACS and presented decision support system (DSS).,e concept of probabilistic approach has been used as a tool to model the identified features for decision-making (DM).,is technique can be further used for DM purposes to RS in the ACS in healthcare. Furthermore, the result of the proposed method has proved closer and more reliable DM of patient and then eventually can be used for advice of medicine and rest accordingly by the doctors.


Introduction
e emergency department is mostly facing the patient of chest pain. One of the main reasons for chest pain is the ACS which occurs in both short-term and long-term adverse prognosis. ACS is a set of symptoms and signs which describe conditions range connected with reduced flow of blood to heart. In ACS, heart muscles are unable to function properly due to decrease in blood flow. e aim of RS in ACS is to identify the patients at high risk of ischemic events.
According to the report of WHO [1], the cardiovascular disease is one of the top causes of death. In the admissions of hospital, the patients of acute chest pain are a big burden.
is range is presented as the number present of 40% of acute medical admissions and 5% of all emergency [2]. On time and efficient treatment is a necessary part to ensure the cost-effective and precise management of the patient. Meanwhile, the poor identification and treatments are linked with increased mortality, morbidity, and cost [3]. e identification of ACS relies on account of ischemic symptoms, such as the ischemic ECG changes, chest pain, and elevated cardiac biomarkers. ere are diverse causes of ACS and the identification of MI is categorized based on the causes [4]. ACS is a multifactorial disease which includes the genetic factor caused by coronary artery obstruction by atheroma. It is influenced by other factors of risk such as diabetes, hypertension, obesity, smoking, and dyslipidemia. Among these, the hypertension and dyslipidemia are considered as the most common factors in the ACS patients, with a suggestion that preventing these factors can put in a decrease in the process of atherosclerosis. Diagnosis of patients with high risk soon after ACS has been a challenging issue for the clinicians for a few decades. Several studies exist which show that more rigorous management of the patient results in considerably good outcomes in the adverse cardiac events. e diagnosis of severity risk of ACS patient is significant, as the patient can get benefit with early and proper treatment.
Lin et al. [5] presented a study containing a dataset of 22 million enrollees in Taiwan from 1999 to 2010. e study has acknowledged 27948 patients with CS and sex, age, and comorbidity matched control. Multivariate regression analysis after amendment for probable cardiovascular confounders and odd ratios was calculated with 95% confidence interval to measure the relationship between CS and ACS. e aim of the ACS risk score is to diagnose the patient with high risk who can be managed with optimum treatment in short time. e patients of ST elevation MI and non-ST elevation MI have been subjected to the RS studies and several risk scores have been proposed. Some of the scores were tested in clinical trial populations, while others were derived from large registries.
Currently, no exploratory study is available to recognize the patients with high risk. To assist this progression, it would be ultimate to have a reliable and trustworthy method through which the practitioners can make early and easy decision for the patient. e proposed study's contribution is to use the features of GRACE Score to RS in the ACS for DM. A probabilistic approach of fuzzy tool has been used to model the feature for DM. e technique can be further used for DM purpose to RS in the ACS in healthcare. Fuzzy set is an effectual tool to convey uncertain and imperfect cognitive information with the membership, nonmembership, and hesitance degrees [6]. e fuzzy logic has a wide range of applications in various areas of research [7]. Technology acceptance model was presented through fuzzy multicriteria DM approach [8].
e study has analysed 136 research articles regarding the software development organization for cloud based technology implementation. is paper is organised as follows: Section 2 describes works related to RS in the patients of ACS. Section 3 describes the methodology used for the planned study. Section 4 depicts the results and discussions, while the paper concludes with Section 5.

