Machine Learning Applied to Data Processing of Patients with Osteoporosis: a correlational and comparative analysis between Osseus and DXA exams
Densitometry, Osteoporosis, Osseus, Machine Learning, Random Forest.
Osteoporosis is a disease characterized by impaired bone microarchitecture and causes high socioeconomic impacts worldwide. Although dual-energy X-ray absorptiometry (DXA) is the gold standard for diagnosing the disease, in developing countries access to this test is limited. In this perspective, the present study aims to propose an auxiliary diagnosis through the exams performed in Osseus, a portable and low-cost equipment developed for osteoporosis screening. Using DXA as a reference, a Random Forest algorithm was trained with risk factors in conjunction with the resulting Osseus measurement. The database consisted of a total of 505 patients who had the exam performed on both devices. Of these patients, 21.8% were healthy and 78.2% were diseased (with low bone mineral density or osteoporosis). The database was separated into 80% for training and validation and 20% for testing, the model was trained and validated (using cross-validation with k fold= 5) and the performance obtained on the test basis matched with AUC=0.80. The most relevant variables to indicate the health status of the individual were age, body mass index and the attenuation measured at Osseus. The model proved to be effective for screening individuals with osteoporosis and facilitating the early diagnosis of the disease, which consequently implies the reduction of costs with surgeries, treatment, hospitalizations, and productivity loss.