The number of convolutional neural networks (CNN) available for fracture detection and classification is rapidly increasing. External validation of a CNN on a temporally separate (separated by time) or geographically separate (separated by location) dataset is crucial to assess generalizability of the CNN before application to clinical practice in other institutions. We aimed to answer the following questions: are current CNNs for fracture recognition externally valid?; which methods are applied for external validation (EV)?; and, what are reported performances of the EV sets compared to the internal validation (IV) sets of these CNNs? The PubMed and Embase databases were systematically searched from January 2010 to October 2020 according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. The type of EV, characteristics of the external dataset, and diagnostic performance characteristics on the IV and EV datasets were collected and compared. Quality assessment was conducted using a seven-item checklist based on a modified Methodologic Index for NOn-Randomized Studies instrument (MINORS).Aims
Methods
Oxford joint scores are increasingly being used in evaluating outcomes following orthopaedic surgery. These patient-reported outcome measures (PROM) have been well validated, but only before and after surgical intervention. We postulated that the scores would deteriorate in the normal population with age. Members of the public accompanying patients to out-patients and the emergency department in our hospitals were asked to complete an Oxford score questionnaire having ascertained that they had no previous problem with that joint. Exclusions included other multiple joint pathologies and known connective tissue disorders. Power analysis advocated 40 subjects per decade per joint for significance at the 80% mark. 993 subjects between 20 and 80 years of age completed the forms. There were more females than males. The scores were analysed using STATA 8 software. Non-parametric tests of variance, regression analysis, and ANOVA were used. The data were analysed by decades.Background
Methods
