scholarly journals Machine learning approaches for the prediction of bone mineral density by using genomic and phenotypic data of 5130 older men

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Qing Wu ◽  
Fatma Nasoz ◽  
Jongyun Jung ◽  
Bibek Bhattarai ◽  
Mira V. Han ◽  
...  
Keyword(s):  
Machine Learning ◽  
Bone Mineral Density ◽  
Linear Regression ◽  
Bone Mineral ◽  
Genomic Data ◽  
Genotype Imputation ◽  
Gradient Boosting ◽  
Mineral Density ◽  
Phenotypic Data ◽  
Individual Snps

AbstractThe study aimed to utilize machine learning (ML) approaches and genomic data to develop a prediction model for bone mineral density (BMD) and identify the best modeling approach for BMD prediction. The genomic and phenotypic data of Osteoporotic Fractures in Men Study (n = 5130) was analyzed. Genetic risk score (GRS) was calculated from 1103 associated SNPs for each participant after a comprehensive genotype imputation. Data were normalized and divided into a training set (80%) and a validation set (20%) for analysis. Random forest, gradient boosting, neural network, and linear regression were used to develop BMD prediction models separately. Ten-fold cross-validation was used for hyper-parameters optimization. Mean square error and mean absolute error were used to assess model performance. When using GRS and phenotypic covariates as the predictors, all ML models’ performance and linear regression in BMD prediction were similar. However, when replacing GRS with the 1103 individual SNPs in the model, ML models performed significantly better than linear regression (with lasso regularization), and the gradient boosting model performed the best. Our study suggested that ML models, especially gradient boosting, can improve BMD prediction in genomic data.

scholarly journals Comparison of Machine Learning Approaches in Prediction of Bone Mineral Density in Elderly Men

2020 ◽  
Author(s):  
Qing Wu ◽  
Fatma Nasoz ◽  
Jongyun Jung ◽  
Bibek Bhattarai
Keyword(s):  
Neural Network ◽  
Machine Learning ◽  
Bone Mineral Density ◽  
Random Forest ◽  
Linear Regression ◽  
Genomic Data ◽  
Gradient Boosting ◽  
Learning Models ◽  
Mineral Density ◽  
Machine Learning Models

AbstractBone mineral density (BMD) is a highly heritable trait with heritability ranging from 50% to 80%. Numerous BMD-associated Single Nucleotide Polymorphisms (SNPs) were discovered by GWAS and GWAS meta-analysis. However, several studies found that combining these highly significant SNPs together only explained a small percentage of BMD variance. This inconsistency may be caused by limitations of the linear regression approaches employed because these traditional approaches lack the flexibility and the adequacy to model complex gene interactions and regulations. Hence, we developed various machine learning models of genomic data and ran experiments to identify the best machine learning model for BMD prediction at three different sites. We used genomic data of Osteoporotic Fractures in Men (MrOS) cohort Study (N=5,133) for analysis. Genotype imputation was conducted at the Sanger Imputation Server. A total of 1,103 BMD-associated SNPs were identified and corresponding weighted genetic risk scores were calculated. Genetic variants, as well as age and other traditional BMD predictors, were included for modeling. Data were normalized and were split into a training set (80%) and a test set (20%). BMD prediction models were built separately by random forest, gradient boosting, and neural network algorithms. Linear regression was used as a reference model. We applied the non-parametric Wilcoxon signed-rank tests for the measurement of MSE in each model for the pair-wise model comparison. We found that gradient boosting shows the lowest MSE for each BMD site and a prediction model built using the machine learning models achieves improved performance when a large number of SNPs are included in the models. With the predictors of phenotype covariate + 1,103 SNPs, all of the models were statistically significant except neural network vs. random forest at femoral neck BMD and gradient boosting vs. random forest at total hip BMD.

scholarly journals Association Between Bone Mineral Density And Serum Creatinine In People < 46 Years Old

2021 ◽  
Author(s):  
Yanru Guo ◽  
Xianyang Zhu
Keyword(s):  
Bone Mineral Density ◽  
Regression Analysis ◽  
Linear Regression ◽  
Serum Creatinine ◽  
Bone Mineral ◽  
Linear Regression Analysis ◽  
Lumbar Bone Mineral Density ◽  
Multivariate Linear Regression Analysis ◽  
Mineral Density ◽  
The Relationship

