Doctor-in-the-Loop: An Explainable, Multi-View Deep Learning Framework for Predicting Pathological Response in Non-Small Cell Lung Cancer

2025-02-21

AI Summary

Existing AI models for predicting pathological response after non-small-cell lung cancer (NSCLC) resection suffer from poor interpretability and insufficient clinical alignment. Method: We propose a multi-view progressive deep learning framework integrating clinical prior knowledge. Innovatively, we establish a “clinician-in-the-loop” paradigm by embedding expert-derived rules throughout training; design a clinical-prior-guided loss function and attention regularization mechanism; and generate anatomically traceable heatmaps via Grad-CAM. A progressive multi-view CNN dynamically shifts focus from macroscopic anatomical context to lesion-level details. Contribution/Results: Evaluated on a real-world NSCLC cohort, our model achieves an AUC of 0.92—significantly outperforming baseline methods. Clinician evaluation confirms high credibility of the generated heatmaps, demonstrating strong predictive accuracy alongside clinically meaningful interpretability.

Technology Category

• Machine Learning: Multi-instance/Multi-view Learning
• Computer Vision: Multi-modal Vision
• Natural Language Processing: Interpretability, Analysis, and Evaluation of NLP Models

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

Non-small cell lung cancer (NSCLC) remains a major global health challenge, with high post-surgical recurrence rates underscoring the need for accurate pathological response predictions to guide personalized treatments. Although artificial intelligence models show promise in this domain, their clinical adoption is limited by the lack of medically grounded guidance during training, often resulting in non-explainable intrinsic predictions. To address this, we propose Doctor-in-the-Loop, a novel framework that integrates expert-driven domain knowledge with explainable artificial intelligence techniques, directing the model toward clinically relevant anatomical regions and improving both interpretability and trustworthiness. Our approach employs a gradual multi-view strategy, progressively refining the model's focus from broad contextual features to finer, lesion-specific details. By incorporating domain insights at every stage, we enhance predictive accuracy while ensuring that the model's decision-making process aligns more closely with clinical reasoning. Evaluated on a dataset of NSCLC patients, Doctor-in-the-Loop delivers promising predictive performance and provides transparent, justifiable outputs, representing a significant step toward clinically explainable artificial intelligence in oncology.

Problem

• Predict pathological response in NSCLC
• Enhance explainability of AI predictions
• Integrate clinical knowledge with AI techniques

Innovation

• Doctor-in-the-Loop framework
• Explainable AI techniques
• Multi-view strategy