A study investigated the possibility of predicting the risk of bone fractures to help with early diagnosis and prevention of postmenopausal osteoporosis.
Affecting over two million Canadians, osteoporosis is known as the “silent thief”.
Increased bone fragility due to deterioration of bone tissue and loss of bone mass, the hallmark characteristics of osteoporosis, can occur for many years without any symptoms showing.
Typically, diagnosis occurs after an osteoporotic fracture occurs. In Canada alone, one in three women will suffer at least one osteoporotic fracture, with the most common osteoporotic fractures occurring in either the spine, hip, wrist, or shoulder.
Given the established relationship between menopause and osteoporosis, predicting future bone mass loss and deterioration to help identify women at high risk of postmenopausal osteoporosis is of significant clinical importance.
The ultimate goal of treatment management is the prevention of future fractures.
By predicting future bone mineral density which is responsible for 70% of bone strength and bone loss rate, early detection and diagnosis would be possible.
This would enable treatment to be customized for high-risk women and therefore reduce risk factors for the severity and incidence of fractures which is particularly important in an aging society.
Use of Artificial Neural Networks
A recent population-based cohort study by Japanese researchers published in BMC Research Notes investigated whether the use of artificial neural networks could be used to predict future bone mineral density and the rate of bone loss in postmenopausal Japanese women.
Previous studies have shown artificial neural networks (biologically inspired computing systems based on the neural brain structure) to be superior to conventional models used to predict bone mineral density in postmenopausal women.
However, none of these previous studies have evaluated the use of artificial neural networks to predict potential future bone mineral density or the rate of bone loss in postmenopausal women.
An artificial neural network statistical model using information collected from 135 female Japanese participants aged 50 years or older was constructed to potentially predict bone mineral density and bone loss rate in 10 years’ time in lumbar (lower spine) and femoral (thigh) sites.
The model included eleven input variables including age, weight, height, age at menopause, age at menstruation first occurred, and others.
Bone mineral density was measured using dual-energy X-ray absorptiometry.
There were two output variables, which included bone mineral density at either the lumbar site or proximal femur site (hip region) and the rate of bone loss.
Predicting Ten Years into the Future
The results showed this statistical model could indeed predict in 10 years future bone mineral density and bone loss rate on an individual basis using the input variables, which are considered conventional parameters which are easy to obtain.
The discovery of this new tool to predict potential high-risk individuals for postmenopausal osteoporosis is of significant importance, particularly in an aging society, and could be used to reduce the incidence of bone fractures.
Furthermore, as the model enables analysis on an individual basis, this means treatment could also be individually customized which could also help with the prevention and early diagnosis of postmenopausal osteoporosis.
The researchers noted that this study included several limitations.
One limitation included the prediction of the study could possibly only apply to women with osteoporosis living in similar rural regions of Japan to those examined in the study due to the artificial neural network processing data so precisely.
It is possible the data may not apply to women living in largely populated cities in Japan.
A second limitation included using only a small sample size and the omission of important variables such as daily physical activity which is believed to have an impact on bone mineral density and the rate of bone loss.
A Potentially New Diagnostic Tool
Consequently, future studies including larger cohorts from various countries and including variables such as physical activity are essential.
However, this study has shown significant advances in the use of statistical analysis models for predicting future bone mineral density.
Meaning, artificial neural networks could be a potential new diagnostic tool for the early diagnosis and intervention of patients at high risk of bone fractures due to bone fragility from postmenopausal osteoporosis.
This, in turn, could potentially increase these patients’ quality of life, as bone fractures, particularly of the hip and lower spine, can significantly impair their quality of life and require increased medical care which can be expensive and have a negative impact on the patient.
Written by Lacey Hizartzidis, PhD
Relevant topics that may be of interest to you:
- Treating Osteoporosis in Men
- Osteoporotic Fracture and Bisphosphonates: What are the Long-Term Risks?
- Exploring the Impact of Carotenoids on Osteoporotic Fractures
- Stopping osteoporosis treatment may increase the risk of vertebral fracture
- Do high blood glucose levels increase your risk of osteoporotic fracture?
- Does A high soy diet influence osteoporotic fracture in breast cancer survivors?
- Monitoring and improving bone mineral density when you have osteoporosis
Shioji M, Yamamoto T, Ibata T, Tsuda T, Adachi K and Yoshimura N.Artificial neural networks to predict future bone mineral density and bone loss rate in Japanese postmenopausal women. BMC Research Notes201710:590.doi:10.1186/s13104-017-2910-4.
Osteoporosis Canada website https://osteoporosis.ca/. Accessed November 20th, 2017.