Abstract Purpose: To research the relationship between serum creatinine and lumbar bone mineral density in people aged <46 years. Methods: A total of 10,968 subjects from the American Nhanes database were included in this cross-sectional study, including 5,744 males (mean age 26.2 years) and 5224 females (mean age 26.7 years). The exposure factor is the serum creatinine value, and the outcome indicator is the lumbar bone mineral density. This study mainly used multivariate linear regression analysis to test the relationship between lumbar bone mineral density and serum creatinine. Results: In the multivariate linear regression analysis, serum creatinine was positively correlated with lumbar bone mineral density (β = 0.122, 95%CI: 0.047-0.198), but in the subgroup analysis stratified by sex, this positive correlation only exists in the female population (Β = 0.186, 95%CI: 0.070-0.301).Conclusions: Our study found that in women aged <46 years with normal renal function, there is a positive correlation between serum creatinine and lumbar BMD. And in those people, the determination of serum creatinine can provide a sensitive biomarker for the early identification and treatment of Osteopenia or osteoporosis.

scholarly journals Associations between serum calcium, 25(OH)D level and bone mineral density in adolescents

2021 ◽  
Vol 61 (1) ◽  
Author(s):  
Kaiyu Pan ◽  
Rongliang Tu ◽  
Xiaocong Yao ◽  
Zhongxin Zhu
Keyword(s):  
Bone Mineral Density ◽  
Vitamin D ◽  
Linear Regression ◽  
Bone Mineral ◽  
Serum Calcium ◽  
Curve Fitting ◽  
Smooth Curve ◽  
Linear Regression Models ◽  
Mineral Density ◽  
Positive Correlation

Abstract Backgrounds It is important to improve our understanding of the roles of calcium and vitamin D in bone health for preventing osteoporosis. We aimed at exploring the associations between serum calcium, vitamin D level, and bone mineral density (BMD) in adolescents included in the National Health and Nutrition Examination Survey (NHANES) 2001–2006. Methods Weighted multivariate linear regression models were used to estimate the associations of serum calcium, 25(OH)D level with total BMD. Smooth curve fitting was used to explore the potential non-linear relationship. Results A total of 5990 individuals aged between 12 and 19 years were included in this study. The fully-adjusted model showed serum calcium positively correlated with total BMD. However, an inverted U-shaped relationship was found when we performed the smooth curve fitting method, and the inflection point was calculated at 9.6 mg/dL using the two-piecewise linear regression model. In contrast, there was a positive correlation between serum 25(OH)D and total BMD after adjusting for potential confounders. Conclusions The present study revealed a positive correlation between serum 25(OH)D level and total BMD, and an inverted U-shaped relationship between serum calcium and total BMD.

scholarly journals Bone Mineral Density and Hemoglobin Levels: Opposite Associations in Younger and Older Women

2021 ◽  
Vol 18 (10) ◽  
pp. 5495
Author(s):  
Tzyy-Ling Chuang ◽  
Malcolm Koo ◽  
Mei-Hua Chuang ◽  
Yuh-Feng Wang
Keyword(s):  
Bone Mineral Density ◽  
Regression Analysis ◽  
Lumbar Spine ◽  
Linear Regression ◽  
Bone Mineral ◽  
Linear Regression Analysis ◽  
Laboratory Data ◽  
Hemoglobin Level ◽  
Adult Women ◽  
Mineral Density

This cross-sectional, retrospective medical record review study aimed to investigate the association between hemoglobin levels and bone mineral density (BMD) in adult women. Medical records obtained from general health examinations conducted from June 2014 to July 2020 at a regional hospital in southern Taiwan were reviewed. Anthropometric and laboratory data were recorded. BMD of the lumbar spine and bilateral femoral neck regions was assessed by dual energy X-ray absorptiometry. Linear regression analysis was used to assess the association between BMD and hemoglobin level with and without adjusting for other anthropometric and laboratory data. The study included 9606 female patients with a mean age of 55.9 years. Of these, 2756 (28.7%) were aged ≤50 years and 6850 (71.3%) were aged >50 years. Results from multiple linear regression analysis showed that hemoglobin and femoral and lumbar spine BMD were significantly correlated. A higher hemoglobin level was significantly associated with a lower BMD level in females aged ≤50 years, but with a higher BMD level in those aged >50 years. Given the relationship between bone metabolism and hematopoiesis, additional research is needed to elucidate the association between hemoglobin and BMD levels in different age groups, particularly in premenopausal and perimenopausal women.

