Seyed Amir Ahmad Safavi-Naini

Seyed Amir Ahmad Safavi-Naini, Shuhaib Ali, Omer Shahab et al.

Background and Aims: This study evaluates the medical reasoning performance of large language models (LLMs) and vision language models (VLMs) in gastroenterology. Methods: We used 300 gastroenterology board exam-style multiple-choice questions, 138 of which contain images to systematically assess the impact of model configurations and parameters and prompt engineering strategies utilizing GPT-3.5. Next, we assessed the performance of proprietary and open-source LLMs (versions), including GPT (3.5, 4, 4o, 4omini), Claude (3, 3.5), Gemini (1.0), Mistral, Llama (2, 3, 3.1), Mixtral, and Phi (3), across different interfaces (web and API), computing environments (cloud and local), and model precisions (with and without quantization). Finally, we assessed accuracy using a semiautomated pipeline. Results: Among the proprietary models, GPT-4o (73.7%) and Claude3.5-Sonnet (74.0%) achieved the highest accuracy, outperforming the top open-source models: Llama3.1-405b (64%), Llama3.1-70b (58.3%), and Mixtral-8x7b (54.3%). Among the quantized open-source models, the 6-bit quantized Phi3-14b (48.7%) performed best. The scores of the quantized models were comparable to those of the full-precision models Llama2-7b, Llama2--13b, and Gemma2-9b. Notably, VLM performance on image-containing questions did not improve when the images were provided and worsened when LLM-generated captions were provided. In contrast, a 10% increase in accuracy was observed when images were accompanied by human-crafted image descriptions. Conclusion: In conclusion, while LLMs exhibit robust zero-shot performance in medical reasoning, the integration of visual data remains a challenge for VLMs. Effective deployment involves carefully determining optimal model configurations, encouraging users to consider either the high performance of proprietary models or the flexible adaptability of open-source models.

Mohammadreza Ghaffarzadeh-Esfahani, Mahdi Ghaffarzadeh-Esfahani, Arian Salahi-Niri et al.

This study compared the performance of classical feature-based machine learning models (CMLs) and large language models (LLMs) in predicting COVID-19 mortality using high-dimensional tabular data from 9,134 patients across four hospitals. Seven CML models, including XGBoost and random forest (RF), were evaluated alongside eight LLMs, such as GPT-4 and Mistral-7b, which performed zero-shot classification on text-converted structured data. Additionally, Mistral- 7b was fine-tuned using the QLoRA approach. XGBoost and RF demonstrated superior performance among CMLs, achieving F1 scores of 0.87 and 0.83 for internal and external validation, respectively. GPT-4 led the LLM category with an F1 score of 0.43, while fine-tuning Mistral-7b significantly improved its recall from 1% to 79%, yielding a stable F1 score of 0.74 during external validation. Although LLMs showed moderate performance in zero-shot classification, fine-tuning substantially enhanced their effectiveness, potentially bridging the gap with CML models. However, CMLs still outperformed LLMs in handling high-dimensional tabular data tasks. This study highlights the potential of both CMLs and fine-tuned LLMs in medical predictive modeling, while emphasizing the current superiority of CMLs for structured data analysis.

Mehran Advand, Zahra Dehghanian, Navid Faraji et al.

Existing medical imaging datasets for abdominal CT often lack three-dimensional annotations, multi-organ coverage, or precise lesion-to-organ associations, hindering robust representation learning and clinical applications. To address this gap, we introduce 3DLAND, a large-scale benchmark dataset comprising over 6,000 contrast-enhanced CT volumes with over 20,000 high-fidelity 3D lesion annotations linked to seven abdominal organs: liver, kidneys, pancreas, spleen, stomach, and gallbladder. Our streamlined three-phase pipeline integrates automated spatial reasoning, prompt-optimized 2D segmentation, and memory-guided 3D propagation, validated by expert radiologists with surface dice scores exceeding 0.75. By providing diverse lesion types and patient demographics, 3DLAND enables scalable evaluation of anomaly detection, localization, and cross-organ transfer learning for medical AI. Our dataset establishes a new benchmark for evaluating organ-aware 3D segmentation models, paving the way for advancements in healthcare-oriented AI. To facilitate reproducibility and further research, the 3DLAND dataset and implementation code are publicly available at https://mehrn79.github.io/3DLAND.

Zahra Atf, Seyed Amir Ahmad Safavi-Naini, Peter R. Lewis et al.

This study investigates uncertainty quantification in large language models (LLMs) for medical applications, emphasizing both technical innovations and philosophical implications. As LLMs become integral to clinical decision-making, accurately communicating uncertainty is crucial for ensuring reliable, safe, and ethical AI-assisted healthcare. Our research frames uncertainty not as a barrier but as an essential part of knowledge that invites a dynamic and reflective approach to AI design. By integrating advanced probabilistic methods such as Bayesian inference, deep ensembles, and Monte Carlo dropout with linguistic analysis that computes predictive and semantic entropy, we propose a comprehensive framework that manages both epistemic and aleatoric uncertainties. The framework incorporates surrogate modeling to address limitations of proprietary APIs, multi-source data integration for better context, and dynamic calibration via continual and meta-learning. Explainability is embedded through uncertainty maps and confidence metrics to support user trust and clinical interpretability. Our approach supports transparent and ethical decision-making aligned with Responsible and Reflective AI principles. Philosophically, we advocate accepting controlled ambiguity instead of striving for absolute predictability, recognizing the inherent provisionality of medical knowledge.

Nariman Naderi, Seyed Amir Ahmad Safavi-Naini, Thomas Savage et al.

