LLMs for Maternal Health Question-Answering and Complications Detection

The high rates of maternal mortality in Africa can be addressed by improving maternal knowledge on obstetric danger signs. Recent research has shown a high rate of acceptance of health question-answering (QA) chatbots. Large language models (LLMs) have also been shown to be effective at detecting diseases from textual and image data. Students taking part in this study will have the opportunity to work on one of two tasks. The first is to continue our research lab's work on evaluating how effective LLMs are in providing context-aware responses (personalized for the users' literacy level, emotion support needs etc.). The second stream of work students can pursue is to assess the effectiveness of LLMs vs other AI tools (like IBM Watson) in detecting pregnancy complications, especially those that are more prevalent in Africa.