Project Inspiration:

Problem Addressed:

Limited access to reliable infant care support during the postpartum period leads to stress, anxiety, and potential health risks for parents and newborns.

Motivation:

Witnessing the struggles of new parents firsthand through your mother's PhD research on postpartum experiences (postpartum depression, lack of 24/7 support, personalized support) highlighted the need for readily available and reliable guidance. Many parents lack access to breastfeeding consultants, expert consultations, or consistent support due to cost or availability.

Importance in Generative AI and SDGs:

This project showcases the potential of generative AI (RAG-based chatbot) to address real-world issues like accessible and trustworthy healthcare information, aligning with UN SDG 3: Ensure healthy lives and promote well-being for all at all ages

Product:

Infant Care Postpartum Chatbot AI

Function:

Acts as a virtual AI baby care consultant, providing personalized, evidence-based advice on infant care for newborns up to 6 months old.

User Experience:

Users interact with the chatbot through a text-based interface, asking questions about various aspects of infant care. The chatbot retrieves relevant information from its knowledge base and generates tailored responses with references to credible sources.

Problem Addressed:

Offers readily accessible, trustworthy infant care information, empowering new parents and reducing stress during the postpartum period.

Innovative Aspects:

Focuses on a specific, curated knowledge base from reliable sources (government and international organizations) to ensure information accuracy. Provides evidence-based advice backed by research and credible resources.

Challenges & Learnings from Product Testing:

Data Preprocessing: The importance of thorough data ingestion with proper text cleaning (removing line breaks and whitespaces) for RAG models handling unstructured data. Size of Context Data: The current knowledge base only allows for answers within the scope of newly born infants upto 6 months. Any queries outside this scope will receive a response from the chatbot indicating this is out of scope. Prompt Engineering: Utilizing advanced prompt engineering techniques like single-shot prompting, few-shot prompting, chain of thought for improved question-answering accuracy.

Future Improvements:

Data Preprocessing: Implement robust text cleaning techniques for better data ingestion. Knowledge Base Expansion: Gather and integrate data from additional reliable sources to answer questions beyond the 0-6 month range, while maintaining focus on trustworthy information. Prompt Engineering: Explore advanced prompt engineering techniques for improved question-answering accuracy.

Additional Features: Incorporate a vision based baby care technique assistant that analysis live video to assist parents with the technique and form the use when performing baby care activities such as calming the baby using burping technique or bottle feeding or kangarooing the baby,etc. Create additional chatbot features such as multilingual support and change the context knowledge based on region specific health authority guidelines.