Introduction Atrial fibrillation (AF) remains a prevalent but frequently undiagnosed condition, increasing the risk of severe complications such as stroke. Identifying patients at high risk of undiagnosed AF by analysis of routine in-patient clinical data offers a promising avenue for early detection and intervention. Methods We will conduct a systematic search of online databases to identify models that estimate the risk of new-onset AF, as developed in adult, routine, in-hospital medical records data. We aim to extract and categorise common risk factors for AF. Generalisability and Implications This systematic review will be the first to analyse routine in-hospital medical data to identify risk factors for undiagnosed AF.