AI-Assisted Preventive Health Advisory

Proposal for Collaboration between Uplift AI and Alkhidmat

Overview

This proposal outlines a collaboration between Uplift AI, a US-based public benefit company focused on leveraging artificial intelligence to address social inequities, and Alkhidmat, a leading NGO providing health services for the underserved in Pakistan. The collaboration aims to develop a voice-based AI system that provides free, accurate, and culturally sensitive preventive healthcare advice in Pakistani languages like Urdu, serving the illiterate and low-literate population.

Background

In Pakistan, approximately 42% of adults are illiterate, and an additional 35% have low literacy levels. Overall a staggering 77% of the population struggles to access vital healthcare information, which is crucial for preventing diseases and maintaining health. Traditional methods of disseminating information are not effective for this demographic, necessitating an innovative approach.

Recent advancements in artificial intelligence have now made it feasible for computer systems to accurately recognize and process spoken Pakistani languages like Urdu, Punjabi and Pashto. This technological advancement enables the creation of voice-based systems that remove literacy barriers, allowing users to access vital information simply by speaking in their native language.

Objective

The primary goal of this collaboration is to make expert-quality preventive healthcare advice freely accessible to anyone with a phone, thereby eliminating the barriers posed by illiteracy.

Proposed Solution

Uplift AI will develop a voice-operated AI system where users can ask preventive health related questions in local Pakistani languages and receive reliable answers. This system will use advanced AI technologies designed to understand and respond appropriately within the local context, ensuring the advice is both accurate and easy to understand. Initially, this service will be available to Urdu-speaking users through WhatsApp voice notes.

The video below demonstrate how the product will work. The actual product works on WhatsApp using voice notes.

A voice operated AI assistant answering a question about antibiotic usage in Urdu.

WhatsApp voice interface for government program information discovery use case.

Roles and Responsibilities

Uplift AI will:

  • Develop the AI system and interface for voice interactions.

  • Fund the development, operation, and initial maintenance of the system.

  • Manage the technical aspects of the system, including updates and quality assurance.

Alkhidmat will:

  • Provide access to healthcare professionals for AI training guidance.

  • Assist in culturally and linguistically tailoring the system via its network of volunteer medical professionals.

  • Promote the system in underserved communities through its nationwide network.

  • Have access to all the data gathered through this collaboration.

Implementation Phases

  1. Initial Setup and Data Collection: Gather and digitize existing healthcare advice with help from Alkhidmat’s medical advisors.

  2. System Development and Training: Utilize feedback from healthcare professionals to ensure the AI responds to queries in ways that are culturally appropriate and easily understandable.

  3. Pilot Testing: Deploy the system to a limited user base of 10,000 people via WhatsApp, monitoring accuracy, engagement, and collecting feedback.

  4. Evaluation and Expansion: Assess the system’s performance post-pilot and plan for scaling up; including extending services beyond WhatsApp users.

Budget Overview

Uplift AI commits to covering:

  • Engineering and development costs: estimated at $85,000.

  • Operational costs for the pilot, including cloud services and WhatsApp integration: approximately $12,400/month.

Alkhidmat’s contributions will be in-kind, involving the recruitment of volunteer medical professionals to provide up-to 500 hours of AI training. Alkhidmat has no direct financial obligations.

Expected Outcomes

  • Increased access to healthcare information for illiterate and low-literate populations.

  • Reduction in preventable health issues through informed decision-making.

  • Long-term sustainability and scalability of the health advisory system

Risk Mitigation

  • Ensure system accuracy through continuous feedback from healthcare professionals during and post-development.

  • Set clear criteria for accuracy and user satisfaction as benchmarks for the pilot’s success.

Evaluation of Success

  • System accuracy will be continually assessed by medical professionals.

  • User engagement and satisfaction will be measured through direct feedback and usage metrics.

Proof of Effectiveness of AI-Assisted Advisories

Below we are showing two examples that highlight the potential impact of AI-Assisted advisories in local languages.

  • AI-assisted agricultural advisory in India: Increased yield by 21% for chili farmers in Telangana through a WhatsApp chatbot providing farming advice in the local language. For more details, read here.

  • AI-assisted fishery advisory in Indonesia: Increased export-qualified shrimp yield by 33%. For more information, read here.

Conclusion

The collaboration between Uplift AI and Alkhidmat represents a transformative opportunity to leverage cutting-edge technology for social good, addressing a critical healthcare need in Pakistan. By combining Uplift AI’s technological expertise with Alkhidmat’s healthcare experience and extensive reach, we aim to significantly improve the accessibility and quality of healthcare information, making a lasting impact on the lives of millions.

Contact

Hammad Malik
hammad@upliftinitiative.com
whatsapp: +1 (607) 280-9922

Uplift AI Team

Hammad Malik

  • Amazon Alexa (~4 years)

  • Apple Siri (~4 years)

  • Masters in Computer Science from Cornell University, USA

  • Bachelors in Computer Science from LUMS, Pakistan

Faizan Ahmad

  • AI Research Engineer at Meta/Facebook (~4 years)

  • Co-Author of Purple LLAMA

  • Masters in Computer Science from University of Virginia, USA (specializing in Deep Learning)

  • Bachelors in Computer Science from FAST, Pakistan

Haseeb Niaz

  • Senior AI/ML Engineer at Apple (~2 years)

  • Previously at Meta/Facebook (~2.5 years) and Microsoft Research (~2.5 years)

  • Lead developer of CrystalNet, Azure’s high-fidelity cloud emulator, showcased at Microsoft Build

  • Masters in Computer Science from Stony Brook University, USA

  • Bachelors from LUMS, Pakistan

Conglo (Media Partner)

  • Conglo is leading media production and marketing company in Pakistan. Conglo has led some of the Pakistan’s most successful marketing campaigns. Notably, their for the Pakistan Super League achieved over 1 billion views.

  • Conglo will help us take our message to every corner of Pakistan.

Dr. Hafiz Asif (Advisor)

  • Professor of AI at Hofstra University, New York, USA

  • Research focuses on AI privacy and safety

  • Author of "Intelligent Pandemic Surveillance via Privacy-Preserving Crowdsensing," which won the 2022 Best Paper Award by the IEEE Computer Society

  • PhD from Rutgers University, USA

  • Bachelor's from LUMS, Pakistan

—

Future Opportunities

  • Automated Voice-Based Explanation of Medical Test Results:
    Develop a system that verbally explains medical test results in simple, clear language, enabling everyone to fully understand their health diagnostics.