Hastane Aciline Gelenlerde Kalp Krizi Risk Olasılıg?ının Belirlenmesi ic?in Bir Karar Destek Sisteminin Olus?turulması

Abstract

Bilindig?i üzere akut kalp krizinden o?lümlerin c?og?u, medikal tedaviye bas?lanmadan gelis?se de hastane ic?i mortalite 50'li yıllarda %30-35 oranında iken günümüzde akut kalp krizinin hastane mortalitesi %5-7 civarına indig?i go?rülmektedir. Bunun bas?lıca nedeni biyokimyasal belirtec? kontrollerinin gelis?en teknoloji ile artması ve tanı sistemlerinin gelis?mesidir. Günümüzde pek c?ok hastalıg?ın referans deg?erlerinin yıllara ya da farklı fakto?rlere go?re deg?is?tig?i bilinmektedir. Ayrıca hasta s?ika?yetleri ile bulguların kars?ılas?tırmalı bir c?alıs?ması yapılmadıg?ından yapay zeka? c?ag?ına gec?ilen günümüzde hastaların acil servislerde belirttig?i s?ika?yetlerin standardize edilerek yazılımsal bir program olus?turulmasının acil servis hekimlerine fayda sag?layacag?ı düs?ünülmektedir. Bu amac?la bu c?alıs?mada,Bezmialem Vakıf U?niversitesi Tıp Fakültesi Hastanesi Acil Servis birimine farklı s?ika?yetlerle bas?vuran hastalarda kalp krizi risk fakto?rleri istatistiksel olarak deg?erlendirilmis?tir. Mayıs– Ekim 2016 tarihleri arasında 6607 hasta verisinin analizi kan tahlilleri, s?ika?yet ve muayene bulguları olarak 3 as?amada düzenlenmis?tir. Bu veriler SPSS 24.0 paket programıyla ki-kare bag?ımsızlık testi ve karar ag?acı modellemesi ile istatistiksel olarak deg?erlendirilmis?tir. Hastaların o?n muayenede belirttig?i s?ika?yetler ile kalp krizi ve cinsiyete go?re ki-kare bag?ımsızlık testi uygulanmıs?tır. Buna go?re; kalp krizi ile go?g?üs ag?rısı, c?arpıntı, terleme ve nefes darlıg?ı s?ika?yetlerinin bag?ımlı oldug?u belirlenirken; cinsiyete go?re de karın ag?rısı, c?arpıntı, sırt ag?rısı, bas? do?nmesi, halsizlik, nefes darlıg?ı, bulantı ve kusma s?ika?yetlerinin bag?ımlı oldug?u go?rülmüs?tür. Karar ag?acı modellemesinde ise, modele bag?ımlı deg?is?ken sadece kalp krizi olarak sec?ilirken bag?ımsız deg?is?kenler olarak yas?, cinsiyet, kan tahlili sonuc?ları, hasta s?ika?yet ve muayene bulguları eklenerek 4 farklı modelde incelenmis?tir. Bu incelemelerin sonucunda yas?, c?arpıntı, terleme, cinsiyet, go?g?üs ag?rısı, diyastolik arter basıncı ve Troponin I’nın o?nemli oldug?u saptanmıs?tır. Bo?ylece elde edilen sonuc?lar deg?erlendirildig?inde acile gelenlerde kalp krizi risk olasılıg?ının belirlenmesi ic?in 4 farklı karar destek sistemi olus?turulmus?tur. Buna go?re hekimler acile gelenlerdeki ilk bulgulara go?re kalp krizi gec?irme riskini karar destek sistemleri vasıtasıyla belirleyebileceklerdir.As it is known, although most of the deaths from acute heart attack occur before medical treatment is started, whilein-hospital mortality was 30-35% in the 1950s, it is seen that hospital mortality of acute heart attack has decreasedto around 5-7% today. The main reason for this is the increase in biochemical marker controls with the developing technology and the development of diagnostic systems. Today, it is known that the reference values of manydiseases change according to years or different factors. In addition, since there is not a comparative study of patient complaints and findings, it is thought that the creation of a software program by standardizing the complaints ofpatients in the emergency services in the era of artificial intelligence will benefit emergency physicians. For thispurpose, in this study, heart attack risk factors were statistically evaluated in patients who applied to the EmergencyDepartment of Bezmialem Vakıf University Medical Faculty Hospital with different complaints. The analysis of6607 patient data between May and October 2016 was organized in 3 stages as blood tests, complaints, andexamination findings. The analysis of 6607 patient data between May and October 2016 was organized in 3 stagesas blood tests, complaints, and examination findings. These data were statistically evaluated by chi squareindependence test and decision tree modeling with SPSS 24.0 package program. Chi- square independence test was applied according to the complaints of the patients in the preliminary examination, heart attack and gender. According to this, it was determined that heart attack and chest pain, palpitation, sweating and shortness of breath were dependent; It was observed that the complaints of abdominal pain, palpitation, back pain, dizziness, weakness, shortness of breath, nausea and vomiting were dependent on gender. In the decision tree modeling examined in 4 different models, while the dependent variable was selected as the only heart attack; the independent variables were added as age, gender, blood test results, patient complaints and medical examination findings. As a result of these examinations, age, palpitation, sweating, gender, chest pain, diastolic arterial pressure and Troponin I were found to be important. Thus, when the obtained results were evaluated, 4 different decision support systems were created to determine the risk of heart attack in those who came to the emergency services. Accordingly, physicians will be able to determine the risk of having a heart attack by means of decision support systems according to the first findings of those who come to the emergency services.