AI BioImage Analysis Hackathon

The team

• Scott Brooks (Project Lead), Warwick Medical School

• Nina Pucekova, PhD Student, Warwick Medical School

The project

The AI BioImage Analysis Hackathon was established to bring together Early Career Researchers from diverse disciplines to tackle genuine biological research challenges using computational approaches. Modern biological research increasingly relies on interdisciplinary collaboration, particularly in fields such as microscopy, image analysis and artificial intelligence. However, researchers often have limited opportunities to work directly with colleagues outside their own discipline or gain practical experience communicating across subject boundaries.

The hackathon aimed to bridge this gap by bringing together researchers from biological sciences, computer science, engineering, mathematics and data science to work intensively on real bioimage analysis problems proposed by biological researchers. Rather than using hypothetical datasets or programming exercises, every challenge was based on genuine microscopy datasets and research questions, allowing participants to experience the practical challenges faced by biological researchers while applying modern computational techniques such as artificial intelligence, computer vision, image analysis and data science.

Participants were organised into interdisciplinary teams, ensuring that each group combined complementary biological and computational expertise. Throughout the three-day event, teams received support from organisers, mentors and judges representing a range of academic backgrounds. The event concluded with project presentations and judging, providing participants with valuable experience of communicating interdisciplinary research to a broader scientific audience.

The primary objectives of the project were to:

- Develop interdisciplinary communication and collaboration skills.

- Introduce computational researchers to genuine biological research challenges.

- Increase awareness of modern bioimage analysis techniques.

- Create new interdisciplinary collaborations that continue beyond the event.

- Demonstrate the value of artificial intelligence and quantitative image analysis within biological research.

- Build confidence among Early Career Researchers working outside their traditional disciplines.

The project also aimed to strengthen interdisciplinary research culture across the University while providing a model for future collaborative training events.

Project outcomes

Overall, the AI BioImage Analysis Hackathon successfully achieved its intended objectives and demonstrated the significant value of bringing together researchers from different disciplines to solve real scientific problems.

The event attracted strong interest from across the University, with 32 participants being selected for the event. Unfortunately, seven participants withdrew shortly before the event after catering arrangements had already been finalised. Despite these late withdrawals, teams were successfully reorganised to maintain a balance of biological and computational expertise across all projects, allowing the event to proceed without compromising the interdisciplinary nature of the hackathon.

Eight interdisciplinary research projects were completed during the event, each addressing a genuine biological image analysis challenge. Participants worked with authentic microscopy datasets rather than simplified educational examples, requiring them to rapidly develop an understanding of both the underlying biology and the computational methods required to address the research question. This created an environment in which participants were constantly learning from one another while producing meaningful scientific outputs.

Throughout the hackathon, participants were supported by organisers, mentors and judges from a range of academic disciplines. The judging panel commented on the exceptionally high standard of collaboration demonstrated by the teams and were particularly impressed by how quickly participants developed effective interdisciplinary working relationships.

Feedback from participants was overwhelmingly positive. Every respondent rated the level of interdisciplinarity within the event as either Excellent or Very Good, while 100% also reported that their interdisciplinary communication skills had improved as a direct result of participating in the hackathon. This was particularly encouraging given that improving communication between researchers from different disciplines was one of the primary aims of the project.

In addition, 86% of respondents reported that they had gained new research skills that they intended to apply within their own work. This demonstrates that the event not only promoted collaboration but also delivered practical training that participants considered valuable for their future research careers.

Perhaps most encouragingly, 43% of respondents indicated that they intended to continue collaborating with their hackathon team following the event, while 71% reported that they intended to continue developing the work independently. These findings suggest that the event generated research ideas with genuine longer-term potential rather than simply producing outcomes confined to the three-day competition.

The overall organisation of the event was also rated extremely highly, with 100% of respondents describing it as either Excellent or Very Good. While organising the hackathon involved numerous logistical challenges, including accommodating participant withdrawals and restructuring teams at short notice, it was rewarding to see these efforts reflected in such positive feedback.

The event concluded with prizes recognising the strongest projects. Following the late withdrawal of an invited external speaker, additional funding was redirected towards participant prizes, allowing an extra award to recognise the project judged to have the greatest future development potential. Participation certificates were also issued to all attendees, enabling participants to formally recognise their involvement and showcase the experience through professional development platforms such as LinkedIn.

Qualitative feedback from judges, Principal Investigators and participants was equally positive. Judges remarked that the Early Career Researchers participating often demonstrated stronger interdisciplinary collaboration than many established interdisciplinary research teams, highlighting the effectiveness of the event in breaking down disciplinary barriers. There was widespread enthusiasm for running the hackathon again in future years, with many suggesting that increased preparation time and greater industrial involvement would further enhance its impact.

Although the event was highly successful, it also highlighted opportunities for improvement. Greater pre-event discussion with biological project leads would allow projects to be refined further before the hackathon, ensuring that every challenge has a realistic scope while remaining scientifically interesting. Similarly, increased organisational time would provide greater opportunities to engage industrial partners, an area highlighted by both judges and participants as having significant potential for future events.

Overall, the hackathon exceeded expectations, demonstrated the value of interdisciplinary training for Early Career Researchers and established a strong foundation for future AI BioImage Analysis Hackathons at the University.

Enhancing interdisciplinarity

Supporting interdisciplinary collaboration was the central aim of the AI BioImage Analysis Hackathon, and this objective was successfully achieved. By bringing together researchers from biological sciences, computer science, engineering, mathematics and data science, the event created opportunities for participants to develop collaborations that would have been unlikely to occur through their normal research activities.

Throughout the event, biological researchers gained first-hand experience of computational approaches including artificial intelligence, machine learning, computer vision and quantitative image analysis. Equally, computational researchers developed a much deeper appreciation of experimental biology, microscopy and the practical constraints involved in generating biological datasets. This mutual exchange of expertise was consistently highlighted within participant feedback as one of the most valuable aspects of the event.

Importantly, the hackathon generated collaborations that are expected to continue beyond the event itself. Nearly half of respondents reported that they intended to continue collaborating with their hackathon team, while over two-thirds intended to continue developing their project independently. Since the event, conversations with participants have reinforced this, with several teams expressing continued interest in progressing their ideas further.

The hackathon also strengthened relationships between departments across the University and demonstrated considerable appetite for future interdisciplinary initiatives. Feedback from judges and Principal Investigators strongly supported repeating the event annually, with suggestions to increase industrial engagement and expand the scale of future hackathons.