Towards predicting neurological outcomes of ICU patients with aneurysmal subarachnoid hemorrhage: Exploratory data analysis with machine learning approaches
Cederlöf, Antti (2023)
2023
Bioteknologian ja biolääketieteen tekniikan maisteriohjelma - Master's Programme in Biotechnology and Biomedical Engineering
Lääketieteen ja terveysteknologian tiedekunta - Faculty of Medicine and Health Technology
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Hyväksymispäivämäärä
2023-11-14
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tuni-202310249017
https://urn.fi/URN:NBN:fi:tuni-202310249017
Tiivistelmä
Aneurysmal subarachnoid hemorrhage (aSAH) is an uncummon but severe health condition that requires emergency treatment. Patients suffering from the condition have a high risk of developing a delayed stroke called delayed cerebral ischemia (DCI), which is a major cause of poor outcomes for the patients. Both aSAH and DCI have been the topic of several recent studies, but their dynamics are still unknown and there are no robust models for predicting the occurrence of a DCI event or the outcomes of aSAH. The aim of this thesis is to find out more about the conditions of aSAH and DCI, to explore and find biomarkers that are connected to the outcomes of the aSAH patients, and to lay the groundwork towards building a predictive model that could help clinicians for early prediction of the outcomes.
With the large amount of data available, machine learning (ML) has recently been used in many healthcare applications. This study conducts an exploratory data analysis using ML approaches and a database of aSAH patients in an intensive care unit (ICU) including vitals, laboratory values, and neurological markers. The data analysis utilizes visualization and clustering methods to find the connections between the biomarker data and the neurological outcomes of the patients. The variables that evaluate the neurological outcome are the development of DCI events, and a neurological outcome grading of recovery after brain injury.
The results of this study suggest that the features in the ICU database are indicative of the neurological outcomes of the patients. They are not predictive of the outcomes on their own, but their combinations can provide good separation between patients with good and bad recovery. However, with the features and analyses used in this thesis, there were no indications for the possible prediction of DCI.
This exploratory study found possible predictive biomarkers that could be used in early prediction of aSAH patient recovery based on neurological outcomes. Predictors for DCI were not found in this study, so the mission of exploring the causes and indicators of the life-threatening event still continues.
With the large amount of data available, machine learning (ML) has recently been used in many healthcare applications. This study conducts an exploratory data analysis using ML approaches and a database of aSAH patients in an intensive care unit (ICU) including vitals, laboratory values, and neurological markers. The data analysis utilizes visualization and clustering methods to find the connections between the biomarker data and the neurological outcomes of the patients. The variables that evaluate the neurological outcome are the development of DCI events, and a neurological outcome grading of recovery after brain injury.
The results of this study suggest that the features in the ICU database are indicative of the neurological outcomes of the patients. They are not predictive of the outcomes on their own, but their combinations can provide good separation between patients with good and bad recovery. However, with the features and analyses used in this thesis, there were no indications for the possible prediction of DCI.
This exploratory study found possible predictive biomarkers that could be used in early prediction of aSAH patient recovery based on neurological outcomes. Predictors for DCI were not found in this study, so the mission of exploring the causes and indicators of the life-threatening event still continues.