آواز صحت  ·  The Voice of Health

AI for
Maternal Health
in Pakistan

Where every voice matters in healthcare

Awaaz-e-Sehat is a speech-first AI platform improving maternal health outcomes in low-resource settings — built at LUMS, deployed in clinics and communities across Pakistan.

Explore the Platform
500+
EMRs Created
96%
EMR Accuracy
300+
Clinical Red Flags Caught
AWAAZ-E-SEHAT
Aap ka naam kya hai aur aap ki pregnancy kitne hafte ki hai?
Mera naam Fatima hai, 28 hafte hogaye hain
Shukria Fatima. Aaj aap ko koi takleef toh nahi? Sar dard ya nazar mein dhundla pan?
Thoda sar dard hai subah se
⚠ Red flag detected — please consult your doctor today

Built for Pakistan's
Fragmented Healthcare
Reality

Pakistan's maternal mortality crisis is not a shortage of knowledge — it's a shortage of access. Clinicians are overwhelmed. Records are paper-based. Frontline workers lack decision support. Women in rural areas navigate care alone.

Awaaz-e-Sehat was designed at LUMS in response to this reality. Our platform uses voice, WhatsApp, and AI to bring structured healthcare tools to clinics and communities where they are needed most — in Roman Urdu, the language people actually speak.

From hospital clinicians creating EMRs by voice, to pregnant women accessing antenatal guidance on WhatsApp, to traditional birth attendants receiving AI-guided support during home deliveries — Awaaz-e-Sehat meets providers and patients where they are.

Voice-First WhatsApp-Based Roman Urdu RCOG-Aligned Low-Literacy Design WHO Triage
Our Journey
1
2023 — PHASE 1
Clinician-facing voice-to-EMR tool piloted at Shalamar Hospital. Fine-tuned Whisper + GPT-4. Strong technical results; informed workflow integration learnings.
2
2024 — PHASE 2
Pivoted to patient-facing platform. Pregnant women create portable digital records via WhatsApp voice, text, or images in Urdu. CRAFT 2024 recognition.
3
2025–26 — PHASE 3
Expanded into comprehensive antenatal health assistant. TBA module in active development.
CURRENT
Gates Foundation-funded. Rural community deployment underway. TBA decision-support tool in active development.

Four Modules.
One Ecosystem.

Each module addresses a distinct link in the maternal care chain — designed to work standalone or as an integrated platform from community to clinic.

💬
Module 1
APPA
WhatsApp-Based Patient Engagement Chatbot

A calm, midwife-style AI communicator that meets pregnant women on WhatsApp — the platform they already use. APPA conducts antenatal health conversations in Roman Urdu, tracks symptoms, delivers educational content, and triages concerns using a WHO-aligned 3-tier framework.

  • Conversational antenatal guidance grounded in RCOG guidelines
  • WHO-aligned triage: routine, urgent, and emergency pathways
  • Symptom check-ins, appointment reminders & follow-up prompts
  • Voice, text, and image input — no app download required
  • Portable digital health records created via WhatsApp
🎙️
Module 2
EMR
Voice-Based Electronic Medical Record Generation

Clinicians speak naturally in Urdu — Awaaz-e-Sehat listens, transcribes, and generates structured electronic medical records. Built on fine-tuned Whisper ASR and a GPT-4 backend with medical dictionaries. Validated at 96% downstream EMR accuracy over 500+ records.

  • Fine-tuned Whisper ASR on Roman Urdu clinical speech
  • 6-section structured EMR: history, vitals, obstetric, labs, socioeconomic, plan
  • Automated red-flag logic with clinician acknowledgement gates
  • Auto-computed gestational age and EDD
  • Image upload for scan and lab report attachment
📊
Module 3
Dashboard
Clinician Oversight & Triage Interface

A centralized view for clinicians to monitor patient panels, review AI-generated EMRs, triage red flags, and manage referrals — all from a single secure interface. Designed to integrate into existing hospital workflows without disruption.

