Machine Learning Predicting Long-term Survival and Length of Stay (LOS) Following the Treatment of Spinal Cord Astrocytomas: A National Analysis of 1,143 Patients

Introduction: Spinal astrocytomas have indistinguishable margins placing sometimes the role of surgery in question in favor of radiation and chemotherapy. Our study aims to investigate the effect of surgery, radiation, and chemotherapy on patients diagnosed with spinal astrocytomas and to determine the predictors of overall survival and postoperative length of stay(LOS).

Methods: We queried the National Cancer Database between 2004-2016 for patients diagnosed with spinal astrocytomas. Five-year survival for each treatment was compared using Kaplan-Meier(KM) curves based on the log-rank test. Multivariate analysis was performed to determine survival determinants. Supervised machine learning(ML) models were trained to predict survival and model performance. Shapley additive explanations (SHAP) were calculated using the best-performing model for feature importance.

Results: We found a total of 1,143 patients with spinal astrocytomas of which 74.3% had surgical resection;44.3% received radiation and 25.5% received chemotherapy. Five-year survival rate was 64.9% and 30-day readmission rate was 5.2%. Age(p < 0.001), radiation(p=0.001); chemotherapy(p < 0.01), and high comorbidity score(0.02) were most likely to predict lower risk of 5-year survival. Age(p=0.004), higher number of days between the diagnosis and definitive treatment(p < 0.01) were more likely to predict higher LOS. Kaplan-Meier survival curves revealed that survival was the best following maximal safe resection(p < 0.005) without radiation or chemotherapy, which was also true for recurrent tumors. The best performing ML models were Random Forest(RF) and Decision tree(DT) (99.9%). ML using RF revealed that radiation alone and chemotherapy alone had the greatest impact on survival leading to greater mortality.

Conclusion : Five-year mortality was increased by older age, radiation therapy, chemotherapy, and high comorbidity score, while higher LOS was increased by older age and delay in treatment. Artificial Intelligence showed that radiation alone and chemotherapy alone lead to higher mortality compared to surgery which indicated that surgery has a fundamental role in treating spinal cord astrocytomas and improving long-term survival.