Related Experiment Video
Updated: Dec 31, 2025
Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
Novel classifiers with clinical laboratory parameters for early detection of osteosarcoma
Lin-Lin Cao1, Zhaoming Chen2, Zhihong Yue1
1Department of Clinical Laboratory, Peking University People's Hospital, Beijing, China.
Background:
Osteosarcoma (OS) is one of the most common malignant bone tumors. It is essential to explore early diagnostic indicators with high sensitivity and specificity due to the rapid progression and metastasis of OS and the poor survival of metastatic OS patients. However, a few indicators of diagnostic significance have been described.
Methods:
A total of 458 OS patients, 312 healthy individuals, and 228 patients with primary benign bone lesions were included. Logistic regression was performed on 46 clinical laboratory parameters to establish the diagnostic classifiers, which were evaluated by analysis of the receiver operating characteristic (ROC) curves.
Results:
We established three diagnostic classifiers, called Cos for all ages, Clos for low ages, and Chos for high ages, with clinical laboratory parameters to distinguish OS from healthy individuals. All classifiers showed better diagnostic performances than alkaline phosphatase (ALP) in the independent validation cohort. In addition, these classifiers had better ability than ALP to discriminate OS from primary benign bone lesions. Furthermore, Cos , Clos, and Chos had larger AUC than ALP to identify small-size and early-stage OS and could also detect ALP-negative OS effectively.
Conclusion:
Our study suggests the potential of Cos , Clos , and Chos as non-invasive biomarkers for early OS.
