Fig. 5: The multimodal platform represented by a nomogram and its predictive performance. | npj Precision Oncology

Fig. 5: The multimodal platform represented by a nomogram and its predictive performance.

From: Tumor cell- and infiltrating immune cell-based supervised learning artificial intelligence multimodal platform for tumor prognosis

Fig. 5

a The multimodal platform, combining clinical parameters and pathomics signatures, was represented as a nomogram. The output results are the probability of survival for less than one year (1- ‘1-year OS probability’), three years (1- ‘3-year OS probability’), and five years (1- ‘5-year OS probability’), respectively. Kaplan–Meier analysis revealed the performance of the multimodal platform in the training (b) and validation (e) cohorts. Time-dependent ROC analysis evaluated the predictive accuracy of the multimodal platform for 1-year and 3-year OS in the training (c, d) and validation (f, g) cohorts. PathS pathomics-based AI platform developed through supervised deep learning, OS overall survival, c-index concordance index, ROC receiver operating characteristic curve, CS clinical signatures, AUC area under the curve, CI confidence interval.

Back to article page