A fresh perspective on humanitarian aid: in conversation with our students

Solid scientific research is integral to the success of humanitarian efforts as it provides an evidence-base and can drive innovation of established practices in this field through new discoveries. By engaging with students and collaborating with universities, 510 ensures access to the newest academic knowledge and actively solicits critical and fresh outside perspectives. Since being founded in 2016, 510 has had the honor to work with 154 students (and counting!), each of whom has contributed unique scientific advances to the work we do. We caught up with three students who have recently completed their research with us about their time at 510:

• Danielle Rachman, MSc in Population Health Management at Leiden University – Danielle explored the implementation of Early Warning Systems to manage Cholera outbreaks in Cameroon.
• Lorette Galois, BSc in Earth, Energy and Sustainability at Leiden University – Lorette researched Nature-based Solutions to manage flooding and mitigate hazard impacts in Zambia and received the University’s Thesis of Merit award for her thesis.
• Polle Dankers, MSc in Business Analytics and Operations Research at Tilburg University – Polle investigated the implementation of active learning for our Automated Damage Assessment tool.

Could you describe the project you have been working on as part of your internship/thesis research?

Danielle: Cholera remains a public health threat. Early warning systems can play a crucial role in minimising the impact of cholera by predicting outbreaks in advance. Nevertheless, in Cameroon, challenges related to data quality, resources and funding hinder the development of early warning systems. Collaborations between state and non-state organisations may accelerate their development, but the factors which promote their collaboration are yet unclear. My research as part of a project with 510, the French Red Cross and the Cameroon Red Cross Society aimed to gain insights into the factors influencing collaboration, and thus inform the design of organisational structures to support improved collaboration outcomes.

Lorette: I conducted research for a project 510 started implement this year, in collaboration with the World Wildlife Fund (WWF) and the Zambia Red Cross Society. The project aims to implement الحلول المستندة إلى الطبيعة to manage flooding and limit its impacts on communities whilst simultaneously protecting biodiversity and ecosystems in Zambia. With 510, I assessed the benefits of these Nature-based Solutions, such as minimizing flood risk or improving food production, and recorded how previous studies have estimated their performance across environmental, social, economic, and technical domains. To understand which benefits were of highest value to various stakeholders, I conducted interviews with representatives from the Netherlands Red Cross, the Zambian Red Cross, الاتحاد الدولي لجمعيات الصليب الأحمر والهلال الأحمر and WWF.

Polle: 510 uses Automated Damage Assessments (أيه دي أيه) to rapidly estimate building damages following a natural hazard, based on deep-learning models analysing satellite imagery. However, no images which are already labelled, i.e. containing information on the damage of pictured buildings, are available for a newly occurred hazard, which means that the أيه دي أيه models cannot be fine-tuned to maximize their accuracy. At the same time, manually labeling a large amount of data is time-consuming and labour-intensive. Therefore, I researched the implementation of active learning for ADA during my time with 510. Active learning involves a model which chooses those satellite images from which it estimates to learn the most, so the expectation was that labeling these images for training purposes would result in quicker performance gains.

510’s purpose is to improve the speed, quality and cost-effectiveness of humanitarian aid by using data and digital products. How has your research helped 510 reach its purpose?

Danielle: Cholera is a complex disease which is impacted by social, environmental and economic factors. In order to develop triggers for early action before a few cases turn into a large outbreak, effective collaboration between state and non-state actors is crucial. Through my research, I provided 510 with the key factors essential for collaboration in the implementation of early warning systems to guide future collaboration efforts between a variety of stakeholders towards minimizing the impact of Cholera.

Lorette: I gathered data on the benefits offered by Nature-based solutions to protect livelihoods and biodiversity by mitigating flood impacts on communities and their environment. These solutions are directly linked to humanitarian aid as they correspond to the prevention/mitigation phase of the Disaster Risk Management cycle, limiting the need for humanitarian aid during phases of preparedness and response and making its actual use more efficient.

Polle: Sadly, active learning did not yield better results compared to training the model using randomly selected data, and should not be implemented in practice. However, combining a model which is pre-trained on images from previous disasters with fine-tuning based on randomly selected images from the new disaster resulted in large performance gains. Using this combination of pre-training and fine-tuning can improve the quality, efficiency and speed of damage assessments made by 510 by keeping the costs and time associated with labelling limited.

Would you like to share an anecdote from your time at 510?

Danielle: A moment which really showed the inclusive culture of 510 was a potluck dinner. It gave every team member a chance to express their culture by contributing a signature dish from their country. Being part of an international team, I loved the focus on cultivating an understanding of each other’s culture, and on making intentional efforts to foster a positive and supportive work environment.

Lorette: I value all the experience I got from this research internship, especially in terms of presentation skills. I never thought that, at 20 years old and with very limited experience in flood management, I would end up in a room presenting my research to experts from the Red Cross and WWF!

Polle: I found striking how closely entangled research is with practice at 510. Throughout my internship, my supervisor Jacopo went to multiple disaster-struck areas such as Ukraine to help people on the ground, while also using the product I was carrying out research for, ADA, in practice – making its tangible impact visible.

We warmly thank Danielle, Lorette and Polle for sitting down to chat with us, and for their tireless efforts in conducting this important research which will ultimately make our data and digital humanitarian aid solutions faster and more efficient, expanding our reach to help more people affected by disasters. We wish them all the best for their future endeavours!