expainable-ai

Master's thesis prototype evaluating Human-in-the-Loop decision-making in a Cognitive Digital Twin architecture, applied to Data. Quality Incident Management. Routes incidents (auto-resolve / escalate / critical) via a RandomForest classifier with SHAP explanations, comparing AI-only, Human-only, and HITL modes. React + FastAPI microservices.

HepatiQ is an explainable AI system that uses Logistic Regression to assess five-year severe outcome risk in PBC patients. Built using the Mayo Clinic dataset of 329 patients and 17 clinical attributes, it uses SHAP to explain predictions and support transparent, data-driven healthcare assessment.

A machine learning framework for credit risk modeling, comparing traditional models with Explainable AI (XAI) approaches to satisfy regulatory compliance.

DAS-Autism is a Java-based expert system designed to support autism diagnosis using multi-valued logic, linguistic modifiers, and approximate reasoning. It employs a knowledge base of diagnostic symptoms and rules to assess symptom intensity and autism severity, providing explainable results through a graphical user interface.