Abstract
In order to devise a method for predicting a patient's fall risk in clinical practice, we have created the technique for construction of a fall risk prediction formula (TCFP) which outlines a prediction formula based on a retrospective cohort study which analyzed data from electronic medical records (EMR), and implements script execution functions of the EMR. No additional burden was placed on patients or staff, and data regarding 10,000 patients were collected and analyzed. The sensitivity of our prediction formula was 73.6% and its specificity was 68.8%. Prediction accuracy was no lower than previous studies in Japan. Because the EMR packages of a leading vendor have a script execution function, similar implementations can be made on such EMR packages. As a result, we consider TCFP to be an applicable technique in other medical institutions.
