Harnessing AI Techniques to Improve Accessibility of Healthcare for Pregnant Women in Kenya
Open Access
Author:
Ranganathan, Prerna
Area of Honors:
Information Sciences and Technology
Degree:
Bachelor of Science
Document Type:
Thesis
Thesis Supervisors:
Amulya Yadav, Thesis Supervisor Marc Aaron Friedenberg, Thesis Honors Advisor
Keywords:
Artificial Intelligence Maternal Health SMS-Based Triage Pregnancy Care
Abstract:
Through the use of artificial intelligence (AI) and natural language processing (NLP), an advanced health-agent system, Promoting Mums through Pregnancy & Postpartum Through SMS (PROMPTS), has been deployed in Kenya to help pregnant mothers, or mums, receive the help they need. Unfortunately, the health care system in Kenya is poor, which leads to challenges in mums getting the necessary care and results in many maternal and neonatal deaths. While the PROMPTS platform has helped to improve the health care that the mums are receiving, there is potential for it to be optimized further by considering the current challenges that the system faces. This paper focuses on the use of traditional classification models to improve classification performance. By analyzing three classification models, Adaptive Boosting (AdaBoost), Random Forest (RF), and k-Nearest Neighbors (k-NN), this work aims to create a new model that combines the three classifiers in an optimal manner. The models discussed in this paper are evaluated using three main performance metrics: precision, recall, and F1 score. With a model that effectively and accurately classifies information, PROMPTS will have a better overall performance.
