TU Darmstadt / ULB / TUbiblio

Towards Automatic Classification of Common Therapy Errors for Diabetes Therapy Support

Heuschkel, Jens and Kauschke, Sebastian and Mühlhäuser, Max (2019):
Towards Automatic Classification of Common Therapy Errors for Diabetes Therapy Support.
Waikoloa, USA, In: 2019 IEEE Global Communications Conference: Selected Areas in Communications: E-Health (Globecom2019 SAC EH), Waikoloa, USA, 9-13 December 2019, [Conference or Workshop Item]

Abstract

Today, one in eleven adults is suffering from diabetes mellitus. Diabetes mellitus is a disease where the body's own insulin control system fails. Incorrectly treated diabetes mellitus will lead to serious complications like strokes, blindness, and ultimately, death. Too high or too low blood glucose levels are dangerous, an insulin over-dose can even be lethal. Hence, the correct dosage of insulin from diabetes patients is the key parameter in therapy. Therefore, the patients get educated regularly by diabetes experts. These training sessions contain data review by the experts in order to identify errors in the patients' dosage behavior. However, this review is time consuming, since the error identification for a wrong dosage is nontrivial. In this paper we investigate the automatic classification of insulin dosage into three categories, representing correctly applied therapy and the most common therapy faults. We provide the experts with a pre-classified overview of the data, where the common errors are visually highlighted. This saves time in the consultation hour, enabling the expert to spend more time on investigating the patients individual problems. In our evaluation we compare multiple classification methods based on dynamic time warping against a convolutional neural network. The results show, that the convolutional neural network can achieve accuracy levels that are promising, although further improvements are required.

Item Type: Conference or Workshop Item
Erschienen: 2019
Creators: Heuschkel, Jens and Kauschke, Sebastian and Mühlhäuser, Max
Title: Towards Automatic Classification of Common Therapy Errors for Diabetes Therapy Support
Language: English
Abstract:

Today, one in eleven adults is suffering from diabetes mellitus. Diabetes mellitus is a disease where the body's own insulin control system fails. Incorrectly treated diabetes mellitus will lead to serious complications like strokes, blindness, and ultimately, death. Too high or too low blood glucose levels are dangerous, an insulin over-dose can even be lethal. Hence, the correct dosage of insulin from diabetes patients is the key parameter in therapy. Therefore, the patients get educated regularly by diabetes experts. These training sessions contain data review by the experts in order to identify errors in the patients' dosage behavior. However, this review is time consuming, since the error identification for a wrong dosage is nontrivial. In this paper we investigate the automatic classification of insulin dosage into three categories, representing correctly applied therapy and the most common therapy faults. We provide the experts with a pre-classified overview of the data, where the common errors are visually highlighted. This saves time in the consultation hour, enabling the expert to spend more time on investigating the patients individual problems. In our evaluation we compare multiple classification methods based on dynamic time warping against a convolutional neural network. The results show, that the convolutional neural network can achieve accuracy levels that are promising, although further improvements are required.

Place of Publication: Waikoloa, USA
Uncontrolled Keywords: Diabetes Mellitus; Classification; Convolutional Neural Network; e-Health
Divisions: 20 Department of Computer Science
20 Department of Computer Science > Telecooperation
Event Title: 2019 IEEE Global Communications Conference: Selected Areas in Communications: E-Health (Globecom2019 SAC EH)
Event Location: Waikoloa, USA
Event Dates: 9-13 December 2019
Date Deposited: 22 Jul 2019 07:15
Export:
Suche nach Titel in: TUfind oder in Google

Optionen (nur für Redakteure)

View Item View Item