Page 124 - FULL REPORT 30012024

Table 4.6 Result of Metrics Evaluation for the Trained Model.

Metrics Result Description
Accuracy 80% Indicates that the model correctly predicted the outcome 80% of
the time.
Precision 75% Out of all the positive stroke predictions made by the model, 75%
were actually correct.

Recall 90% This metric signifies that the model successfully identified 90% of
all actual positive cases.
F1 Score 82% The F1 score is the harmonic mean of precision and recall,
providing a single score that balances both the false positives and
false negatives.

Ultimately, the trained model and the label encoders associated with it are

serialised into joblib files. This phase is critical for preserving the model's state
and allowing for further deployment or additional assessment without the need

for retraining.

4.3.3 Prototype Development

The web development chapter delineates the comprehensive process
undertaken to construct the web application. Utilising Flask, a micro-

framework in Python, the development focused on creating a lightweight yet
robust platform capable of handling various healthcare data interactions.

Moreover, both the prediction module and dashboard module have been
integrated into the prototype.

i. Flask

The web development of this web application was executed using Flask, a

Python micro-framework. The application utilises Flask's routing features
to handle several endpoints, enabling user interactions such as submitting

risk predictions and doing administrative tasks. The system incorporates
integration with a MySQL database to provide long-term data storage,

107