Our Theory of Change
How do we turn artificial intelligence into effective public health interventions? Our Theory of Change maps the logical pathway from our immediate technical development to our ultimate goal: better control and prevention of zoonotic diseases in Mongolia.
The Baseline Context
The bubonic plague, caused by the bacterium Yersinia pestis, is transmitted through fleas that infest small mammals such as marmots. While not globally prevalent, it remains a major public health threat in endemic regions like Mongolia.
Mongolia faces unique challenges: while it is illegal to hunt marmots, they are still widely consumed in the countryside. People often neglect or hide their exposure due to legal fears, creating life-threatening hazards. Traditional methods of contact tracing rely heavily on human labor, which is slow and time-consuming, allowing the highly infectious disease to spread rapidly before targeted responses can be deployed.
The Five Steps
Identify the Problem & Need
Addressing the slow speed of manual tracing and the hidden nature of marmot exposure. AI provides real-time surveillance, analyzing large datasets to identify patterns missed by human analysts, allowing for targeted responses before outbreaks become widespread.
Develop the AI Technology
Collaborating across disciplines to build algorithms that analyze data from medical records, social media, and surveillance systems. A core focus is on mobile phone data based on cellular proximity to assess the chances of exposure.
Implement the Technology
Deploying the technology to public health agencies and integrating it into existing workflows. Strict adherence to patient privacy, confidentiality, and non-discriminatory practices is enforced. The initial pilot invites four provinces of Mongolia for geographical coverage.
Monitor & Evaluate
Tracking the performance of AI algorithms using Key Performance Indicators (KPIs) such as the number of cases identified and the timeliness of the response. Continuous assessment includes gathering stakeholder feedback and reviewing ethical implications.
Articulate Assumptions & Handover
Focusing on policy development to formalize the system. The final AI product will be registered, documented, and handed over as a public good for non-commercial use. Maintenance and nationwide utilization will be led by the Ministry of Health, Mongolia.
The Outcome Chain
Short-Term
Identify needs, establish KPIs, and secure collaboration among all stakeholders (MOH, MUST, NCZD, MNUMS).
Medium-Term
Develop AI technology, deploy early notification services, and ensure timely responses in rural settings.
Long-Term Impact
Effective public health interventions, better control of zoonotic spread, and AI-based standard contact tracing.