Gendered Data in Falls Prediction Using Machine Learning

Adults over the age of 65 years may be considered as a vulnerable population prone to having falls which may have huge consequences. Machine learning is being explored as an approach to understanding better the specific risk factors for falling. However most studies use composite population data rather than including data on male or female gender in the analysis. This study focused on using machine learning models utilising healthcare data to establish whether gendered data gives a more accurate prediction of falling. Splitting the data into male and female gives slightly higher predictive accuracy, however reducing the size of the dataset is likely to give a lower prediction. Such models could usefully inform health and social care professionals in their daily decision-making with individuals and families about optimal care arrangements.