The Use of an Information-Theory based Diagnostic Rule for Hip Fracture Evaluation and Prediction

Objective: The present work explores the application of a new methodology, based on an information theoretical measure, the normalized mutual information, for hip fracture risk evaluation and prediction. Methods: A dataset on geriatric hip fracture patients was analyzed, including diagnostic parameters routinely available to physicians, such as physiological, biochemical, immunological and hematological parameters. Information-theory based methods were utilized to establish correlations between the parameters and to construct a diagnostic decision rule for hip fracture risk evaluation at different age groups. Results: The use of information-theoretical methods, utilizing normalized mutual information, revealed the exact amount of information that various diagnostic parameters contained about the presence of hip fracture at different ages. Based on those exact informative values for hip fracture evaluation, we constructed a diagnostic rule (a decision tree) to estimate a person’s risk for hip fracture at different ages. We established a risk group for hip fracture at a relatively “younger” age (below 80 years old). We developed an algorithm (decision tree) that can be used to evaluate whether a subject can be categorized as belonging to the risk group for hip fracture under 80 years old. The algorithm’s sensitivity was 58.8% and its specificity was 54.6%. Conclusion: With the addition of further data and validation, algorithms constructed by this methodology can be used to help predict the risks of hip fracture in elderly subjects, in order to optimize preventive interventions. Keywords Information theory; Hip fracture; Diagnostic rule; Normalized mutual information; Prediction