Προβλεπτικά μοντέλα χρήσης ιατρικής εφαρμογής
View/
Keywords
Μηχανική μάθηση ; Επιλογή χαρακτηριστικών ; Κατηγοριοποίηση ; Stacked generalization ; Σύστημα έγκαιρης προειδοποίησηςAbstract
The rapid development of technology with the parallel increase in the volume of medical data in the information systems of hospitals, marked the era of Big Data in the field of health. The analysis of medical data is a significant opportunity worldwide for national health systems to reduce costs and at the same time improve healthcare. The utilization of these technologies is done in the context of monitoring health issues, counting health goals and fitness, as well as for recording medical data. In such a context, early detection of users at risk of lower compliance rates and patterns of use of a health monitoring application suggesting a risk of abandonment is an invaluable opportunity to implement tailored intervention strategies aimed at recovering and avoiding abandonment thoughts. This dissertation deals with this year's edition of the global competition (IFMBE Science Challenge 2022), which aims to identify patterns of early dropout in users of an application for mobile intervention called Active and Healthy Aging (AHA). Contestants have access to a database of more than 150 users in Madrid who have experienced the impact of a digital AHA application to improve their quality of life for at least 6 months on the MAHA (Moving Active & Healthy Aging) network. The challenge of the competition is given a window of n = 12 consecutive scheduled moments of data acquisition, to predict the user's compliance during the next 3 scheduled data acquisition. At the experimental stage many different approaches to early dropout prediction were implemented and presented. First in the initial data set and then with a different set of features to choose the best one. Specifically, the current thesis proposes a methodology using the Neighborhood Cleaning Rule (NCR) and a specific classification algorithm with the Stacked Generalization learning method to predict the early abandonment of users of the health monitoring application. The results shown that the proposed algorithm was able to predict the early dropout of users from the application with an accuracy of 97.6%, while at the same time, based on the challenge metric at a rate of 93,4%, makes it reliable enough to be used as an early warning system. At the end of this paper are presented in detail the conclusions that emerged from the research.
Postgraduate Studies Programme
Πληροφοριακά Συστήματα και ΥπηρεσίεςDepartment
Σχολή Τεχνολογιών Πληροφορικής και Επικοινωνιών. Τμήμα Ψηφιακών ΣυστημάτωνNumber of pages
69Language
GreekCollections
Except where otherwise noted, this item's license is described as
Αναφορά Δημιουργού 3.0 Ελλάδα
Αναφορά Δημιουργού 3.0 Ελλάδα