Diagnosis of osteoporosis using intelligence of optimized extreme learning machine with improved artificial algae algorithm

Osteoporosis is the silent killer disease that mostly occur in elderly people because of bone fragility and fracture. Early and accurate diagnosis of osteoporosis saves the patient life. This work focuses on developing an efficient classifier model to support this issue. For this, the proven Extreme learning machine (ELM) is selected. Then a novel hybrid metaheuristic optimization algorithm is developed on fusing two nature inspired metaheuristic optimization algorithms namely Artificial algae algorithm with multi-light source and Monarch butterfly optimization algorithm. It is named as HMBA algorithm. To further increase the diagnostic accuracy of ELM, it is optimized using HMBA. This proposed HMBA-ELM classifier model is used to diagnose osteoporosis from normal subjects. The discrimination efficiency of proposed classifier is compared with other similar classifiers based on the results produced. It is found that the proposed HMBA- ELM has yielded outstanding results mainly in terms of (sensitivity ​± ​SD/specificity ​± ​SD/precision ​± ​SD/MCR±SD/accuracy ±SD) as (99.45 ​± ​0.69/99.77 ​± ​0.31/96.32 ​± ​0.12/0.30 ​± ​0.18/99.70 ​± ​0.21), (98.11 ​± ​0.91/99.56 ​± ​0.28/90.03 ​± ​0.19/0.51 ​± ​0.09/99.49 ​± ​0.18) and (99.26 ​± ​1.13/99.54 ​± ​0.33/97.38 ​± ​0.22/0.4 ​± ​0.31/99.6 ​± ​0.32) respectively for three osteoporosis datasets namely Femoral neck, Lumbar spine and Femoral & Spine. This is highest among all other approaches with less computation time. © 2020 The Author(s)