ADS Capstone Chronicles Revised

37

additional nutritional information for each food item such as glycemic index, which would allow for a more precise calculation of suitability. Future studies that target a similar topic or attempt to expand the work done in this study should aim to increase the amount of diverse, representative data for both patients and food, and gain access to additional nutritional information such as glycemic index or glycemic load. The inclusion of glycemic index and glycemic load will properly represent the ebbs and flows of patient data rather than the strict guidelines used to calculate food and patient scores in this project. Additionally, future studies might have access to more computational power to accommodate the increase in data, which is something this study could not perform due to limited resources. Another area of improvement is the calculation for determining personalized suitability of food items for patients. Due to the limitation of resources for this study, working alongside a healthcare professional to assist in navigating the appropriate metrics for determining suitability was not plausible. As a result, the calculations are done using the best judgement of individuals with no medical experience, which can impact the accuracy and reliability of this model. For future studies, collaboration with a healthcare professional is highly recommended to ensure safe advice. Additionally, the application developed during this study provides an interactive prototype that demonstrates the potential for personalized recommendations. Users input health data such as BMI, glucose values, and food preferences to receive specific meal recommendations. While the current version runs locally on personal computers, the long-term vision is to deploy this application as a real-time web-based tool. By integrating feedback from healthcare

professionals, the application can be refined further to ensure usability, reliability, and alignment with clinical guidelines. Lastly, another area of improvement is efficiency. Due to the methods used to aggregate data and conduct feature engineering, some parts of the code are computationally expensive and execute slowly. While this can be improved with more computational power, the efficiency of the code can also be improved. With more time and resources, future studies can enhance the currently used methods to optimize execution. ACKNOWLEDGMENTS We would like to express our profound gratitude to Dr. Ebrahim Tarshizi, Program Director of the Applied Data Science program at the University of San Diego’s Shiley-Marcos School of Engineering. His continuous guidance and expertise were instrumental in the successful completion of our Capstone and in shaping our development as data science professionals. References American Diabetes Association. (2024). Statistics about diabetes . Diabetes.org. https://diabetes.org/about-diabetes/statist ics/about-diabetes Centers for Disease Control and Prevention. (2024, May 15). National diabetes statistics report . https://www.cdc.gov/diabetes/ php/data-research/ Denniss, E., Lindberg, R., & McNaughton, S. A. (2023). Quality and accuracy of online nutrition-related information: A systematic review of content analysis studies. Public Health Nutrition, 26 (7), 1345–1357. https://doi.org/10.1017/s1368980023000 873

239