Using Support Vector Machine "SVM" and Autoregressive Integrated Moving Average "ARIMA" to predict number of males and females who will be have Stroke at European Gaza Hospital

Khaled I.A. Almghari, Adnan Elsibakhi


Stroke is among the world leading cause of morbidity and mortality and the principal cause of long-term disability. Stroke patients need to stay for long time at hospitals for treatment and rehabilitation. Hospitals should be prepared for receiving stroke patients. Stroke forecast prediction can contribute to better preparedness of hospitals in treatment of stroke which lead to improving the functional outcome of the patient, thereby decreasing the national expenses in health.

Gaza strip have a special case due to clashes escalation between Palestinian protesters and Israeli troops at the border of Gaza, because of that, the Ministry of Health at Gaza needs to temporary release patients from hospitals on Thursday weekly to  receive hundreds of wounded people on Friday.

We used two methods of time series forecast; ARIMA and Support vector machine. Real data of stroke patient from European Gaza Hospital in Khan Younis city from 1/1/2012 to 31/03/2019 were used to examine the forecasting accuracy of the models, our sample included all the patient hospitalized at hospital through the study period, which consists of (1141 patients) 521 Males and 620 females.

The main findings are SVM performs better than ARIMA method, therefore we used SVM to predict number of patients who will have stroke next nine months.

 Keywords: Stroke, ARIMA, Support Vector Machine (svm), MASE, MPE, RMSE, KPSS,AIC

النص الكامل:


المراجع العائدة

  • لا توجد روابط عائدة حالياً.

شبكة المؤتمرات العربية © 2019