Back to Search
Start Over
Stimated Body Fat Percentage using Mechine Learning Techniques
- Source :
- journal of ilam university of medical sciences. 25:171-178
- Publication Year :
- 2017
- Publisher :
- CMV Verlag, 2017.
-
Abstract
- Introduction: Doctors undertake calculation of body fat percentage by using BIA (Bioelectrical Impedance Analysis) equipment. In this study, we measured body fat percentage without using equipment. For this purpose, an artificial neural network has been used to estimate the exact amount of fat. Materials & methods: The sample was selected from patients admitted in a nutrition clinic in Tehran. 400 patients took part in this study. MLP neural network was used to estimate body fat percentage. The used neural network had five input neurons and ten neurons in the hidden layer. Also, cross validation method for evaluating the proposed method has been used. Findings: The proposed method is efficient because of the results that demonstrate 2.5 units error based on cross validation. The results of experiments show that the proposed neural network for estimating body fat percentages has an average accuracy of 93%. Therefore the proposed method can accurately estimate body fat percentage of people with very high accuracy. Discussion & conclusions: The results of this research show that the proposed method as the first method used in machine learning technique, can estimate fat percentage with high accuracy. This method can be used as a useful method without using BIA device.
- Subjects :
- Artificial neural network
business.industry
Sample (material)
Pattern recognition
030204 cardiovascular system & hematology
Body fat percentage
Cross-validation
03 medical and health sciences
0302 clinical medicine
030212 general & internal medicine
Artificial intelligence
Hidden layer
business
Bioelectrical impedance analysis
Mathematics
Subjects
Details
- ISSN :
- 25883135 and 15634728
- Volume :
- 25
- Database :
- OpenAIRE
- Journal :
- journal of ilam university of medical sciences
- Accession number :
- edsair.doi...........149342665e7e2303137654aec6f6e7cd