scholarly journals Bone mineral density response prediction following osteoporosis treatment using machine learning to aid personalized therapy

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Thiraphat Tanphiriyakun ◽  
Sattaya Rojanasthien ◽  
Piyapong Khumrin
Keyword(s):  
Machine Learning ◽  
Bone Mineral Density ◽  
Treatment Regimen ◽  
Bone Mineral ◽  
Demographic Data ◽  
Osteoporosis Treatment ◽  
Response Prediction ◽  
Receiver Operating Curve ◽  
Mineral Density ◽  
Treatment Responses

AbstractOsteoporosis is a global health problem for ageing populations. The goals of osteoporosis treatment are to improve bone mineral density (BMD) and prevent fractures. One major obstacle that remains a great challenge to achieve the goals is how to select the best treatment regimen for individual patients. We developed a computational model from 8981 clinical variables, including demographic data, diagnoses, laboratory results, medications, and initial BMD results, taken from 10-year period of electronic medical records to predict BMD response after treatment. We trained 7 machine learning models with 13,562 osteoporosis treatment instances [comprising 5080 (37.46%) inadequate treatment responses and 8482 (62.54%) adequate responses] and selected the best model (Random Forests with area under the receiver operating curve of 0.70, accuracy of 0.69, precision of 0.70, and recall of 0.89) to individually predict treatment responses of 11 therapeutic regimens, then selected the best predicted regimen to compare with the actual regimen. The results showed that the average treatment response of the recommended regimens was 9.54% higher than the actual regimens. In summary, our novel approach using a machine learning-based decision support system is capable of predicting BMD response after osteoporosis treatment and personalising the most appropriate treatment regimen for an individual patient.

scholarly journals Trends in bone mineral density among nutritional status categories of Vojvodina elderly population

2020 ◽  
Vol 148 (9-10) ◽  
pp. 577-583
Author(s):  
Zoran Gojkovic ◽  
Radmila Matijevic ◽  
Vladimir Harhaji ◽  
Branislava Ilincic ◽  
Ljubisa Barisic ◽  
...  
Keyword(s):  
Bone Mineral Density ◽  
Linear Regression ◽  
Nutritional Status ◽  
Bone Mineral ◽  
Cross Sectional Study ◽  
Normal Weight ◽  
Nutritional Risk ◽  
Regression Equations ◽  
Mineral Density ◽  
Cross Sectional

Introduction/Objective. Low bone mineral density (BMD) is commonly associated with alterations of nutritional status. The aims of the present study were to evaluate the prevalence of low BMD and its associated nutritional risk factors in Vojvodina population and to use linear regression equations to predict the BMD by using a simple marker of nutritional status, body mass index (BMI). Methods. In this retrospective, cross-sectional study, the study population included subjects who were undergoing assessment of BMD between January and December 2017, and who have met the study inclusion criteria. A total of 1974 patients (1866 women and 108 men) were included in this analysis of nutritional status according to anthropometry and BMI index, and dual-energy X-ray absorptiometry (DEXA) measurements of BMD of the femoral neck and lumbar spine. The relationship between BMI and BMD was analyzed by linear regression equation. Results. Median age was 63 (56?70) years. Considering nutritional status category, there were 40% overweight, 31% obese and 29% normal weight subjects. In most of the sample, the subjects had low BMD, 37% had osteopenia, and 25% had osteoporosis. In both bone areas we observed trends of lowering BMD as the subjects BMI decreased. Subjects with osteoporosis are more prone to BMI depended BMD changes, concerning subjects with osteopenia and normal BMD. In addition, normal weight subjects compared to overweight and obese had the highest prediction coefficients of BMI-depended changes on BMD. Conclusion. High prevalence of low BMD coexists with overweight and obese elderly females in Vojvodina. Prediction equations for the calculation of BMD can be used to evaluate the effect of BMI changes on BMD in clinical settings.

Sign in / Sign up

Export Citation Format

Share Document