Analysis of an Application for Fall Risk Screening in the Elderly for Clinical Practice: A Pilot Study

To test the accuracy of an App in determining the risk of falls in elderly women, 24 elderly women admitted to a hospital were evaluated by the TechOne App, regarding self-report of falls, by the clinical questionnaire ("TFQ") and subjective motor evaluations by the questionnaire "TMPSE". Then, in one faller and one non-faller volunteer, stabilometric variables were measured in the static bipedal posture by the inertial sensing algorithm "TBM". The scores of the questionnaires were compared between fallers and non-fallers by the ANOVA test for independent measures and by the ability in discriminate fallers by the Area Under de Curve (AUC) of the ROC curve and traditional classification measures. The TechOne App demonstrated to be technically viable to obtain stabiliometric variables, demonstrating clear difference between one faller and one non-faller volunteer. The TFQ showed excellent metrics (Sensitivity: 90%; Specificity: 71.4%; Accuracy: 79.17%; AUC: 0.875) and difference between groups (p < 0.05). The TMPSE score showed no significant differences and presented lower classification metrics (Specificity: 64.3%; AUC: 0.654). It is concluded that TBM is potentially viable and the TFQ is accurate to predict falls.