From Indonesia to the Philippines: How an Interdisciplinary Team is Redefining Dengue Prediction with AI
On November 16, 2024, the AI for Global Health Innovation (AI4GHI) Challenge concluded with its Student Summit, showcasing the ingenuity of 12 finalist teams tackling some of the most pressing global health issues. Among the highlights of the event was the People’s Innovation Award, which went to the team led by Diyah Utami Kusumaning Putri, a Computer Science student from Universitas Gadjah Mada and the University of Vienna. Her team included Guntur Budi Herwanto (Software Engineering, Universitas Gadjah Mada/University of Vienna) and Annisa Maulida Ningtyas (Computer Science, Universitas Gadjah Mada/TU Wien). Together, they developed ExplainDengue, a predictive AI system designed to revolutionize how dengue outbreaks are anticipated and managed.
A Critical Health Concern in the Tropics
Dengue fever is one of the fastest-spreading vector-borne diseases globally, disproportionately affecting tropical and subtropical regions. The disease is driven by an intricate web of climate, environmental, socioeconomic, and behavioral factors. Traditional models often fail to capture the complex spatio-temporal patterns of these variables, leading to inaccurate predictions and limited practical utility.
Enter ExplainDengue: a novel system leveraging neurosymbolic Graph Neural Networks (GNNs) to enhance the accuracy and interpretability of outbreak predictions. Using data from the Project CCHAIN dataset, which maps climatic impact drivers, environmental conditions, and socioeconomic vulnerabilities in 12 Philippine cities, the team created a model that doesn’t just predict outbreaks—it explains why they are likely to occur.
This dual capability offers public health workers a powerful tool to make timely, informed decisions and strengthen their understanding of disease transmission dynamics.
The Science Behind the Solution
ExplainDengue’s innovation lies in its neurosymbolic AI approach, combining deep learning with symbolic reasoning to address the limitations of traditional AI models. The GNNs architecture, enhanced by Gated Recurrent Units, captures both the spatial relationships and temporal dynamics of dengue outbreaks. Unlike purely mechanistic models, ExplainDengue provides interpretable results, offering actionable insights for public health management.
The team already developed a prototype to generate early warnings, alerts, and response strategies, with pilot testing conducted on data from Philippine cities.
The Human Element
For Diyah and her team, the competition was as much about personal growth as it was about technological innovation.
“The most exciting aspect was developing AI solutions for strengthening the dengue prediction system while considering climate-sensitive factors that cause infectious disease transmission,” the team shared. “The challenge in our project was integrating epidemiological factors with dengue cases while using neuro-symbolic AI to provide accurate predictions and explanations.”
The team’s journey offered a crash course in epidemiology, neuro-symbolic AI, and the nuances of real-world problem-solving. It also honed their communication skills, particularly in distilling complex ideas into a three-minute presentation video and a live pitch.
The Future of Innovation
While the team praised the competition’s organization, they suggested adding mentorship sessions to facilitate knowledge exchange and networking. “Sharing progress and challenges with mentors could enhance the learning experience,” they noted.
The AI4GHI Challenge provided not just a platform for innovation but a supportive environment that recognized and celebrated participants’ efforts. “The committee’s guidance and encouragement motivated us to give our best,” Diyah reflected.
A Model for Future Success
The ExplainDengue project exemplifies the potential of interdisciplinary, interregional collaboration in tackling global health challenges. By integrating computer science with public health and climate science, Diyah’s team has created a system that is as insightful as it is practical. Their victory in the People’s Innovation Award underscores the importance of not just building solutions but communicating their value effectively to a wide audience.
As climate change continues to exacerbate vector-borne diseases like dengue, tools like ExplainDengue will become increasingly indispensable. The work of Diyah and her team offers a glimpse into a future where artificial intelligence is not just a tool but a bridge connecting data to actionable insights.
🎥 Watch their research video: https://www.youtube.com/watch?v=DmAKKwg_ofY
📄 Read their abstract: https://drive.google.com/file/d/1zn38cbXEe-VyY80HontwoneS6Y4-RDgK/view
✨ Explore all finalist projects: AI4GHI Finalists