Generalising Health Events by Using Frequent Itemset Mining
Name
Oliver-Erik Suik
Abstract
The digitisation of health data has enabled us to conduct studies that enhance healthcare practices and make clinical processes more efficient. However, the diverse types of health data and their sparse nature create challenges in understanding a patient's health status and utilising it in data mining tasks and analytics techniques. The primary objective of this research is to use frequent itemset mining to generalise similar health events into a higher-level event and assess its practicality and limitations. The study involves extracting health event transactions from Estonian healthcare data using a sliding window technique and applying the FP-Max algorithm to identify frequent itemsets of health concepts. These itemsets are clustered into higher-level events, providing a generalised representation of a patient's health event timeline. Experimenting with different parameters resulted in clusters with varying levels of generalisation which ultimately helped to describe a patient's health status by reducing elements and giving them generalised labels.
Graduation Thesis language
English
Graduation Thesis type
Master - Software Engineering
Supervisor(s)
Sulev Reisberg
Defence year
2024