Outcome Classification of Spontaneous Bleeding in Meningiomas Using a Supervised Machine Learning Model

Resumo

Introdução: Os meningiomas são tumores benignos; porém, quando associados à hemorragia intracraniana espontânea, uma complicação rara e grave, representam ameaça significativa à sobrevivência. Devido à baixa incidência, as diretrizes clínicas são escassas e a predição de desfechos desafiadora. Objetivo: Desenvolver e avaliar uma abordagem de aprendizado de máquina supervisionado para prever desfechos binários (morbimortalidade ou recuperação) em meningiomas hemorrágicos. Métodos: Um conjunto de dados foi construído a partir de relatos publicados na literatura médica e validados por especialistas em neurocirurgia. Algoritmos de classificação foram aplicados e o desempenho comparado. Resultados: O AdaBoost apresentou os melhores resultados: 85% de acurácia, 97% de recall para morbimortalidade e 68% para recuperação. A análise de importância das variáveis identificou déficit focal pré-operatório, volume tumoral e do sangramento, idade, nível de consciência e categoria da Escala de Glasgow como principais preditores. Conclusão: Além de propor um modelo preditivo clinicamente relevante, o estudo disponibiliza conjunto curado com 294 casos e 15 características clínicas, oferecendo à comunidade médica recurso para pesquisas. A abordagem demonstra boas práticas em engenharia de atributos, pré-processamento e otimização de modelos, apoiando a tomada de decisão em casos neurocirúrgicos de alto risco e aprimorando a estratificação de pacientes e o planejamento de cuidados.

Palavras-chave

Cérebro; Meningiomas; Sangramento; Espontâneo; Hemorragia; Desfecho

Abstract

Introduction: Meningiomas are typically benign tumors, however, when associated with spontaneous intracranial hemorrhage, a rare but severe complication, they pose a significant threat to survival. Due to the low incidence of this condition, clinical guidelines are scarce, and outcome prediction remains difficult. Objective: To develop and evaluate a supervised machine learning approach for predicting binary outcomes (Morbimortality or Recovery) in hemorrhagic meningiomas. Methods: A novel dataset was constructed from curated case reports published in the medical literature and manually validated by neurosurgery specialists. Classification algorithms were applied, and performance was compared. Results: The AdaBoost classifier achieved the best results, with 85% Accuracy, 97% Recall for the Morbimortality class, and 68% for Recovery. Feature importance analysis identified preoperative focal deficit, tumor and hemorrhage volume, patient age, level of consciousness, and Glasgow Scale category as the most influential predictors. Conclusion: In addition to proposing a clinically relevant predictive model, this study contributes a curated dataset of 294 cases and 15 clinical features, offering a valuable resource for future research. The approach demonstrates best practices in medical feature engineering, preprocessing, and model optimization, supporting clinical decision-making in high-risk neurosurgical cases and improving patient stratification and personalized care.

Keywords

Brain; Meningiomas; Bleeding; Spontaneous; Hemorrhagic; Outcomes

References

1. Buerki RA, Horbinski CM, Kruser T, Horowitz PM, James CD, Lukas RV. An overview of meningiomas. Future Oncol. 2018;14(21):2161-77. https://doi.org/10.2217/fon-2018-0006. PMid:30084265.

2. Goran A, Cimenello VJ, Fisher RG. Hemorrhage into

meningiomas. Arch Neurol. 1965;13(1):65-9. https://doi.org/10.1001/ archneur.1965.00470010069009. PMid:14314276.

3. Pressman E, Penn D, Patel NJ. Intracranial hemorrhage from meningioma: 2 novel risk factors. World Neurosurg. 2020;135:217-21. https://doi.org/10.1016/j.wneu.2019.10.173. PMid:31698125.

4. Maggio I, Franceschi E, Tosoni A, et al. Meningioma: not always a benign tumor. A review of advances in the treatment of meningiomas. CNS Oncol. 2021;10(2):CNS72. https://doi.org/10.2217/cns-2021-0003. PMid:34015955.

5. Pereira BJA, de Almeida AN, Paiva WS, de Aguiar PHP, Teixeira MJ, Marie SKN. Assessment of hemorrhagic onset on meningiomas: systematic review. Clin Neurol Neurosurg. 2020;199:106175. https:// doi.org/10.1016/j.clineuro.2020.106175. PMid:33027728.

6. Navi BB, Reichman JS, Berlin D, et al. Intracerebral and subarachnoid hemorrhage in patients with cancer. Neurology. 2010;74(6):494-501. https://doi.org/10.1212/WNL.0b013e3181cef837. PMid:20142616.

7. Kulkarni VY, Sinha PK. Pruning of random forest classifiers: A survey and future directions. In: 2012 International Conference on Data Science Engineering (ICDSE); 2012; Cochin, India. Manhattan: IEEE; 2012. p. 64-8. https://doi.org/10.1109/ICDSE.2012.6282329.

8. Naidu G, Zuva T, Sibanda EM. A review of evaluation metrics in machine learning algorithms. In: Silhavy R, Silhavy P, editors. Artificial intelligence application in networks and systems. Cham: Springer International Publishing; 2023. p. 15-25. https://doi.org/10.1007/978-3-031-35314-7_2.

9. Pienaar JA, Varghese J. Intracranial meningiomas at a tertiary hospital: spectrum of MRI findings with histopathologic correlation. SA J Radiol. 2024;28(1):2812. https://doi.org/10.4102/sajr.v28i1.2812. PMid:38628266.

10. Ferrufino-Mejia BR, Rodríguez-Rubio HA, López-Rodríguez R, et al. Spontaneous intracranial hemorrhage associated with an intracranial meningioma. Cureus. 2023;15(6):e40472. https://doi.org/10.7759/ cureus.40472. PMid:37456393.

11. Angonese S, Galante R. Composition of heterogeneous node embeddings - unlocking the power of heterogeneous graph representation. In: Anais-XXXIX Simpósio Brasileiro de Bancos de Dados; 2024; Florianópolis. Porto Alegre: Sociedade Brasileira de Computação; 2024. p. 626-638. https://doi.org/10.5753/sbbd.2024.243436.

12. Siedlecki Z, Nowak K, Grzyb S, Śniegocki M. Intracranial hemorrhages associated with meningiomas: own experience and literature review.

J Cancer Res Ther. 2022;18(3):820-6. https://doi.org/10.4103/jcrt. JCRT_958_20. PMid:35900566.

13. Halalmeh DR, Alrashdan M, Kharouf M, Sbeih I, Molnar PT, Moisi MD. Brain meningiomas manifesting as intracranial hemorrhage: comprehensive systematic review and report of the first case of hemorrhagic meningiomatosis. World Neurosurg. 2023;169:73-86.e6. https://doi.org/10.1016/j.wneu.2022.10.113. PMid:36332779.

14. Leclerc A, Gohel H, Malczuk J, Anzalone L, Emery E, Gaberel T. Systematic review of meningiomas revealed by spontaneous intracranial hemorrhage: clinicopathological features, outcomes, and rebleeding rate. World Neurosurg. 2023;172:e625-39. https:// doi.org/10.1016/j.wneu.2023.01.100. PMid:36738963.

15. Pedregosa F, Varoquaux G, Gramfort A, et al. Scikit-learn: machine learning in python. J Mach Learn Res. 2011;12:2825-30.