The Identifying and Measuring Playful Parenting Using Machine Learning Project is pioneering innovative methods to develop an objective measure of playful parenting.
The research is led by Co-PIs Professor Mark Tomlinson, Dr Caspar Addyman, and Dr Daniel Stamate working with two Global South researchers Jeremiah Ishaya and Irene Uwerikowe. Over the past year, efforts have focused on supporting the development of two doctoral students – both of whom are GPI Playful Parenting Scholars – as well as advancing knowledge sharing and dissemination, particularly through work on the BabyJokes dataset.
BabyJokes Hackathon
In early June, Jeremiah and Irene, alongside co-PI Dr Caspar Addyman, participated in the SAGE Sustainability and Society Hackathon. This event challenged teams to develop predictive models for the BabyJokes dataset and to explore the transition from a machine learning model of parent-child interaction to a practical, real-world application. The first challenge focused on building models to analyse parent-infant joke interactions, while the second examined how predictive models could be applied in real-world settings. Initial work has centred on an open-source demonstration dataset of parents engaging in joke-based interactions with their infants. At the same time, the team has been laying the ethical and infrastructure groundwork necessary to work with larger and more sensitive datasets of parents and infants, a process set to begin in early 2025.
Publications
Predicting High vs Low Mother-Baby Synchrony with GRU-Based Ensemble Models
2023 | International Conference on Artificial Neural Networks
2024 | Engineering Applications of Neural Networks
Team members
Supporting emerging researchers
Jeremiah Ishaya, a GPI Playful Parenting Scholar, is applying machine learning to study parent-child interactions and their role in cognitive development. His research focuses on developing models that extract meaningful data from full parent-child interaction videos. Once proof of concept is established, complete interaction videos will be analysed to deepen understanding of how these interactions shape early learning and development. As part of his work, Jeremiah has developed a Docker container for BabyJokes code, improving portability and consistency across research applications.
Irene Uwerikowe, supported by the GPI Capacity Sharing Fund, is conducting doctoral research on microcoding parent-child interactions from joke demonstrations. Her work explores the application of existing eye-gaze measurement libraries to assess their relevance for infants, particularly in African contexts where datasets on child interactions remain limited. Over the past year, Irene was invited to speak at the African Conference of Women Engineers and Scientists and at a meeting focused on AI approaches to the Sustainable Development Goals, reflecting the broader significance of her research.
Their research has progressed significantly, and they are expected to publish first machine learning papers, expected for submission in early 2025. In April, Jeremiah and Irene attended an African Initiative for Mathematical Sciences (AIMS) workshop in Kigali, which provided an opportunity to strengthen their research networks and technical skills. This was followed by a research visit to Cape Town in June, where they collaborated with AIMS colleagues and the ParentApp for Kids team at Stellenbosch University. They also participated in the SAGE Sustainability and Society Hackathon, which focused on developing predictive models for the BabyJokes dataset and exploring real-world applications of machine learning in parent-child interactions.