Related Work
Early DM of patient has immense prospective for refining the value of care, decreasing the care cost, and dropping error and waste. Several studies were related to RS of patient for ACS. Singh and Guttag [9] presented an algorithm to hold class imbalance in medical datasets and assessed the nonsymmetric based classification tree algorithm performance. e performance of the said algorithm is evaluated and compared with those of SVM based classifiers applied to 4219 non-ST elevation ACS patients. ree strategies were used for management class imbalance. ese strategies were synthetic minority oversampling, cost-sensitive SVM learning, and random majority undersampling. Khalil et al. [10] assessed the validity and accuracy of Canadian-ACS risk score as a prognostic score for the patient in hospital with RS with ACS as compared to TIMI and GRACE risk scores in patients of Egypt. e authors of [11] assessed the clinical utility of H-FABP for suspected ACS. e H-FABP demonstrated the added value to cardiac troponin in the initial stage after onset of symptoms. e H-FABP also detected the patients at enlarged risk for upcoming actions of cardiac diseases. J. Přeček et al. [12] demonstrated a summary of patients with ACS of prognostic stratification according to baseline levels of renal function biomarkers. e authors of [13] provided a model with interpretable representation of long-term ECG recording and discovered the associations among all short heart rate patterns across the population of patients. e experiment was performed on 4557 patients with non-ST elevation ACS and confirmed that heart rate topic models considerably get better RS.
Macdonald et al. [14] compared the two clinical scoring methods of RS of suspected ACS in the emergency department. Cullen et al. [15] performed a diagnostic study in the emergency department for the patient with symptoms of ACS with two-hour sensitive Troponin Assay Results. e results were elaborated using standard measures of test accuracy. e authors of [16] reviewed up-to-date methods in the emergency care for the identification and RS of NSTE-ACS. Further, the authors highlighted the limitations of the existing approaches and suggested the improvements in them. An approach was presented which quantified the level to which subtle ECG morphology changes from beat to beat. e time and shape characteristics were considered as morphological variability to measure the changes in these variables in the cardiac events in sequential pairs of heart beats. e proposed method tested 400 patients of ACS [17]. e research explored the challenges that manipulate the dietary choices from patient attending a secondary prevention clinic following MI [18].
Dellas et al. [19] studied the predictive performance of Heart-type Fatty Acid Binding Protein and integrated it into the ESC algorithm. A subanalysis study was presented for the disease of diabetes and peripheral arterial disease in the patient of RS admitted after ACS [20]. Hill et al. [21] presented a roadmap for computational cardiology. e approach consisted of three stages: (i) development of baseline model, (ii) disease simulation, and (iii) translation. e authors of [22] presented that there were some barriers in the RS appropriate implementation with ACS, so they proposed developing a simple and early score for RS of the ACS patient. e study was conducted to diagnose the risk factors for the ACS for the age below 40 years in the Slemani Cardiac Hospital, Sulaimaniyah, Iraq [23]. Pezawas et al. [24] presented a study on the risk of arrhythmic death in ischemic heart disease. e study was conducted for prospective, controlled, observer-blind RS for 10 years. Lin et al. [5] presented a study for the quantification of the relationship between cervical spondylosis and ACS. Details of some of the existing approaches are shown in Table 1. ese approaches along with the publication year are given.

Methodology
ACS is a set of symptoms and signs which describe variation of conditions associated with the sudden, reduced blood flow to the heart. In ACS, the muscles of the heart are unable to function appropriately due to the reduction of blood flow. MI is a situation under the umbrella of ACS. e proposed research work uses the features of GRACE Score to RS in the 2 Scientific Programming ACS. e concept of probabilistic approach (fuzzy tool) has been used to model the identified feature for DM. e following subsections briefly describe the details of the proposed method.

RS in the ACS.
e aim of RS in ACS is to identify the patients at high risk of ischemic events. Early DM in healthcare system has massive potential for refining the quality of care and reducing the cost of care, waste, and error. e risk score of Global Registry of Acute Coronary Events (GRACE) is a tool of validated RS which has incremental predictive value for RS compared with the evaluation of clinical testing [16]. Some countries have partial implementation of GRACE Score in their emergency medicine. e age factor formulates an exponential input to the GRACE Score. e addition of new cardiac biomarkers into this identification pathway can allocate to early treatment stratification. e advantages of the strong GRACE Score [30,31] are based on large registry, but the disadvantage lies in including variables at the time of existing admission (laboratory results), so it is not applicable for instant stratification. e related studies to RS are discussed in Section 2 of the paper.

Fuzzy Logic.
e fuzzy logic (FL) is a mathematical tool used to resolve circumstances of vagueness and uncertainty. It was developed by Zadeh in 1965 [32,33]. FL converts the specific analysis, deals with vague information, and offers a best decision in the presence of the vague and incomplete data. Some of the applications of FL are in control system, car transmission system, washing machines, vacuum cleaner, and software system [7,[34][35][36][37]. Details concerning fuzzy idea can be found in the study in [32]. Messaoud et al. [38] used fuzzy logic for the application of cognitively stimulated use of speech information and to enhance microphone system. Liu and Zhang [39] integrated the picture fuzzy set and the linguistic term; the study has defined the picture fuzzy linguistic set (PFLS) and operations of picture fuzzy linguistic numbers for developing the Archimedean picture fuzzy linguistic weighted arithmetic averaging operator and presented various properties and special cases of this operator. ey also proposed a method which deals with problem of multiattribute group DM on the basis of the developed A-PFLWAA operator. e reason behind the fuzzy logic used in the proposed is that it works in the situation of uncertainty and vagueness, while the other methods like AHP and ANP work in complex situations.

