Kalp, insan vücudunda kritik kompleks görevlerde yer alan hayati bir organdır. Dolayısıyla kalp rahatsızlıkları hayati riskler oluşturmaktadır. Kalıtsal ve çevresel kalp hastalıkları durumunda tanımlanamayan ani kardiyak ölümler artmakta ve kategorileştirilememektedir. Bu bağlamda kalp hastalıklarının takibi ve kalp hastalıklarının birbiriyle ilişkilendirilmesi önem taşımaktadır. Bu çalışmada kalp hastalıkları açısında günümüzde en popüler olan birbiri ile ilişkili 18 parametre (kolesterol, hipertansiyon, ateroskleroz/damar tıkanıklığı, kardiyomiyopati, atriyal fibrilasyon, yaş, diyabet, obezite, sigara tüketimi, alkol tüketimi, stres yoğunluğu ve yaşam temposu, fibrozis/pulmoner fibrozis, ventrikül duvar kalınlığı, Galaktin-3 seviyesi, miyokard iskemi, miyokard enfarktüsü geçmişi, taşikardi, PQRST diyagramı, COVİD öyküsü) kendi içinde gruplandırılıp sınıflandırılarak bulanık sistem tabanlı ani kalp krizi skor değerlendirmesi yapmaktadır.. Çalışmada bulanık mantığın kullanılma sebebi, bu sistemin çalışma mekanizmasına uygun olan ve kesin sonuç vermeyen yaklaşımsal üyelik fonksiyonlarını değerlendirmesidir. Bulanık mantığa göre üyelik derecesi 1'e eşit ise üyelik fonsiyonu bulanık kümeye tamamen ait, 0 ise bulanık kümeye ait değildir. Yapılan çalışmada kalp hastalığı verilerinin klasik Aristo yöntemi olan kesin karar mekanizmasına uygun olmadığından bulanık mantık yöntemi tercih edilmiştir. Aristo mantığında kesin sonuçlar veren çıkış fonksiyonlarını 1-0, iyi-kötü, beyaz-siyah, var-yok, evet-hayır değerlendirmelerinde bulunurken, bulanık mantık yaklaşımı Aristo mantığının aksine kesin olmayan sonuçların koalisyona dahil olmasına olanak verir. Bulanık mantık ve çalışmada kullanılan ve seçilen parametreler yüksek riskli kalp hastalıkları olan kolesterol, hipertansiyon, ateroskleroz/damar tıkanıklığı, kardiyomiyopati, atriyal fibrilasyon, yaş, diyabet, obezite, cinsiyet, alkol tüketimi, stres yoğunluğu ve yaşam temposu, fibrozis/pulmoner fibrozis,ventrikül duvar kalınlığı, Galaktin-3 seviyesi, miyokard iskemi, miyokard enfarktüsü geçmişi, taşikardi gibi teşhisler doğrultusunda olması gereken insan kalbi testlerine ve hastanın kalıtsal özelliklerine göre ani kardiyak ölüm oranlarını öngörmek amacıyla tasarlanan sistemin verimli uygulaması olarak görülmüştür. Bu tez çalışmasında kullanılan 18 parametre giriş üyelik fonksiyonlarını ifade ederken, Ani Kardiyak Ölüm (AKÖ) risk grupları ise çıkış fonksiyonu olarak değerlendirilmiştir. Çalışma sonuçları teşhis olarak değerlendirilmeyip, alınan teşhislerin birbiri ile ilişkilerini öngörmektedir. Tüm üyelik fonksiyonlarının dahil edildiği Fuzzy System (FS) sistemde oluşturulan kurallar, sisteme dahil edilen verilerle beklenen yüksek oranda doğruluk değerini sağlamıştır.
The heart is a vital organ involved in critically complex tasks in the human body. Therefore, heart diseases pose life-threatening risks. In the case of hereditary and disabling heart diseases, unidentified sudden cardiac deaths increase and are not categorized. In this context, it is important to explain the relationship between the follow-up of heart diseases and heart diseases. In the case of hereditary and environmental heart diseases, unidentified sudden cardiac deaths increase and cannot be categorized. In this context, it is important to monitor heart diseases and correlate heart diseases with each other. In addition to medical medical research, issues such as the development of intelligent computer systems and their support with biomedical studies, the continuity of follow-up-treatment of diseases, early diagnosis and prevention of diseases are the subject of many studies. In this study, 18 interrelated parameters (cholesterol, atherosclerosis, cardiomyopathy, atrial fibrillation, age, diabetes, obesity, alcohol consumption, stress intensity and lifestyle, fibrosis/pulmonary fibrosis, ventricular wall thickness, Galactin-3 level, myocardial ischemia, myocardial infraction history, tachycardia, PQRST diaphragm, COVİD history), which are the most popular in terms of heart diseases today, were grouped and classified within themselves and classified the estimation of sudden cardiac death based on the fuzzy system. The reason for using the fuzzy logic in the work is to allow the storage of inconclusive approximate functions that are suitable for the workspaces of the execution logic. If the membership degree is equal to 1, the member function fully belongs to the fuzzy cluster, if 0, it does not belong to the fuzzy cluster. In the study, heart disease data is not suitable for the definitive decision mechanism, which is the classical Aristotelian method. While the output functions that give definite results in Aristotelian logic are evaluated as 1-0 good-bad, white-black, yes-no, yes-no, fuzzy logic approach allows inconclusive results to be included in the coalition, unlike Aristotelian logic. Fuzzy logic with the parameters used and selected in the study is seen as the most efficient application of the system designed to predict sudden cardiac death rates according to human heart tests, which should be compatible with diagnoses such as high-risk heart diseases. such as cholesterol, atherosclerosis, cardiomyopathy, atrial fibrillation, age, diabetes, obesity, alcohol consumption, stress intensity and lifestyle, fibrosis/pulmonary fibrosis, ventricular wall thickness, Galactin-3 level, myocardial ischemia, myocardial infraction history, tachycardia, PQRST diaphragm, COVID history.While 18 parameters used in this thesis represent input membership functions, SCD risk groups are evaluated as output