Please use this identifier to cite or link to this item: http://dspace.iua.edu.sd/handle/123456789/5736
Title: Disease prediction using support vector machines Covid-19 case study = التنبؤ بالامرض بإستخدام آلة المتجهات الداعمة دراسة حالة علي مرض كوفيد -19
Authors: إسراء محمد احمد محمد خوجلي
Keywords: الحاسوب
هندسة البرمجيات
Issue Date: 2021
Publisher: جامعة إفريقيا العالمية
Citation: جامعة افريقيا العالمية ـ عمادة الدراسات العليا والبحث العلمي والنشر ـ كلية إقرا لدرسات الحاسوب ـ قسم علوم الحاسوب
Abstract: The number of people infected with the Corona virus pandemic in various regions of the world has exceeded four million and 636 thousand, and the disease has eliminated approximately 312,000, while about one million and 700 thousand infected have recovered. Signs and symptoms of the Corona virus disease may appear after two to 14 days of exposure. Although symptoms for most people with COVID-19 range from mild to moderate, the disease can cause severe medical complications and be fatal for some people. Older people or those who already have health problems are more likely to become seriously ill when infected with COVID-19. A disease prediction model has been developed that combines several features to estimate the risk of infection. The goal is to assist the medical staff in triaging patients, especially in the context of limited healthcare resources (ie, screening work is not available). We created a machine-learning approach trained on records from 599 individuals, that predicts COVID-19 test results with high accuracy using only eight binary traits: gender, age >60, is there known contact with an infected individual, and five primary clinical symptoms are fever. Shortness of breath, headache, sore throat, and cough. In this study, the SVM model was used, and an accuracy of 82.30%, 76.82% precision and 95.61% recall were obtained.
URI: http://dspace.iua.edu.sd/handle/123456789/5736
Appears in Collections:أطروحات الماجستير

Files in This Item:
File Description SizeFormat 
Research.pdf1.88 MBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.