Machine learning technology and its application to computer games for health education

Abstract

Driven by an initiative of the Adult & Paediatric Diabetes Psychology Service of New Zealand, research has been performed to develop new mechanisms, in the form of computer games, to educate children and teenagers about living with diabetes. Aimed at achieving maximum education effectiveness, the ultimate goal of our research is to develop innovative machine learning algorithms that can be used in games to improve their ability to learn about the changing needs of children and to incorporate this intelligence into the game interface. We also plan to collect and report on the information collected from testing our computer games within a small group of children who have been diagnosed with type I diabetics.

Our research plan has been structured into three major stages, starting from modifying some open-source games towards constructing brand new 3D games that are powered by intelligent machine learning technologies for highly effective education. Specifically, at stage one, a joint collaboration has been established between healthcare professionals at the Starship hospital and Unitec staffs from both the Department of Computing and the Department of Performing and Screen Arts. The main objective of this collaboration is to develop a working game prototype that can deliver essential health knowledge to children and meet reasonable usability requirements.

In line with our research plan, fast development of a game prototype becomes the first technical challenge faced by the research team. To tackle this challenge, a decision is made to embed education features, including a variety of visual effects that provide knowledge-rich feedbacks to game players, into an existing 2D game. After evaluating and comparing a number of open-source games, a Java-based Mario Bros game has finally been identified as an ideal base for the game prototype. The Mario Bros game has long been considered as an engaging game for children and can be easily migrated to different computing platforms including mobile systems.

For the purpose of enriching the Mario Bros game with essential diabetes knowledge, we have proposed three important strategies, namely structure enhancement, feedback enhancement, and challenge enhancement, to guide our design of education games. The three strategies are derived naturally from Malone’s conditions that help to induce the flow state, which is marked by children’s intensive involvement in a series of game-playing activities. Extensive use of the three strategies is clearly evidenced in the design of our first game prototype, which is aimed at educating children with the right skill to manage their diabetes through regular exercise, consuming healthy food, and daily insulin intake.

Implementation of the modified Mario Bros game has been completed successfully. Our game is able to show the blood sugar level of the main character which is called Mario in the game. The dynamic change of Mario’s blood sugar level is directly related to Mario’s physical activities, such as walking and jumping. When the blood sugar level falls outside a safe zone, the blood sugar indicator will change its colour to warn children. Meanwhile a window will pop up and children will be asked to either eat some food or inject a certain amount of insulin in order to solve the problem. Our game also features a stage-based design. At early stage of the game, a small set of recommended food or insulin injection will be provided when Mario has abnormal blood sugar level. At later stages of the game, the challenge increases and children need to choose among a large variety of food and need to decide the right amount of insulin injection.

Our diabetes education game was successfully presented at Unitec's Kaleidoscope event in July 2011, at URC research forum in October 2011, and at the OZHI conference in December 2011. It has also attracted wide public interests and has been reported in the Advance magazine and the Aucklander. Inspired by the success of the first game prototype, great efforts have been made to design and develop a 3D game packed with more interactive, entertaining and educational features. The 3D game was completely designed and developed by the research team. The implementation of the game will be finalized soon.

In the future, we plan to evaluate our game prototypes with a small group of children at the Starship hospital. Hopefully, the evaluation results will shed some new lights on the effectiveness of computer games as mainstream tools for child education. The lessons learned from the game prototypes will also pave the way towards developing powerful machine learning technologies and smart learning environments that promote adaptive and prolonged learning experience.