Skip to content

Read-only version: For security reasons, the Research Information System (FIS) has been temporarily replaced by this read-only version. All public content and full texts remain available; login, editing, submission and statistics are not available. Questions: fis@uni-bamberg.de

Home › Publications › Symbolic AI in Critical Healthcare Setting : Towards Automated Governance of Spontaneous Breathing Trials for Streamlined Extubation

Publication

Symbolic AI in Critical Healthcare Setting : Towards Automated Governance of Spontaneous Breathing Trials for Streamlined Extubation

Stößel, Patrick Salvatore (2025): Symbolic AI in Critical Healthcare Setting : Towards Automated Governance of Spontaneous Breathing Trials for Streamlined Extubation, Bamberg: Otto-Friedrich-Universität, doi: 10.20378/irb-110690.

Details

Faculty/Chair
Faculty/Professorship
Author
Stößel, Patrick Salvatore
Corporate Body
Universität Bamberg
Publisher Information
Otto-Friedrich-Universität : Bamberg
Year of publication
2025
Pages
XII, 133, Tabellen, Diagramme
Supervisor
Language
English
Remark

Masterarbeit, Otto-Friedrich-Universität Bamberg, 2025

DOI
Licence
URN
Abstract

This thesis presents a novel approach to clinical decision support in critical care by leveraging symbolic artificial intelligence (AI) for the formalization and automated verification of Spontaneous Breathing Trial (SBT) readiness assessments. It introduces a prototype system that transforms Intensive Care Unit (ICU) checklist logic into machine-verifiable higher-order logic (HOL) using Isabelle/HOL, ensuring correctness-by-construction, traceability, and semantic fidelity. The system was implemented in Java and supports modular, extensible workflows encompassing Health Level 7 (HL7)-compatible data ingestion, logic generation, and automated proof execution. A simulated environment was used to evaluate the prototype’s performance across clinical and technical dimensions. Results demonstrate consistent and explainable classifications aligned with clinical expectations, and profiling indicates linear scalability with respect to patient count and checklist complexity. Qualitative analysis highlights the system’s interpretability, workflow compatibility, and auditability. Limitations concerning real-world integration, proof traceability, and regulatory readiness are critically examined. The work establishes a reusable methodology for the formalization of protocol-based reasoning and outlines future directions including live deployment, integration with subsymbolic models, and regulatory certification. This research contributes a technically sound and ethically aligned foundation for the development of transparent, verifiable, and adaptable AI systems in healthcare.

GND Keywords
Erklärbare künstliche Intelligenz
Verifikation
Logik
Softwareentwicklung
Medizin
Keywords
Symbolische Künstliche Intelligenz
Erklärbarkeit
Verifizierbarkeit
Höherstufige Logik
Medizinanwendung
DDC Classification
004 Computer science
RVK Classification
ST 302
Type
Masterthesis
URI
Activation date
2026-01-12

Permalink: https://fis.uni-bamberg.de/handle/uniba/110690 · Data as JSON · Cite: BibTeX · RIS

Data as of: 2026-11-16

Restricted read-only access: login, editing and submission are temporarily disabled.