Auditing Sybil: Explaining Deep Lung Cancer Risk Prediction Through Generative Interventional Attributions

2026-05-13 · ICML 2026 · anchor

Note

date and anchor point to v2, the version held in the project and verified against

Findings

• SY-001Objects outside the patient, gown snaps and ECG electrodes, drive Sybil's risk predictions
• SY-002Sybil responds more weakly to nodules near the pleura, where adenocarcinoma tends to appear
• SY-003Sybil processes pulmonary nodules almost additively, with limited pairwise interactions