Photo Credit: AND-ONE
The following is a abstract of “Clinical utility of machine studying fashions in sufferers with prostate most cancers earlier than prostatectomy,” revealed in the February 2024 concern of Oncology by Guerra et al.
This retrospective observational examine aimed to assemble machine studying predictive fashions to evaluate surgical threat associated to extracapsular extension (ECE) in prostate most cancers (PCa) sufferers earlier than radical prostatectomy. Two unbiased datasets have been analyzed: one comprising 139 contributors from a single establishment (used for coaching) and one other consisting of 55 sufferers from 15 completely different establishments (utilized for exterior validation), all handled with Robotic Assisted Radical Prostatectomy (RARP). Five machine studying fashions, incorporating numerous mixtures of medical, semantic (interpreted by a radiologist), and radiomics options derived from T2W-MRI photos, have been developed to foretell extracapsular extension in prostatectomy specimens (pECE+).
Decision curve evaluation (DCA) plots have been employed to guage the fashions’ web profit in assigning sufferers to prostatectomy with both non-nerve-sparing surgical procedure (NNSS) or nerve-sparing surgical procedure (NSS) primarily based on the expected ECE standing. The rankings of fashions derived from DCA have been in contrast with these derived from the receiver working attribute (ROC) space beneath the curve (AUC). Results indicated that the mannequin integrating medical, semantic, and radiomics options demonstrated the very best web profit values throughout related threshold chances in the coaching information, with related efficiency noticed in the exterior validation set.
However, the mannequin rating primarily based on AUC differed between the invention teams and favored the mannequin using solely medical and semantic options. In conclusion, the mixed mannequin incorporating medical, semantic, and radiomic options reveals promise in predicting pECE+ in PCa sufferers, providing a optimistic web profit when figuring out between prostatectomy with NNSS or NSS.
Source: cancerimagingjournal.biomedcentral.com/articles/10.1186/s40644-024-00666-y
https://www.physiciansweekly.com/utilization-of-machine-learning-models-in-pre-prostatectomy-patients-with-prostate-cancer/