This study evaluated self-reported response certainty across several large language models (GPT, Claude, Llama, Phi, Mistral, Gemini, Gemma, and Qwen) using 300 gastroenterology board-style questions. The highest-performing models (GPT-o1 preview, GPT-4o, and Claude-3.5-Sonnet) achieved Brier scores of 0.15-0.2 and AUROC of 0.6. Although newer models demonstrated improved performance, all exhibited a consistent tendency towards overconfidence. Uncertainty estimation presents a significant challenge to the safe use of LLMs in healthcare. Keywords: Large Language Models; Confidence Elicitation; Artificial Intelligence; Gastroenterology; Uncertainty Quantification

Seyed Amir Ahmad Safavi-Naini, Elahe Meftah, Josh Mohess et al.

The competency of any intelligent agent is bounded by its formal account of the world in which it operates. Clinical AI lacks such an account. Existing frameworks address evaluation, regulation, or system design in isolation, without a shared model of the clinical world to connect them. We introduce the Clinical World Model, a framework that formalizes care as a tripartite interaction among Patient, Provider, and Ecosystem. To formalize how any agent, whether human or artificial, transforms information into clinical action, we develop parallel decision-making architectures for providers, patients, and AI agents, grounded in validated principles of clinical cognition. The Clinical AI Skill-Mix operationalizes competency through eight dimensions. Five define the clinical competency space (condition, phase, care setting, provider role, and task) and three specify how AI engages human reasoning (assigned authority, agent facing, and anchoring layer). The combinatorial product of these dimensions yields a space of billions of distinct competency coordinates. A central structural implication is that validation within one coordinate provides minimal evidence for performance in another, rendering the competency space irreducible. The framework supplies a common grammar through which clinical AI can be specified, evaluated, and bounded across stakeholders. By making this structure explicit, the Clinical World Model reframes the field's central question from whether AI works to in which competency coordinates reliability has been demonstrated, and for whom.

Nariman Naderi, Zahra Atf, Peter R Lewis et al.

This paper investigates how prompt engineering techniques impact both accuracy and confidence elicitation in Large Language Models (LLMs) applied to medical contexts. Using a stratified dataset of Persian board exam questions across multiple specialties, we evaluated five LLMs - GPT-4o, o3-mini, Llama-3.3-70b, Llama-3.1-8b, and DeepSeek-v3 - across 156 configurations. These configurations varied in temperature settings (0.3, 0.7, 1.0), prompt styles (Chain-of-Thought, Few-Shot, Emotional, Expert Mimicry), and confidence scales (1-10, 1-100). We used AUC-ROC, Brier Score, and Expected Calibration Error (ECE) to evaluate alignment between confidence and actual performance. Chain-of-Thought prompts improved accuracy but also led to overconfidence, highlighting the need for calibration. Emotional prompting further inflated confidence, risking poor decisions. Smaller models like Llama-3.1-8b underperformed across all metrics, while proprietary models showed higher accuracy but still lacked calibrated confidence. These results suggest prompt engineering must address both accuracy and uncertainty to be effective in high-stakes medical tasks.

Saeide Danaei, Zahra Dehghanian, Elahe Meftah et al.

This systematic review critically evaluates publicly available abdominal CT datasets and their suitability for artificial intelligence (AI) applications in clinical settings. We examined 46 publicly available abdominal CT datasets (50,256 studies). Across all 46 datasets, we found substantial redundancy (59.1\% case reuse) and a Western/geographic skew (75.3\% from North America and Europe). A bias assessment was performed on the 19 datasets with >=100 cases; within this subset, the most prevalent high-risk categories were domain shift (63\%) and selection bias (57\%), both of which may undermine model generalizability across diverse healthcare environments -- particularly in resource-limited settings. To address these challenges, we propose targeted strategies for dataset improvement, including multi-institutional collaboration, adoption of standardized protocols, and deliberate inclusion of diverse patient populations and imaging technologies. These efforts are crucial in supporting the development of more equitable and clinically robust AI models for abdominal imaging.

Mohammad Amin Khalafi, Seyed Amir Ahmad Safavi-Naini, Ameneh Salehi et al.

Introduction: This study provides a comprehensive performance assessment of vision-language models (VLMs) against established convolutional neural networks (CNNs) and classic machine learning models (CMLs) for computer-aided detection (CADe) and computer-aided diagnosis (CADx) of colonoscopy polyp images. Method: We analyzed 2,258 colonoscopy images with corresponding pathology reports from 428 patients. We preprocessed all images using standardized techniques (resizing, normalization, and augmentation) and implemented a rigorous comparative framework evaluating 11 distinct models: ResNet50, 4 CMLs (random forest, support vector machine, logistic regression, decision tree), two specialized contrastive vision language encoders (CLIP, BiomedCLIP), and three general-purpose VLMs ( GPT-4 Gemini-1.5-Pro, Claude-3-Opus). Our performance assessment focused on two clinical tasks: polyp detection (CADe) and classification (CADx). Result: In polyp detection, ResNet50 achieved the best performance (F1: 91.35%, AUROC: 0.98), followed by BiomedCLIP (F1: 88.68%, AUROC: [AS1] ). GPT-4 demonstrated comparable effectiveness to traditional machine learning approaches (F1: 81.02%, AUROC: [AS2] ), outperforming other general-purpose VLMs. For polyp classification, performance rankings remained consistent but with lower overall metrics. ResNet50 maintained the highest efficacy (weighted F1: 74.94%), while GPT-4 demonstrated moderate capability (weighted F1: 41.18%), significantly exceeding other VLMs (Claude-3-Opus weighted F1: 25.54%, Gemini 1.5 Pro weighted F1: 6.17%). Conclusion: CNNs remain superior for both CADx and CADe tasks. However, VLMs like BioMedCLIP and GPT-4 may be useful for polyp detection tasks where training CNNs is not feasible.