functions. The results of the study are not considered as a diagnosis, but predict the relationships between the diagnoses received. The rules created in the FS system, in which all membership functions are included, provided the expected high accuracy value with the data included in the system. Cholesterol, hypertension, atherosclerosis/vascular occlusion, cardiomyopathy, atrial fibrillation, age, diabetes, obesity, alcohol consumption, cigarette consumption, stress intensity and pace of life, which are evaluated as environmental and hereditary parameters for high-risk heart diseases based on the rates stated and discovered in the literature, fibrosis/pulmonary fibrosis, ventricular wall thickness, Galactin-3 level, myocardial ischemia, myocardial infarction history, tachycardia, PQRST diagram problems, COVID history are grouped and classified within itself to determine the prediction of sudden heart attack based on fuzzy system. SCD values grouped as three outputs in the study are evaluated as low, medium and high risk groups together with the interactions of the parameters. The system aimed at predicting SCD based on the results of diagnosing high-risk heart disease, standard human heart tests, and other hereditary problems has been studied. In addition to medical studies on this subject, studies on the diagnosis and treatment of heart diseases continue in the biomedical field. In this study, sudden cardiac deaths, which are aimed to be predicted, are classified by studying the fuzzy logic-based smart system application designed to predict SCD based on the results of the diagnosis of high-risk heart diseases, standard human heart tests and other hereditary problems. In the study, high-risk heart diseases were primarily classified within themselves. It was aimed to express the direct and secondary relationship of the determined heart diseases with each other and the risk levels in terms of sudden cardiac death. Diagnoses obtained during the study were blurred at certain intervals by the parameters given in Chapter 3 by the user, and 448 rules were created. According to the given rules, the relationships of heart diseases identified in Chapter 2 are divided into 3 groups. In the created program, 448 results validating rules were specified and verified. Membership functions used in the study are numbered to simplify the program interface. Number 1 for cholesterol. It is divided into two groups as low and medium. Hypertension number is 2. It is divided into two groups as primary and secondary. Vascular occlusion number is 3. It is blurred in two groups as low and medium. The number of age membership functions is 4 and it has been examined in three groups as low, medium and high. Diabetes membership function number is 5. Pregnancy is divided into three groups as Type-1 and Type-2. The number of obesity membership functions is 6. It was examined in three groups as level-1, level-2 and level-3. Alcohol and cigarette membership function numbers are 7 and 8. The use-case risk domain was analyzed in three groups: never-users, occasional and frequent users. The stress membership function number is 9. It is grouped as Level-1, Level -2, Level 3. Cardiomyopathy membership function number is 10. It is divided into 2 groups as initial and diagnosed cardiomyopathy. Fibrosis membership function number is 11. Graded as pulmonary fibrosis. The ventricular wall thickness membership function number is 12. The thickness value can take values between 6-20 mm. The Galectin-3 membership function number is 13. It is grouped as low, medium and high. Myocardial ischemia membership function number is 14. It is grouped as low and medium. Myocardial infraction history membership function number is 15, Covid membership function number is 16. Yes – Grouped as No. The membership function number of the PQRST diagram is 17. It is grouped as level-1, level -2 and level-3. The tachycardia membership function number is 18. It is grouped as level-1 and level-2. All membership functions were blurred in the specified groups and rules were created. SCD results obtained after clarification were evaluated as low, medium and high risk groups. In the grouping stages of the 18 selected input membership functions, non-real data were approximated. In the study, the 'Mamdani' approach was used in the fuzzification process. In this study, the 'Centroid' method was used in the defuzzification process. . In Figure 4.4, some of the 18 membership functions of the low risk group are expressed according to the results; age 45, alcohol use sometimes (7→2), smoking sometimes (8→4), stress and pace of life level 2 (9→5), no history of myocardial infarction but genetic history (15→3), COVID-19 A result value of 1.85 was obtained for the patient who had no history (16→1), whose ECG signs were considered level 2 in terms of normality (17→3), who had obesity level 1 (6→28), who was diagnosed with diabetes type-1 (5→175). It is classified as a low risk group. According to the result output in Figure 4.5; Cholesterol level 175 mg/dl (1→175), which is considered medium in blurred condition, Hypertension group secondary (2→0.5), 45 years old, diagnosed with type-1 diabetes (5→175), obesity level 2, ventricular The SCD value was calculated as 5 for the patient with a wall thickness of 10 mm, a history of COVID-19 (16→5), and a tachycardia level of 2 (18→5). It is classified as a medium risk group. The system confirms the desired results when the fuzzy groups are clarified.
Except where otherwise noted, this item's license is described as http://creativecommons.org/licenses/by/4.0/
