Fig. 4: The prediction performance of the IPM and ACPM methods. | npj Precision Oncology

Fig. 4: The prediction performance of the IPM and ACPM methods.

From: AI-powered prediction model for neoadjuvant chemotherapy efficacy: comprehensive analysis of breast cancer histological images

Fig. 4: The prediction performance of the IPM and ACPM methods.

ad ROC plot of the IPM for predicting pCR in four datasets; eh The clinical prediction model ACPM for the four datasets centrally predicted the pCR ROC diagram; the ACPM model yielded AUC values higher than 0.7 in the four datasets but lower than those of the IPM. Especially for the V1 data, the IPM AUC was as high as 0.802, which was significantly greater than that of ACPM (p < 0.001). i, j Show the performance of IPM and ACPM in the overall validation set, respectively, and the AUC of IPM was significantly greater than that of ACPM (p < 0.001). k Shows the feature ranking of IPM. DC discovery cohort, VCs validation cohort, TR tumor region, HER2 human Epidermal Growth Factor Receptor 2, HR hormone receptor.

Back to article page