Faced with the challenge of population ageing, healthcare providers are increasingly in need for evidence-based artefacts to support the decision making process. In this regard, the paper avails of machine learning techniques in a bid to support the elderly care planning with application to hip fracture care in Ireland. Specifically, the inpatient length of stay (LOS), and discharge destination are aimed to be predicted based on learning from patient historical data. The accuracy of various regression and classification techniques was investigated. Random Forests proved to provide a considerable higher accuracy in comparison to other algorithms in our case. The prediction models were trained using the Azure Machine Learning Studio. Furthermore, the models were published as predictive web services on top of the Azure cloud platform. The developed predictors are claimed to make predictions on the LOS and discharge destinations with high accuracy.