Prediction of Osteoporosis by K- NN Algorithm and Prescribing Physical Activity for Elderly Women

Purpose: A number of clinical decision tools for osteoporosis risk assessment have been developed to select postmenopausal women for the measurement of bone mineral density. We developed Data mining algorithm with the aim of more accurately identifying the risk of in postmenopausal women compared to the ability of conventional clinical decision tools. Materials and Methods: The present study was a cross-sectional development study conducted in the second half of 2018. In the present study, first, by identifying the influential variables, a survey questionnaire was prepared to select the most important clinical factors. Bone mineral density information of women referred to the bone density measurement unit of Khatam Al-Anbia Hospital in Tehran was used to teach the K-Nearest Ne neighbors (K-NN) algorithm (based on simple studies). Evaluation was based on accuracy. We also reviewed the results of several scientific articles and suggested the best sports activities according to the bone density of individuals. Results: The K-NN algorithm with sub-curve surface (AUC) showed significant performance. The algorithm predicted the risk of osteoporosis with an accuracy of 61.7% in the femoral neck for women participating in the experiment. Also, regular resistance and endurance training exercises repeated for 2-3 times a week for a year can have significant effects on maintaining or increasing hip BMD in postmenopausal women. Conclusion: Considering various pre‌dictors associated with low bone density, the K-NN algorithm may be an ef‌fective tool for identifying women at high risk for osteoporosis. This method widely recommends and predicts regular resistance and endurance training exercises for women with a high risk of osteoporosis.

Prediction of Osteoporosis by K- NN Algorithm and Prescribing Physical Activity for Elderly Women Keywords:

Data mining , bone mineral density , exercise , aging