This study explored the hidden biomedical information from knee MR images for osteoarthritis prediction. We have computed the Cartilage Damage Index (CDI) information from 36 informative locations on tibiofemoral compartment from 3D MR imaging reconstruction and used PCA analysis to process the feature set. The processed feature set and original raw feature set were severed as input to four machine learning methods (artificial neural network (ANN), support vector machine (SVM), random forest and naïve Bayes) respectively. To examine the different effect of medial and lateral informative locations, we have divided the 36-dimensional feature set into 18-dimensional medial feature set and 18-dimensional lateral feature set and run the experiment on four classifiers separately. Experiment results showed that the medial feature set generated better prediction performance than the lateral feature set, while using the total 36-dimensional feature set generated the best. PCA analysis is helpful in feature space reduction and performance improvement. The best performance was achieved by ANN with area under the receiver operating characteristic (ROC) curve = 0.761 and F-measure = 0.714. Experiment results indicated that the informative locations on medial tibiofemoral compartment contain more valuable information than informative locations on lateral tibiofemoral compartment, for OA severity prediction. Therefore, to improve the design of the clinically used CDI, it could be considered to select more points from the medial tibiofemoral compartment while reduce the number of points selected from the lateral tibiofemoral compartment.