Table 1.
Research on patent value prediction based on structured patent features.
Fig 1.
Overall process of the proposed approach.
Table 2.
Biomedical textiles patent information search.
Fig 2.
Data processing flow.
Fig 3.
(a) Average annual number of forward citations for biomedical fabric patents; (b) Grading the average annual number of forward citations for biomedical fabric patents.
Table 3.
Training and validation sets.
Fig 4.
Stacking strategy.
Fig 5.
Five-fold cross validation.
Table 4.
Performance matrices for other machine learning models.
Table 5.
Results of XGBoost evaluation (Accuracy and F-1 score).
Table 6.
Confusion matrix of BioTexVal.
Table 7.
Feature importance ranking (Top20).
Fig 6.
Contribution of base models to BioTexVal.
Fig 7.
Comparison of feature variances in correct vs. incorrect sample classifications.
Table 8.
Features with significant variance discrepancies affecting classification accuracy.