  • Patient panel view with risk stratification
  • Red flag queue with priority sorting and acknowledgement workflow
  • QR-code consent system for secure patient data access
  • EMR review, edit, and sign-off workflow
  • Referral tracking and inter-facility handover support
🤝
Module 4 In Development
TBA Tool
AI Decision-Support for Traditional Birth Attendants

31% of deliveries in Pakistan are attended by traditional birth attendants (dais) — unskilled, often low-literacy, and working without decision support. This speech-first module guides TBAs through safe delivery protocols and postpartum monitoring in Urdu, with tap and voice inputs designed for low-literacy users.

  • Step-by-step delivery guidance with voice prompts in Urdu
  • Multi-method blood loss estimation (visual, voice, tap/photo)
  • Danger sign recognition and escalation protocols
  • Feeds into shared EMR and Clinician Dashboard
  • Feeds into shared EMR and Clinician Dashboard
Research

Peer-Reviewed &
Field-Validated

500+
EMRs generated in 7-month deployment at Shalamar Hospital
96%
Downstream EMR field-level accuracy on clinical data
300+
Clinical red flags identified by the AI system
ACM UbiComp 2026  ·  Published
Awaaz-e-Sehat: A Mobile Voice-based AI System for EMR Generation and Clinical Decision Support in Low-resource Maternal Healthcare
Maryam Mustafa, Amna Shahnawaz, Umme Ammara, Moaiz Abrar, Bakhtawar Ahtisham, Fozia Umber Qurashi, Mostafa Shahin, Beena Ahmed  ·  LUMS & UNSW · March 2026
doi.org/10.1145/3790115 →
ACM CHI 2026  ·  Under Review
Beyond Euphemisms: Rethinking LLMs for SRH in Conservative Contexts
Two-stage study in Pakistan: qualitative analysis of indirect communication strategies in sexual and reproductive health, followed by performance evaluation of five LLMs — LLaMA 3.2, Gemma 3, GPT-OSS, GPT-4o, and Claude Sonnet 4 — on Roman Urdu clinical data.
Contact for preprint →

In the News &
Featured Stories

Awaaz-e-Sehat has been featured across international media, foundation platforms, and public-interest coverage highlighting its work in maternal healthcare innovation.

Featured Audio
BBC World Service — The Conversation: Pregnancy by Numbers
A BBC World Service conversation featuring Dr. Maryam Mustafa on pregnancy, data, AI, and why women need better information and support.
Listen now →
Foundation Feature
How a computer scientist is using AI to save mothers' lives in Pakistan
The Gates Foundation profiled Dr. Maryam Mustafa and Awaaz-e-Sehat, highlighting how the platform helps close dangerous gaps in maternal care in low-resource settings.
Read feature →
Press Coverage
From dai to AI: Lahore experiments with a virtual midwife
Express Tribune covered Awaaz-e-Sehat as part of a broader story on how AI is reshaping pregnancy care and recordkeeping in low-resource settings.
Read article →
Additional Coverage

Supported by
Global Leaders

Our work is funded and supported by leading global health organizations, technology foundations, and clinical institutions committed to equitable healthcare innovation.

🏥
Bill & Melinda Gates Foundation
Primary Funder
🔬
Digital Square at PATH
Technical Partner & Funder
💡
Google Research
Research Partner & Funder
🌐
Patrick J. McGovern Foundation
Funder
🎓
LUMS
Research & Development Home
Team

The People Behind
Awaaz-e-Sehat

Dr Maryam Mustafa
Dr. Maryam Mustafa
Co-Founder
Director, Saida Waheed Gender Centre & Associate Professor of Computer Science, LUMS
Dr Fozia Qureshi
Dr. Fozia Qureshi
Co-Founder
Professor, Shalamar Institute of Health Sciences
Saad Hussain
Saad Hussain
Co-Founder
Contact

Let's Build Better
Healthcare Together

We welcome clinical partnerships, research collaborations, and conversations about deployment and scale in low-resource healthcare settings.

Contact Us