A Deep Neural Network-Based Method for Early Detection of Osteoarthritis Using Statistical Data.

Area of Science:

• Medical informatics
• Artificial intelligence in healthcare
• Osteoarthritis research

Background:

• Osteoarthritis (OA) affects many, particularly knee, hip, and spine.
• Current diagnosis relies on manual image inspection, which is time-consuming.
• Image-based AI detection requires hospital visits and medical images.

Purpose of the Study:

• To develop a method for predicting osteoarthritis occurrence using non-imaging data.
• To leverage medical utilization and health behavior data for OA prediction.
• To create a more accessible and efficient prescreening tool for osteoarthritis.

Main Methods:

• Utilized a deep neural network model.
• Employed Principal Component Analysis (PCA) with quantile transformer scaling for feature generation.
• Trained the model on statistical data from 5749 subjects, including medical utilization and health behavior information.

Main Results:

• The proposed deep neural network with scaled PCA achieved an Area Under the Curve (AUC) of 76.8%.
• The method effectively minimized the effort required for feature generation.
• Demonstrated the feasibility of predicting OA using statistical patient data.

Conclusions:

• The developed method is a promising tool for prescreening osteoarthritis.
• It can help reduce healthcare costs and time spent by patients in hospitals.
• Facilitates proactive and preventive medical care for diverse forms of OA.