M. Nabi

Open access Aug 2026

Toward AI-Assisted Precision Diagnostics in Breast Cancer: Source-Group-Preserving Evaluation of Class-Imbalance Strategies for Ordered ER-IHC Segmentation

Background/Objectives: Estrogen receptor immunohistochemistry (ER-IHC) exhibits heterogeneous staining and substantial class imbalance, complicating segmentation of ordered expression categories. We evaluated whether class weighting, minority-focused crop sampling, adaptive minority curriculum (AMC), and Focal–Tversky optimization improved a common ResUNet-DS backbone for segmenting background/non-target pixels and C1 ER-negative, C2 weak-positive, C3 moderate-positive, and C4 strong-positive foreground categories. Methods: The dataset comprised 220 paired 512×512 image–mask patches organized into 44 recovered five-patch source groups. A source-group-preserving nested five-fold design used outer folds of 45, 45, 45, 45, and 40 patches. Six controlled training conditions were compared, with four independently selected inner models ensembled for each condition and outer fold. Results: Unweighted random training achieved the best numerical mean for all five primary endpoints: foreground Dice (0.7479±0.0123), C2–C4 Dice (0.7098±0.0128), foreground IoU (0.6074±0.0147), foreground quadratically weighted kappa (0.9761±0.0040), and foreground-ordinal MAE (0.0446±0.0062). Weighted random training was the strongest weighted/minority-sensitive condition but did not exceed the unweighted reference. None of 25 paired outer-fold comparisons reached p<0.05; the minimum raw p-value was 0.0625, and all Holm-adjusted p-values were 0.3125. A 20,000-replicate paired source-group bootstrap preserved the same overall ordering. Conclusions: More complex imbalance-handling strategies did not improve aggregate segmentation over unweighted training. Post hoc analyses showed scope-dependent probability quality and error ranking. IHC4BC provided expression-ordering consistency but not direct external segmentation validation. The findings support preliminary source-group-preserving methodological evidence rather than patient-level, whole-slide, or clinical validity.

Md Saiful Arefin, Mohammad Saiful Islam, M. Nabi et al. · Diagnostics · 0 citations

Open access Jul 2026

Machine learning for chemotherapy decision-making in breast cancer using large language model

Introduction Breast cancer chemotherapy decision-making remains challenging due to biological heterogeneity and variability in clinical practice. This study proposes a hybrid framework integrating machine learning (ML), causal reasoning, and large language models (LLMs) to improve treatment recommendations. Methods Using the METABRIC dataset, eleven pre-treatment clinicopathologic variables were selected. A Random Forest classifier was developed and compared with baseline ML models. Individualized treatment benefit was estimated through inverse probability-weighted causal survival analysis, while GPT-4 was employed using few-shot prompting to generate clinical rationales. Results The Random Forest achieved an AUC of 0.91, outperforming benchmark models. Causal analysis identified heterogeneous treatment benefits and patient groups where chemotherapy could potentially be deprioritized. GPT-4 showed moderate agreement with the Random Forest (Cohen's κ = 0.13) while consistently highlighting clinically relevant factors. Uplift-based ML policies outperformed treat-all and treat-none strategies, and GPT-4 improved interpretability through rationale-driven explanations. Discussion By combining predictive ML, causal survival modeling, and LLM-based rationale generation, the proposed framework provides a promising approach for personalized and transparent chemotherapy decision support in oncology.

U. Aickelin, Tekoshin Ammo, M. Nabi et al. · Frontiers in Digital Health · 0 citations