M.S. AAI Capstone Chronicles 2024

First page Table of contents Next page Last page
Table of Contents3
Spring 20243
Summer 20243
Fall 20243
Background Information77
Data Summary80
Experimental Methods81
Results83
Future Enhancements87
Figure 1184
● 3K Conversations Dataset for ChatBot. (n.d.). Kaggle. Retrieved July 15, 2024, from https://www.kaggle.com/datasets/kreeshrajani/3k-conversations-dataset-for-chatbot204
Deep Learning Image Captioning206
Introduction207
Introduction236
Dataset Summary237
While exploring our chosen datasets we discovered some data issues and had to implement some preprocessing methodologies to prepare our data for training.238
Background Information239
Managing diabetes has been a focus of both academic research and commercial innovation. Numerous efforts have explored predictive models and automated systems for blood glucose management, meal analysis, and long-term glycemic control.240
Experimental Methods242
This project employs three core machine learning models, each tailored to address specific tasks in diabetes management:242
Results and Discussion245
Tidepool Dataset245
250
DiaTrend Dataset250
Figure 6: This scatter plot displays the relationship between actual and predicted HbA1c values for the LSTM model. The red dashed line represents the ideal line of equality, where predictions match the actual values. The concentration of points near the line indicates strong predictive accuracy, while slight deviations, particularly at higher HbA1c ranges, suggest minor variability in the model’s performance for extreme cases. This visualization underscores the model’s capability in accurately predicting HbA1c trends, contributing to improved long-term glycemic control.260

Made with FlippingBook - professional solution for displaying marketing and sales documents online