In a UK multicentre trial, an unconstrained XGBoost model incorrectly learned that higher tumour stage and carcinoma in situ predicted lower bladder cancer recurrence risk, a finding that conventional metrics such as discrimination, calibration, and SHAP failed to detect. The authors introduced a counterfactual direction test and a monotonic‑constraint framework, which removed the inversion without harming model performance and even outperformed established risk systems. The study demonstrates that such tests should be routine before deploying predictive models in clinical settings.
arXiv:2608.28592v1 Announce Type: new Abstract: Large language models (LLMs) achieve high scores on medical knowledge examinations, yet real-world oncology is not a knowledge test--it is a sequence o...
The paper introduces FCA‑Guided Counterfactual (FCA‑CF) explanations for multi‑modal breast cancer diagnosis, leveraging a Formal Concept Analysis lattice as a hard structural constraint to search for counterfactuals. On the TCGA‑BRCA dataset, FCA‑CF achieves perfect validity (100% prediction flips), the lowest average feature changes (2.37), and competitive proximity (0.900) compared to four other methods. Ablation studies show the lattice constraint drives sparsity, while a greedy refinement phase further improves results.
The paper introduces FCA‑Guided Counterfactual (FCA‑CF) explanations for multi‑modal breast cancer diagnosis, leveraging a Formal Concept Analysis lattice as a hard structural constraint to generate counterfactuals. On the TCGA‑BRCA dataset, FCA‑CF achieves perfect validity (100% prediction flips), the lowest average feature changes (2.37), and competitive proximity (0.900), outperforming four established counterfactual methods. Ablation studies show the lattice constraint and a greedy refinement phase are key to its sparsity and validity.
arXiv:2607. 08602v1 Announce Type: new Abstract: Hepatocellular carcinoma (HCC) is a common malignancy and a leading cause of cancer-related mortality.
The paper introduces Cros, a risk‑constrained stopping layer for sequential clinical diagnosis agents that determines when to stop testing and make a diagnosis. Cros combines state‑wise error ranking, policy design on disjoint development splits, and exact tests of selective diagnostic error to provide finite‑sample guarantees. On a MIMIC‑derived abdominal‑pain benchmark, Cros achieves higher state‑error AUROC and lower selective error rates compared to baseline stopping methods, though its performance varies across development resplits.
arXiv:2609.09510v1 Announce Type: cross Abstract: High-risk non-muscle-invasive bladder cancer (HR-NMIBC) carries substantial risks of recurrence and progression, while current clinical risk stratifi...
arXiv:2606. 08305v1 Announce Type: cross Abstract: Externally controlled survival trials are increasingly used when concurrent randomized controls are infeasible, particularly in oncology and rare-disease settings with time-to-event endpoints.
OpenMTB‑Audit is an open‑source benchmark that tests large language models on 500 synthetic non‑small cell lung cancer cases, covering five adversarial error categories and four safety labels: Supported, Partially Supported, Unsupported, and Insufficient Information. The study found that all eight tested LLMs over‑refused Partially Supported recommendations, collapsing labels to achieve high safety scores. A deterministic seven‑module framework, MTB‑AuditAgent, was introduced to reduce over‑refusal to 6.7% and reach 91.2% accuracy, while an oncologist annotation study highlighted disagreement around the boundary between information sufficiency and treatment optimization.
arXiv:2608. 03079v1 Announce Type: cross Abstract: Breast core needle biopsy (CNB) is central to breast cancer diagnosis yet remains challenging because limited tissue sampling, lesion heterogeneity, and subtle morphologic overlap can obscure subtype distinctions.
arXiv:2606. 19300v1 Announce Type: cross Abstract: Glioma segmentation in multiparametric MRI is a critical component of treatment planning.
The paper introduces TRUST, a threshold‑recalibrated training method that dynamically adjusts the dismissal threshold during training to penalize cancer‑positive images near the dismissal region. Evaluated on NLBS and RSNA datasets, TRUST achieved higher case‑level dismissal rates while maintaining 98% and 95% recall, outperforming a cross‑entropy baseline. External validation on RSNA→NLBS data confirmed improved dismissal rates at both recall targets, demonstrating the effectiveness of closed‑loop threshold‑aware training for selective dismissal in breast cancer screening.
arXiv:2608.21497v1 Announce Type: cross Abstract: Biochemical recurrence (BCR), defined as any detectable prostate-specific antigen level after prostatectomy with confirmatory elevation, is widely us...