A doctoral candidate at the University of Georgia’s Franklin College of Arts and Sciences, Jane Odum, has developed a mobile-first AI platform aimed at improving disease surveillance in low-resource settings. Her innovation, EpiCast, recently earned first place and a $30,000 prize at the Google-sponsored MedGemma Impact Challenge.
The global competition, which attracted more than 850 teams, called on developers to design human-centered AI applications capable of addressing complex healthcare challenges. Odum’s solution distinguished itself by directly tackling long-standing gaps in how disease data is collected, processed, and shared.
Gagan Agrawal, director of the university’s School of Computing where Odum is pursuing her doctorate, described the recognition as a reflection of the institution’s academic strength and the impact of its research. Similarly, Professor John Miller, who supervises Odum’s doctoral work, praised her ability to combine deep technical knowledge with practical, real-world applications. Her research on diffusion-based generative models for epidemiological forecasting played a key role in shaping the development of EpiCast.
Rooted in frontline health realities
Odum’s innovation is informed by personal and regional experiences with disease outbreaks. Growing up in Nigeria, she witnessed the devastating effects of the 2014 Ebola outbreak, where fear spread rapidly alongside the virus and communities relied heavily on local monitoring systems.
Her understanding deepened during the early stages of COVID-19. While visiting Nigeria in 2020, she observed that community health workers had already begun identifying symptoms such as fever, cough, and respiratory distress weeks before official confirmations were made.
However, these early observations were often recorded manually in notebooks and across multiple local languages, creating delays before the information could reach national systems or the World Health Organization. By the time laboratory confirmations were completed, critical opportunities to contain the outbreaks had often passed.
This disconnect between grassroots detection and formal reporting systems became the driving force behind the creation of EpiCast.
How EpiCast works
EpiCast is designed to bridge the gap between informal clinical observations and structured public health surveillance. The platform enables community health workers to describe patient symptoms in their native languages, removing language and technical barriers to reporting.
The system then uses AI to convert these inputs into standardized clinical data aligned with global health frameworks. Within seconds, it can identify symptoms, assign severity levels, and map cases to recognised diagnostic codes—allowing for faster interpretation and response.
A major innovation lies in its offline capability. Unlike many traditional AI tools that depend on cloud computing, EpiCast operates directly on mobile devices. Odum optimised advanced medical language models to function without internet connectivity, reducing processing times from minutes to seconds and making the tool viable in remote or resource-limited environments.
She emphasised that speed and reliability are critical for adoption, noting that even small delays can disrupt workflows in busy clinics.
Enabling faster, cross-border health responses
Beyond improving local reporting, the platform also has the potential to enhance cross-border information sharing. By standardising data from diverse sources and languages, EpiCast helps ensure that early warning signals can be communicated quickly and effectively across regions.
Odum noted that empowering health workers to report in the languages they actually use significantly improves surveillance efficiency and increases the chances of detecting outbreaks early.
A vision shaped by experience
Reflecting on her journey, Odum recalled how both the Ebola crisis and the COVID-19 pandemic highlighted the importance of community-level health intelligence. In many cases, frontline workers were the first to detect unusual patterns, yet lacked the tools to escalate that information in real time.
Her goal with EpiCast is to ensure that such early signals are no longer delayed or lost, but instead integrated seamlessly into national and global health systems.
She stressed that early detection remains the most critical factor in outbreak response, adding that capturing real-time data at the community level could significantly alter the trajectory of epidemics.
Implications for global health
EpiCast represents a growing trend toward leveraging artificial intelligence to strengthen healthcare systems, particularly in regions where infrastructure gaps persist. By combining accessibility, speed, and linguistic inclusivity, the platform offers a scalable solution to one of public health’s most enduring challenges.
As global health stakeholders continue to prioritise preparedness and rapid response, innovations like EpiCast may play a pivotal role in ensuring that the next outbreak is identified—and contained—before it escalates into a full-scale crisis.