With an ageing population, the number of fall injuries and fractures in Sweden is expected to increase rapidly, causing disability, and increased morbidity and mortality. Tools for prediction of injurious falls are not available. While effective and inexpensive osteoporosis medications are available, only a small fraction of eligible women receive treatment, mostly due to low awareness of osteoporosis, and lack of case-finding strategies in the healthcare system. A risk algorithm, that relies on data available in electronic patient records, would enable automatic screening for high risk of injurious falls and fractures in older patients to reduce the treatment gap and prevent injurious falls and fractures. The aims of the proposed project are to use (i) machine learning algorithms and (ii) traditional statistical models, and a massive data set based on 7 million individuals, 40 years and older from national electronic health records: • to identify novel risk factors for injurious falls and fractures • to develop accurate models to predict the risk of injurious falls, imminent risk of fractures, and hip fractures in particular. An automated system to identify patients at high risk of injurious falls and fracture would likely increase the proportion of patients receiving falls interventions and preventive osteoporosis medication, thereby mitigating suffering and reducing cost.