DSS for RS in ACS.
Fuzzy logic is used as a DSS for the RS of the ACS. Liu et al. [40] extended the ensemble-based scoring system by using a fast extreme learning machine algorithm for neural networks. e patients of emergency department of Singapore General Hospital were examined and the features of 12-lead ECG parameters, heart rate variability, and vital signs were extracted. An ensemblebased scoring system-extreme learning machine (ESS-ELM) algorithm was proposed for the prediction of cardiac events. e features of GRACE Score are considered to model the proposed DSS. Figure 1 shows the European Society of Cardiology (ESC) rapid rule-out of ACS with hs-cTn algorithm [41]. Rapid rule-out of ACS with high-sensitivity troponin. Grace � Global Registry of Acute Coronary Events; ULN � upper limit of normal, 99th percentile of  [15] healthy controls; hsTn � high-sensitivity troponin. Figure 1 shows the graphical representation of heart diseases and their symptoms. e GRACE 1.0 Score is broadly certified [41]. is score diagnoses the patient in hospital as a 6-month risk of death by using the following factors: heart rate, age, Killip class, systolic BP, creatinine, elevated cTn, ST-segment deviation, and cardiac arrest.
e new GRACE 2.0 Score revised GRACE algorithms for diagnosis of death at 1 and 3 years and is termed at medical device (available at http://www. gracescore.org/WebSite/WebVersion.aspx). Figure 2 shows the GRACE score, which consists of the factors age, heart rate, systolic BP, Killip class, and creatinine [16]. Figure 3 shows the generic process of fuzzification and defuzzification modeling for the proposed study of RS for ACS. e figure consists of fuzzification, database, rule base, defuzzification, and output. Once the rules are designed based on the membership function, they can be stored in database provided by the fuzzy inference system. After that, the inputs can be passed from the designed model and the output will be displayed.
Membership functions are designed individually for each input based on the score provided in the GRACE Score table. e MFs are the degree of representations of each input. e starting and ending boundaries of a particular input can be shown. All the inputs are categorized into four membership functions. e reason behind this is that some of the inputs have fewer features, while some of them are having more detailed features. So, to make uniformity in the model, all inputs were equally categorized into four membership functions. ese categories show the level and severity of a specific input. Figure 4 shows the memberships functions designed for the input age. e rest of the membership functions for the other inputs can be drawn the same as Figure 4. e same can be plotted in the same way for the remaining membership functions.  (1) . . . and so on e proposed system is based on five inputs (age, Killip class, heart rate, systolic BP, and creatinine); rules and their output are shown in Figure 5. is system was designed in Matlab to deal with the situation of uncertainty and vagueness arising for the RS in the ACS. In this system, initially the membership functions are defined followed by the rules development from the MF, and the rules are combined to form the model. Inputs can be passed through the model and resulted output will be used to facilitate patient accordingly. Figure 6 depicts the nomenclature of the proposed inputs, membership functions, rules, and output. e proposed system is generated in the fuzzy toolbox by using Matlab. In the figure, the initial black circles show the inputs that are the age, heart rate, systolic BP, Killip class, and creatinine, the second blank circles show the membership functions defined for the rules modeling, and the light-blue circles show the rules generated. Finally the output is shown in the figure and highlighted in black circle. Figure 7 shows the graphical representation of the rules view obtained from membership function to design the rules and fuzzy inference system.

Results and Discussions
Once the model is designed from the membership functions for the variables of GRACE Score consisting of age, heart rate, systolic BP, Killip class, and creatinine. e fuzzy inference system of membership functions and rules are obtained. Different inputs of the same format are passed through the designed model to check the RS of the ACS. e range of the input and output values is between 0 and 1, where 0 shows the lowest range, while 1 shows the highest    Other approaches are available as mentioned in Section 2 and Table 2, which measure the RS in ACS from different perspectives.
None of the approaches mentioned in this paper calculate the RS in ACS based on the defined criteria in complex and uncertain situation. e proposed approach can easily and effectively measure the RS in ACS. Table 3 shows the inputs that passed through the model and their output. In

Conclusion
ACS consists of signs and symptoms to define a range of situations linked with the sudden, reduced blood flow to the heart. e heart muscles are unable to function properly due to the reduction of blood flow in the ACS. e RS in ACS aims to recognize the patients at high risk of ischemic events. On time and initial DM in healthcare has enormous prospective for enlightening the quality of care, reducing the

Reference
Method Year [42] Biomarker algorithm encompassing the clinical chemistry score 2020 [43] High-sensitivity cardiac troponin T in patients with suspected ACS 2020 [44] Comorbidity assessment for mortality RS 2019 [45] RS in ACS 2018 [46] Role of cardiovascular reactions to mental stress challenge [43] RS through high-sensitivity cardiac troponin T in patients with suspected ACS 2020 Our Fuzzy logic based decision for RS in ACS 2021 Scientific Programming cost of care, and reducing waste and error. To assist the said process, the proposed research work uses the features of GRACE Score to RS in the ACS. It will be a reliable and trustworthy method by the help of which the doctor can make early and easy DM for the patient and for detecting related disease. FL approach has been used as a tool to model the feature of GRACE score for DM in the ACS. e result of the proposed method has proved more close and authentic DM of the patient and can be used for advice of medicine and rest accordingly by the doctors. In the future, the proposed research can be explored to a more reliable and efficient system with large sample size to show the effectiveness and usefulness of the system for enhancement in healthcare. is will be further an enhanced step toward improvement in healthcare.

Data Availability
No data are available.

Conflicts of Interest
e authors declare no conflicts of interest.