CrowdMom - A Generative-AI & Volunteer Crowdsourcing Assistant for Maternal Support

A mobile platform that combines AI guidance with peer support to improve maternal wellbeing, decision confidence, and access to trustworthy information.

Objective

Design and evaluate a scalable, human-centred assistant that combines retrieval-augmented AI with peer support to improve maternal wellbeing, decision confidence, and access to trustworthy information.

CrowdMom uses Retrieval-Augmented Generation (RAG) to provide up-to-date, source-linked answers drawn from vetted health guidance and programme materials. A structured knowledge base supports explainable responses, while community features invite peer mentoring and moderation. The platform’s UX emphasises warmth, clarity, and low cognitive load. Evaluation covers information quality, empathy, safety, and user outcomes (e.g., anxiety reduction, self-efficacy, and help-seeking).

Description

Pregnancy and early motherhood can be joyful—and also confusing, emotional, and sometimes isolating. Many women face symptoms they don’t fully understand, decisions they’re unsure about, or simply wish they could talk to someone who has been through it before. While clinical care is essential, it rarely covers everyday emotional and practical needs. CrowdMom explores how to combine smart tools with real human connection in one supportive mobile platform.

The project designs and evaluates a system that blends generative AI with volunteer crowdsourcing, so women can ask questions, track wellbeing, share worries, receive daily tips, and feel supported—not only medically, but emotionally and socially.

What’s Unique:

  • AI guidance: Personalized insights, symptom interpretation, and trusted suggestions grounded in health knowledge and user context
  • Peer connection: A network of real mums offering encouragement, validation, and practical tips via chat, forums, and story-sharing
  • Comprehensive support: Spans the full journey—from trying to conceive to pregnancy and early parenting—and translates needs into concrete features

Key Features:

  • Mood & symptom tracking with gentle, personalized nudges
  • Symptom checker with plain-language explanations and links to sources
  • Feeding & sleep support; daily tips and activity ideas
  • Autism-aware guidance and resource signposting
  • Private/anonymous peer chat and moderated forums
  • Crisis-aware safety rails with escalation to professional help lines
  • Inclusive content style (readability, localization, tone control)

Methods & Evaluation:

  • RAG over curated sources: Prompt templates optimized for empathy + clarity
  • Safety layers: Content filters, medical disclaimers, escalation logic
  • Metrics: Answer quality (citation accuracy, groundedness), retrieval (P@k, nDCG), usability (SUS), perceived empathy, and qualitative interviews
  • Privacy by design: Minimal data collection, transparent settings, opt-in sharing

Project Details

Collaborator(s): Dr. Sara Seaid (University of Calgary)

Highlights:

  • Prototype app + knowledge base development
  • Formative user interviews and research
  • Initial evaluation plan and methodology design

Date: 2025-Ongoing