Reasoning and causal inference regarding surgical options for patients with low‐grade gliomas using machine learning: A SEER‐based study

Abstract

BackgroundDue to the heterogeneity of low‐grade gliomas (LGGs), the lack of randomized control trials, and strong clinical evidence, the effect of the extent of resection (EOR) is currently controversial.AimTo determine the best choice between subtotal resection (STR) and gross‐total resection (GTR) for individual patients and to identify features that are potentially relevant to treatment heterogeneity.MethodsPatients were enrolled from the SEER database. We used a novel DL approach to make treatment recommendations for patients with LGG. We also made causal inference of the average treatment effect (ATE) of GTR compared with STR.ResultsThe patients were divided into the Consis. and In‐consis. groups based on whether their actual treatment and model recommendations were consistent. Better brain cancer‐specific survival (BCSS) outcomes in the Consis. group was observed. Overall, we also identified two subgroups that showed strong heterogeneity in response to GTR. By interpreting the models, we identified numerous variables that may be related to treatment heterogeneity.ConclusionsThis is the first study to infer the individual treatment effect, make treatment recommendation, and guide surgical options through deep learning approach in LGG research. Through causal inference, we found that heterogeneous responses to STR and GTR exist in patients with LGG. Visualization of the model yielded several factors that contribute to treatment heterogeneity, which are worthy of further discussion.

Individualized survival prediction and surgery recommendation for patients with glioblastoma

“…The prediction and explanation of ITE from censored time-to-event outcomes have received little attention in the data science domain ( 19 , 30 ), which is surprising when one considers the enormous practical relevance of the subject ( 31 , 32 ). The BITES framework uses strong ignorability ( 33 ) to remove confounding artifacts ( 34 ) and IPM to sufficiently balance the generating distributions of treatment groups on both latent representations ( 35 , 36 ) and covariates ( 37 ).…”

Personalized surgical recommendations and quantitative therapeutic insights for patients with metastatic breast cancer: Insights from deep learning

“…As our research progresses, we aim to refine the DL model, broadening its applicability to a wider range of diseases [ 46 , 47 ]. Moreover, the development of user‐friendly client software for clinical use is anticipated.…”

Optimizing adjuvant treatment options for patients with glioblastoma

“…T-learner excludes some confounding artifacts; however, it can still be affected by inconsistent predictive performance and biased treatment allocation ( 14 ). To address this issue, we utilized Balanced Individual Treatment Effect for Survival data (BITES) ( 20 ), one of the recently proposed DL models capable of making individual-level causal inferences, so as to predict each patient’s ITE and to make treatment recommendations for GBM patients ( 24 ). BITES combines both representation-based and CATE-based causal inference methods, therefore providing more unbiased ITE inferences.…”