Determining of bone mineral density with using neural networks

SUMMARYDetermining of Bone Mineral Density with Using Artificial Neural NetworksResearch Assistant Dr. Veysi AKPOLATOsteoporosis, especially could be observed in older ages period, appears as a serious andcostly public health problem. In recent years, the growth rate of the clinical case:osteoporosis increases because of the following reasons: widespread inactive life, the habitof nutritional deficiency, obesity, diabetes, negligence of active physical exercise havingmechanic stress on bone. The most important outcome of osteoporosis is bone fracturewhich effects morbidity and mortality on a large scale.In order to determine osteoporosis, bone mineral density measurement and diagnosticcriteria of World Health Organization are used. Osteoporosis is generally observed inelderly women of population, for that reason this part of population is having different typeof risk factors. Predetermining of osteoporosis is possible with epidemiological data ofpopulation. However, it is very expensive for both diagnosis and treatment of osteoporosis.For that reason, economically and practically determining of the person having risk of lossbone is very important.Consequently, the aim of the current study is to determine of women having risk of lossbone with using Artificial Neural Network (ANN).In this thesis, the parameters having correlation with osteoporosis are evaluated. Theseparameters are weight, height, age, number of pregnancy, age of menopause, fat mass andbasal metabolic rate.It this study, two different ANN analysis have done with two different data groups. In thefirst part, totally 765 data are evaluated and these data includes weight, height, age, age ofmenopause and bone mineral density (BMD). Mean of weight, height, age, age ofmenopause and BMD are 70.34 kg, 1.57m, 54.94 year, 35.06 year and 0.853 g/cm2respectively.In the second part of study, totally 442 data are evaluated and these data includes weight,height, age, age of menopause, number of pregnant, fat mass, basal metabolic rate andBMD. Mean of weight, height, age, age of menopause, number of pregnant, fat mass, basalmetabolic rate and BMD are 70.7014 kg, 1.57m, 54.56 year, 6.73, 35.62 year, 35,25, 35.62and 0.908 g/cm2, respectively.The results of success rate for first study is 85.96% and for the second one is 70%.As a conclusion, for determining of the women having risk of loss bone, the patterns basedon data taken by women and measurements are used for input data in ANN. ArtificialNeural Network (ANN) having recognition and learning features is used for determiningthe women having risk of loss bone. In this study, with the increasing number of inputpatterns, the success rate of capturing those patterns are being improved gradually. Thismethod can be also use in determining other type of medical diseases.Key Words: Bone-density, Determining, Artificial Neural Network