Enabling personalized health-care planning for older adults using population risk tools: development, validation, and implementation of prediction algorithms in the Canadian Longitudinal Study on Aging

Many clinical prediction tools are developed using large health administrative databases, as these are often the most comprehensive sources of information on the health and health outcomes of a population. However, algorithms developed using health administrative databases are context-specific and may only capture information pertinent to service delivery, not the full health information on the patient. They also do not reflect the patient’s understanding and assessment of their own health or well-being. As such, when they are applied in a non-clinical setting and for self-completion by patients, the risks presented to the patients may be misestimated. When possible, clinical prediction algorithms intended for patients’ use should be validated against patient-reported data. In this project, we will use the CLSA data to validate and improve clinical risk tools previously developed in population-level health administrative databases.