Can AI determine whether or not someone has osteoporosis? Newly developed deep learning model shows promise

Osteoporosis is called a “silent disease” because it is very difficult to detect in the early stages. What if artificial intelligence could help predict the likelihood of a patient developing osteoporosis before they even visit a doctor?

Tulane University researchers have made progress toward realizing that vision by developing a new deep learning algorithm that outperforms existing computer-based methods for predicting osteoporosis risk, potentially leading to earlier diagnosis and better outcomes for patients at risk of osteoporosis.

Their findings recently came to light: The cutting edge of artificial intelligence.

Deep learning models have gained attention for their ability to mimic human neural networks and find trends in large datasets without being specially programmed. Researchers tested a deep neural network (DNN) model against four traditional machine learning algorithms and a traditional regression model using data from more than 8,000 participants over the age of 40 from the Louisiana Osteoporosis Study. DNN achieved the best overall predictive performance, measured by scoring each model's ability to identify true positives and avoid mistakes.

“The earlier osteoporosis risk is detected, the more time patients have to take preventive measures,” said Chuang Qiu, lead author of the paper and a research assistant professor in the Center for Biomedical Informatics and Genomics at Tulane University School of Medicine. “We are pleased to see that our DNN model outperformed other models in accurately predicting osteoporosis risk in an aging population.”

After testing their algorithm on a large sample of real-world health data, the researchers were also able to identify the 10 most important factors in predicting osteoporosis risk: weight, age, sex, grip strength, height, beer intake, diastolic blood pressure, alcohol consumption, years of smoking, and income level.

Notably, the simplified DNN model using these top 10 risk factors performed nearly as well as the full model including all risk factors.

Chiu acknowledged that there is still a lot of work to be done before AI platforms can be used by the general public to predict an individual's risk of osteoporosis, but said identifying the benefits of deep learning models is a step in that direction.

“Our ultimate goal is for people to be able to enter their information, receive a highly accurate osteoporosis risk score, and then receive treatment to strengthen their bones and reduce further damage,” Qiu said.