Clinical Decision Support System using Real-Time Data Analysis for a Neonatal Intensive Care Unit

It appears your Web browser is not configured to display PDF files. Download adobe Acrobat or click here to download the PDF file.

Click here to download the PDF file.


Gilchrist, Jeffrey Spencer




There are two main contributions to knowledge presented in this thesis: (1) an efficient Clinical Data Repository (CDR) with a novel data storage format, and (2) new neonatal mortality risk estimation models that use data analysed in real-time from the CDR.

The CDR collects, stores, and retrieves clinical data in real-time. It uses a novel hybrid Entity-Attribute-Value (EAV) storage format that is faster, uses less complex queries, and allows the data to remain in its original data type while using the same amount of storage space as other popular storage formats. The CDR can collect and store data from patient monitors and laboratory results in real-time using only free open-source tools, and is compatible with industry standard protocols. The CDR design accommodates the fast-changing medical domain, and supports the addition of new attributes automatically. Private patient information is segregated automatically from raw research data before it is stored. This provides the ability to give researchers and physicians access to the real-time research data without further processing, and without violating patient privacy.

We developed novel real-time neonatal mortality risk estimation models using Decision Trees (DTs) and Artificial Neural Networks (ANNs) that met criteria for clinically useful results (>60% sensitivity, >90% specificity) for individual patients. Results showed that mortality models using summary data (data from admission until up to the first 48 hours after Neonatal Intensive Care Unit (NICU) admission) provided, on average, the highest sensitivity and specificity with the least number of false positives, exceeding the performance criteria set by our clinical partners mentioned above. The DT model produced the best results on average (sensitivity=75%, specificity=96%), while the AN N model produced lower but still clinically useful results (sensitivity=68%, specificity=97%). Three attributes were found to be most important to estimate the risk of mortality during the first 48 hours after NICU admission: lowest serum pH, lowest blood pressure, and lowest heart rate.

In the future, a prototype of the real-time Clinical Decision Support (CDS) system framework should be implemented with the ability to generate intelligent alerts and warnings from medical events as they occur, such as when the risk estimation for a patient changes significantly.


Biomedical engineering




Carleton University

Thesis Degree Name: 

Doctor of Philosophy: 

Thesis Degree Level: 


Thesis Degree Discipline: 

Engineering, Electrical and Computer

Parent Collection: 

Theses and Dissertations

Items in CURVE are protected by copyright, with all rights reserved, unless otherwise indicated. They are made available with permission from the author(s).