Using Data Mining process to investigate the predictors for early discharge planning for patients in stroke ward.

This abstract has open access
Abstract Summary
Abstract ID :
Submission Type
Presentation Upload :
If the file does not load, click here to open/download the file.
Authors (including presenting author) :
Chan HY(1), So YNL(1), So CT(1), Li WW(1), Chan KH(1), So YC(1), Yu TT(1), Cheng WCS(1)
Affiliation :
(1)Occupational Therapy Department, Princess Margaret Hospital
Introduction :
The prevalence rate of stroke rapidly increases due to aging population and stroke is one of the costly diseases resulting in prolonged hospitalization. Accurate and early prediction of the discharge destination could facilitate early discharge.
Objectives :
To investigate and identify potential predictors for early discharge based on data mining procedures.
Methodology :
This was a 1-year (15/9/2017 – 22/12/2018) retrospective cohort study using Data Mining analysis. This study recruited 302 subjects. All variables were taken on admission. Independent factors were age, gender, scorings of Montreal Cognitive Assessment HK version (HK-MoCA), Modified Barthel Index (MBI) and the Functional Test for the Hemiplegic Upper Extremity (FTHUE). The data mining procedures - Auto Clustering and Classification & Regression (C&R) tree analysis were utilized.
Result & Outcome :
Analysis by Data Mining: Auto-clustering, Result 1: By using ‘Auto Cluster’, 2 groups were clustered and 4 important factors were identified; C&R tree analysis, Result 2: The characteristics of majority group for discharging home were MBI > 42. Result 3: The characteristics of majority group discharging to Old Age Home were MBI=< 42, initial HK-MoCA between 4 to 19, and initial FTHUE< 5. Result 4: The characteristics of minority group discharging to Old Age Home were MBI=< 42, initial HK-MoCA < 4 and age>=63.5. The result identified a few conditions for accurate discharge planning during the early admission phase, in which the ADL independence, cognitive level, hand function level as well as age were good predictors for discharge destination. The data mining analysis simplified the process in stratifying the predictors especially in finding out the cut-off scorings. The study can be further improved by recruiting more patients to identify a more accurate reference of the predictors of discharge destination for facilitating early discharge.

Similar Abstracts by Type

Abstract ID
Abstract Title
Abstract Topic
Submission Type
Primary Author
Clinical Safety and Quality Service I
HA Staff
Maria SINN Dr
Enhancing Partnership with Patients and Community
HA Staff
Donna TSE
Enhancing Partnership with Patients and Community
HA Staff
Clinical Safety and Quality Service II
HA Staff