<p>Over a million of children die of a preventable injury every year all over the world. One of effective preventive measure is product improvement. Risk assessment for product improvement requires predicting child interactions with products. However, it is still very difficult even for advanced simulation systems to precisely predict how children interact with consumer products in daily life environments. In this study, the authors are developing a data-driven method as one of promising approaches. The authors collected children’s climbing behavioral data and developed a data-driven simulation system of predicting children's climbing behaviors in daily living environments.</p>