{"title":"Sovereignty in Automated Stroke Prediction and Recommendation System with Explanations and Semantic Reasoning","authors":"Ayan Chatterjee","doi":"10.1016/j.procs.2025.02.079","DOIUrl":null,"url":null,"abstract":"<div><div>Personalized approaches are required for stroke management due to the variability in symptoms, triggers, and patient characteristics. An innovative stroke recommendation system that integrates automatic predictive analysis with semantic knowledge to provide personalized recommendations for stroke management is proposed by this paper. Stroke exacerbation are predicted and the recommendations are enhanced by the system, which leverages automatic Tree-based Pipeline Optimization Tool <strong>(TPOT)</strong> and semantic knowledge represented in an <strong>OWL Ontology (StrokeOnto). Digital sovereignty</strong> is addressed by ensuring the secure and autonomous control over patient data, supporting data sovereignty and compliance with jurisdictional data privacy laws. Furthermore, classifications are explained with Local Interpretable Model-Agnostic Explanations <strong>(LIME)</strong> to identify feature importance. Tailored interventions based on individual patient profiles are provided by this conceptual model, aiming to improve stroke management. The proposed model has been verified using public stroke dataset, and the same dataset has been utilized to support ontology development and verification. In TPOT, the best <strong>Variance Threshold + DecisionTree Classifier</strong> pipeline has outperformed other supervised machine learning models with an accuracy of <strong>95.2%,</strong> for the used datasets. The Variance Threshold method reduces feature dimensionality with variance below a specified threshold of 0.1 to enhance predictive accuracy. To implement and evaluate the proposed model in clinical settings, further development and validation with more diverse and robust datasets are required.</div></div>","PeriodicalId":20465,"journal":{"name":"Procedia Computer Science","volume":"254 ","pages":"Pages 201-210"},"PeriodicalIF":0.0000,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Procedia Computer Science","FirstCategoryId":"1085","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S1877050925004296","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
Abstract
Personalized approaches are required for stroke management due to the variability in symptoms, triggers, and patient characteristics. An innovative stroke recommendation system that integrates automatic predictive analysis with semantic knowledge to provide personalized recommendations for stroke management is proposed by this paper. Stroke exacerbation are predicted and the recommendations are enhanced by the system, which leverages automatic Tree-based Pipeline Optimization Tool (TPOT) and semantic knowledge represented in an OWL Ontology (StrokeOnto). Digital sovereignty is addressed by ensuring the secure and autonomous control over patient data, supporting data sovereignty and compliance with jurisdictional data privacy laws. Furthermore, classifications are explained with Local Interpretable Model-Agnostic Explanations (LIME) to identify feature importance. Tailored interventions based on individual patient profiles are provided by this conceptual model, aiming to improve stroke management. The proposed model has been verified using public stroke dataset, and the same dataset has been utilized to support ontology development and verification. In TPOT, the best Variance Threshold + DecisionTree Classifier pipeline has outperformed other supervised machine learning models with an accuracy of 95.2%, for the used datasets. The Variance Threshold method reduces feature dimensionality with variance below a specified threshold of 0.1 to enhance predictive accuracy. To implement and evaluate the proposed model in clinical settings, further development and validation with more diverse and robust datasets are required.