Related Experiment Video
Updated: May 29, 2025
Author Spotlight: Bridging Gaps in Anatomy and Establishing a Foundation for Algorithmic Studies
Published on: December 15, 2023
Using Artificial Intelligence to Identify Three Presenting Phenotypes of Chiari Type-1 Malformation and Syringomyelia
Vivek Prakash Gupta1, Ziqi Xu2,3, Jacob K Greenberg1,3
1Department of Neurosurgery, Washington University School of Medicine in St. Louis, St. Louis , Missouri , USA.
Insights
Artificial intelligence identified three distinct patient groups for Chiari malformation type-1 with syringomyelia, aiding diagnosis and treatment. This AI approach combines data analysis and expert input to classify complex pediatric neurosurgical conditions.
Area of Science:
- Neurosurgery
- Medical Informatics
- Pediatric Neurology
Background:
- Chiari malformation type-1 (CM1) and syringomyelia (SM) are common pediatric neurosurgical conditions with varied presentations.
- Diagnosis and management of CM1 + SM pose challenges due to clinical and radiological heterogeneity.
Purpose of the Study:
- To differentiate presenting phenotypes of CM1 + SM using a novel AI clustering approach.
- To combine data-driven feature selection with clinician input for improved phenotype identification.
Main Methods:
- Utilized a registry of 1340 patients with CM1 + SM, divided into internal and external cohorts.
- Employed a data-driven feature selection process (Laplacian scores) combined with expert clinician survey input.
- Applied a K-modes clustering algorithm to identify distinct patient phenotypes.
Main Results:
- Identified 33 key features from an initial pool of 582 variables.
- Discovered an optimal of 3 distinct clusters in both internal and external cohorts.
- Characterized clusters by age at diagnosis, syrinx size, tonsil position, symptoms (headaches, bulbar), comorbidities, and motor deficits.
Conclusions:
- This study pioneers the use of AI clustering with combined data-driven and clinical feature selection for CM1 + SM phenotypes.
- The identified clusters offer a refined approach to understanding and managing CM1 + SM presentations.
Background And Objectives:
Chiari type-1 malformation (CM1) and syringomyelia (SM) are common related pediatric neurosurgical conditions with heterogeneous clinical and radiological presentations that offer challenges related to diagnosis and management. Artificial intelligence (AI) techniques have been used in other fields of medicine to identify different phenotypic clusters that guide clinical care. In this study, we use a novel, combined data-driven and clinician input feature selection process and AI clustering to differentiate presenting phenotypes of CM1 + SM.
Methods:
A total of 1340 patients with CM1 + SM in the Park Reeves Syringomyelia Research Consortium registry were split a priori into internal and external cohorts by site of enrollment. The internal cohort was used for feature selection and clustering. Features with high Laplacian scores were identified from preselected groups of clinically relevant variables. An expert clinician survey further identified features for inclusion that were not selected by the data-driven process.
Results:
The feature selection process identified 33 features (28 from the data-driven process and 5 from the clinician survey) from an initial pool of 582 variables that were incorporated into the final model. A K-modes clustering algorithm was used to identify an optimum of 3 clusters in the internal cohort. An identical process was performed independently in the external cohort with similar results. Cluster 1 was defined by older CM1 diagnosis age, small syringes, lower tonsil position, more headaches, and fewer other comorbidities. Cluster 2 was defined by younger CM1 diagnosis age, more bulbar symptoms and hydrocephalus, small syringes, more congenital medical issues, and more previous neurosurgical procedures. Cluster 3 was defined by largest syringes, highest prevalence of spine deformity, fewer headaches, less tonsillar ectopia, and more motor deficits.
Conclusion:
This is the first study that uses an AI clustering algorithm combining a data-driven feature selection process with clinical expertise to identify different presenting phenotypes of CM1 + SM.
