AI-Augmented Genomic Screening: Leveraging Vision-Language Models for Early Disorder Detection

Abstract

This project, titled “ AI-Augmented Genomic Screening: Leveraging Vision-Language Models for Early Disorder Detection”, presents a robust intelligent system designed to predict the likelihood of multiple genetic disorders using structured medical data. Leveraging multi-label classification with LightGBM, the system achieves over 84% accuracy in identifying up to nine disorders simultaneously, addressing the clinical need for early and comprehensive diagnosis. The solution integrates a modular pipeline covering preprocessing, automated training, and real-time inference. Inputs such as patient age, symptoms, blood test results, and family history are processed to generate predictions, which are visualized through an intuitive Streamlit interface using animated charts. Technologies like Python, Scikit-learn, LightGBM, and SHAP contribute to the system’s accuracy and explainability. The final model enables interpretable and actionable insights, supporting better clinical decisions and early interventions.

Ähnliche Arbeiten