Knowledge representation and learning of operator clinical workflow from full-length routine fetal ultrasound scan videos

•A fully-automatic framework analysed operator clinical workflow in fetal ultrasound.•More than 200 hours of routine second-trimester scan video recordings were analysed.•Knowledge representation scheme was developed solely from full-length scan videos.•Dimensionality and complexity of raw video data was significantly reduced.•Machine learning models were learnt to distinguish workflow based on operator skill.