University of Twente Student Theses


Automatic Clinical Deterioration Monitoring using Machine Learning Techniques Post Surgery

Prasad, Pralad (2022) Automatic Clinical Deterioration Monitoring using Machine Learning Techniques Post Surgery.

[img] PDF
Abstract:The detection time of clinical deterioration (clinical adverse events) has long been viewed as a key factor in indicating the response rate of necessary interventions. However, this detection time is impacted by the frequency of nurse assessment and the number of patients in a given ward. To improve assessment time, nurses employ threshold-based track-and-trigger systems to monitor adverse events, but they suffer from high false alarm rates given the heterogeneity of population. It is desired to improve the performance of clinical deterioration monitoring systems. This study aimed to develop a fully automatic machine learning based system to detect clinical adverse events in 60 postoperative patients with short-termed (two minutes per hour) vital sign data from wearable sensors. Our system focused on extracting and highlighting the most important features from the vital sign data, and using these features, perform a comprehensive test of decision support models from the machine learning sphere. This includes models from the classical statistical machine learning, deep learning and time-series classification domains. Finally, the top three model’s performances were compared to existing threshold-based systems. Overall, the best decision support model in the system exhibits a significant boost in performance compared to existing threshold-based systems. It could detect all clinical adverse events ahead of time with an accuracy of 86% and a precision of 42%. The system model also reported an average false positive rate of 15%, almost 67% lesser than existing threshold-based systems
Item Type:Essay (Master)
Ziekenhuis Groep Twente, Almelo, The Netherlands
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:54 computer science
Programme:Embedded Systems MSc (60331)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page