BHW Assistant 🏥

An AI-powered platform providing real-time protocol support for over 200,000 Barangay Health Workers (BHWs) serving as frontline healthcare providers across rural Philippines.

The Challenge

The Philippines faces a critical healthcare crisis with just 1 doctor per 25,280 people. As a result, 200,000 Barangay Health Workers - 90% of whom are women - must deliver essential primary care despite receiving only 5 days of formal training. These dedicated frontline workers serve 70% of Filipinos living in rural and underserved communities.¹

Our Solution

The Healthworker Assistant processes health worker-patient conversations, comparing them against Department of Health (DOH) protocols to provide relevant guidance while respecting BHW expertise. By combining generative AI with offline-capable speech recognition, it delivers context-specific support in both Tagalog and English, even in areas with limited connectivity.

Studies show this type of AI guidance can:

• Improve protocol adherence by 67%²
• Enhance diagnostic accuracy by 70%²
• Increase BHW confidence by 75%³
• Reduce preventable complications through earlier detection and referral³

Focus Areas

• Prenatal care
• Communicable diseases
• Non-communicable diseases

Key Features

Minimally Intrusive Design

• Simple audio recording interface
• Alerts only for critical danger signs
• Post-interaction guidance review
• Manual verification and editing of captured data

Protocol-Based Analysis

• Real-time processing of health conversations
• Automatic extraction of:
  • Symptoms and complaints
  • Vital signs and measurements
  • Risk factors
  • Current medications
  • Recent health history
• Comparison against DOH protocols
• Generation of:
  • Missing information checklist
  • Symptom-specific guidance
  • Education topic suggestions
  • Protocol recommendations

Multilingual Support

• Handles Tagalog, English, and code-switching
• Context-aware medical terminology processing
• Culturally appropriate health guidance
• Natural language understanding for Filipino medical terms

Offline Capabilities

• Local audio processing
• Offline transcription
• Local protocol storage and matching
• Secure data handling

System Requirements

Hardware

• Android/iOS smartphone with:
  • Adequate microphone
  • Sufficient storage for offline models
  • Full-day battery life

Software Dependencies

• Python 3.11+
• OpenAI API key (for GPT-4 and Whisper)
• Anthropic API key (for Claude)

Installation

1. Clone the repository:

git clone https://github.com/your-username/HealthWorkerAssistant.git
cd HealthWorkerAssistant

2. Install dependencies:

pip install -r requirements.txt

3. Create a .env file with your API keys:

OPENAI_API_KEY=your_openai_key_here
ANTHROPIC_API_KEY=your_anthropic_key_here

Usage

Testing Mode

# Test with sample transcripts
python src/main.py --mode test

# Generate and test with synthetic data
python src/main.py --mode synthetic

Production Mode

# Run with real-time audio processing
python src/main.py --mode production

Output Format

The system provides analysis output containing:

Extracted Information:
- Condition type
- Measurements
- Symptoms
- Covered topics
- Risk factors
- Trimester (if applicable)
- Danger signs

Realtime Alerts:
- Critical danger signs requiring immediate attention

Missing Information:
- Required measurements not yet taken
- Protocol-required questions not yet asked

Symptom-specific Guidance:
- Recommendations based on reported symptoms

Education Topics:
- Suggested health education topics to cover

Protocol Suggestions:
- Next steps based on DOH protocols

Contributing

We welcome contributions! Key areas include:

• Protocol implementation improvements
• Language processing enhancements
• UI/UX refinements
• Documentation updates

License

This project is licensed under the MIT License - see the LICENSE file for details.

Footnotes

1. Department of Health Philippines. (2023). Ratio of active barangay health workers to population in the Philippines in 2023, by region. Statista. https://www.statista.com/statistics/1370633/philippines-number-of-health-workers-by-region ↩

2. Babel et al. (2021). Artificial Intelligence Solutions to Increase Medication Adherence in Patients With Non-communicable Diseases. Frontiers in Digital Health, 3, 669869. https://pmc.ncbi.nlm.nih.gov/articles/PMC8521858 ↩ ↩²

3. Kearns et al. (2024). Bridging the Skills Gap: Evaluating an AI-Assisted Provider Platform to Support Care Providers with Empathetic Delivery of Protocolized Therapy. arXiv:2401.03631. https://arxiv.org/abs/2401.03631 ↩ ↩²