Page 137 - FULL REPORT 30012024

CHAPTER 5

RESULTS AND FINDINGS

This chapter outlines the results and insights from the study which focused on

evaluating both the prediction model and user acceptance of a web application
designed for stroke risk assessment. The evaluation is twofold: firstly, the prediction

model is assessed during its training phase, and secondly, user acceptance is examined

using the Technology Acceptance Model (TAM) through data gathered from an online
survey. The survey attracted participants from a wide range of backgrounds,

encompassing both medical professionals and laypersons, ensuring a well-rounded
perspective on the application's usability, functionality, and overall impact. The

findings are meticulously organized to provide insights into the prediction model's
performance and the user acceptance test that focused on three critical aspects:

Perceived Usefulness (PU), Perceived Ease of Use (PE), and Behavioural Intention to

Use (BI). This chapter aims to critically evaluate the web application, particularly
assessing the efficacy of its dashboard and prediction tool, thereby addressing a

fundamental objective of the project.

5.1 Prediction Model Evaluation

This section outlines the evaluation process of the developed Random Forest
model used for stroke risk prediction. Various metrics and validation methods

were utilized to assess the accuracy and reliability of the model's predictions.

The evaluation was carried out using Python within Visual Studio Code
environment, and the results were displayed in the terminal. This process was

conducted during training the model, ensuring that the predictive outcomes
are both robust and dependable.

120