SMART MEDICAL MONITORING SYSTEMS: A realtime classification approach based on Support Vector Machine and concept drift for data imbalance

الموضوعات

إدارة

الملخص

Critical care within hospital is facing one of the major problems that is persisting even with the introduction of information technology devices based on healthcare. This issue is known by the name of alarm fatigue, reflecting the high rate of falsealarms in patient monitoring. One of the main causes of such problem is due tothe imbalance of the monitored patient data. The objective in this work is to ad-dress this issue through the use of an anomaly detection method, namely one-class classification with one-class Support Vector Machine (OCSVM) for monitoring patient data, and the utilization of the Stochastic Gradient Descent (SGD) methodfor model update within the stream of data coming over time. Even though thegoal is to reduce false alarms by not missing true positives as well as not generat-ing a single false positive, patient data might be subject to a change of behavior. In that case, concept drift is involved in this problem which also calls for a model update by detecting data drift whose samples must be distinguished from a usual outlier. To do this adaptive windowing (ADWIN) is included within the stream learning process allowing the final solution to correctly update its model. Thisfinal proposition detects anomalies within a stream of patient data but also can differentiate between outliers and normal data drifts.

روابط وملفات

التعريف والنوع

رقم الوثيقة

040bac1a-6d2a-44c5-8455-cf477d4088c0

رقم العقد

0

نوع الوسائط

Crawler

نوع المحتوى

الرسائل العلمية

صيغة المصدر

رسائل دكتوراة

نوع الملف

pdf image

أسماء الملفات

040bac1a-6d2a-44c5-8455-cf477d4088c0_1.pdf

بيانات النشر

ألقاب المؤلفين

[{"name_ar":"Fahmi Ben Rejab","title_ar":"اشراف","title_en":"Supervision"},{"name_ar":"Kaouther Nouira","title_ar":"اشراف","title_en":"Supervision"},{"name_ar":"Walid Ksiaa","title_ar":"اعداد","title_en":"Preparation"}]

اللغة

English

المصدر والدورية

اسم المصدر

SMART MEDICAL MONITORING SYSTEMS: A realtime classification approach based on Support Vector Machine and concept drift for data imbalance

المحتوى والصفحات

عدد الصفحات

0

إشراف وإعداد

الإشراف

Fahmi Ben Rejab, Kaouther Nouira

الإعداد

Walid Ksiaa

الاقتباسات الببليوغرافية