Jul 2026· Computer Methods and Programs in Biomedicine· Vol 285, pp.
109541
· 0 citations· 129 references
Medicine
Abstract
Medical Question Answering (MQA) has emerged as a critical artificial intelligence (AI) capability for supporting clinicians, researchers, and the general public with timely and evidence-based responses to medical queries. Recent advances in natural language processing (NLP), computer vision, and large language models (LLMs) have expanded MQA from text-only systems to multimodal frameworks. This survey aims to provide a comprehensive and structured review of MQA systems, covering both text and image-based approaches. We present a systematic review of MQA literature, including applications, datasets, and modeling paradigms. We introduce a unified taxonomy categorizing MQA systems into scientific, clinical, consumer, and examination-oriented tasks. We also analyze representative datasets for text-based and vision-based question answering, focusing on data sources, annotation strategies, task formulations, and evaluation protocols. Furthermore, we review methodological developments ranging from classical and transformer-based models to multimodal vision-language systems and LLM-driven approaches. The analysis highlights a rapid evolution of MQA systems toward multimodal and LLM-based frameworks, particularly in medical visual question answering. Existing datasets and models demonstrate strong progress but also reveal limitations in generalization, reasoning, and real-world clinical applicability. Key challenges remain, including reliability, hallucination, explainability, fairness, and clinical safety. This survey identifies open research directions such as improved data quality, knowledge-grounded reasoning, trustworthy evaluation, and real-world deployment. The study provides a comprehensive reference and roadmap for developing reliable and clinically applicable MQA systems.
Medical intelligent question-answering (QA) systems have become important tools for improving the efficiency of healthcare services, and recent research has increasingly emphasized performance optimization and multimodal integration. However, existing systems still face several challenges in intent recognition, entity extraction, and multimodal knowledge fusion, particularly reduced accuracy in multi-label classification, heavy reliance on large-scale annotated data, and limited support for cross-modal retrieval. To address these issues, this study proposes a medical intelligent QA framework that integrates a dual-layer attention mechanism, a large language model, and a multimodal medical knowledge graph to improve system understanding and response generation in complex clinical scenarios. Specifically, we develop a text-based intent recognition model with a dual-layer attention architecture, in which a global contextual attention module is introduced to capture long-range semantic dependencies and improve multi-label classification performance. In addition, an instruction-tuned large language model is employed for zero-shot medical entity recognition, thereby reducing dependence on manually annotated datasets. Building on this foundation, we construct a multimodal medical knowledge graph comprising more than 15,000 associated medical images and develop a visualization-oriented retrieval interface using Flask and ECharts. Experimental results show that the proposed intent recognition model achieves a peak Micro-F1 of 94.42% on multiple benchmark datasets, outperforming several baseline methods. The LLM-based entity recognition module achieved competitive recall in medical entity extraction, demonstrating strong capability in identifying medical entities. User evaluation results further indicate that the system is effective and practical across a variety of medical query types. This study provides a feasible framework for advancing medical QA systems through improved intent recognition, low-resource entity extraction, and multimodal knowledge integration.
Guoqiang Qiu, Qingni Yuan, Yi Wang et al.· PLoS ONE· 0 citations
An engineering-oriented, end-to-end roadmap that structures the full lifecycle of clinical language model systems—from model design and domain adaptation to optimization and real-world evaluation is introduced.
A dual-view approach that connects clinical practice with computational methods is presented, establishing a five-level competency scheme following Miller’s Pyramid and linking deductive, inductive, and abductive reasoning patterns to common medical goals and tasks.
The integration of large language models (LLMs) into healthcare is hindered by their tendency to generate hallucinations, a critical issue in medical question answering (MedQA). Retrieval-Augmented Generation (RAG) frameworks address this limitation by grounding LLM outputs in external, verifiable knowledge sources. In this study, RAG pipeline implemented without fine-tuning the base model is evaluated to enhance factual fidelity in MedQA. GPT-4 is integrated with a dense passage retrieval module based on Facebook AI Similarity Search (FAISS) and the BAAI/bge-small-en embedding model, and compared against a non-retrieval GPT-4 baseline.Evaluation is conducted on a held-out set of 1,000 questions from the MedQuAD dataset using widely adopted computational evaluation metrics, including BLEU-1 to BLEU-4, ROUGE-L, METEOR, text-level F1, Exact Match (EM), and embedding-based semantic similarity metrics such as BERTScore and SBERT. The RAG-enhanced configuration consistently outperforms the baseline across all evaluation metrics, with notable improvements including an increase in BLEU-4 from 0.3015 to 0.4224 and BERTScore from 0.8487 to 0.8691, indicating improved contextual coherence and semantic alignment. Qualitative analysis further indicates fewer omissions and more accurate use of domain-specific terminology. To assess generalization, cross-dataset experiments were conducted using the PubMedQA benchmark. The results reveal that while RAG significantly improves performance in in-domain settings, its effectiveness decreases under domain shift when the retrieval
corpus
is not well aligned with the target question distribution. Additional experiments with varying retrieval depths (k = 1, 3, and 5) demonstrate that increasing the number of retrieved passages can improve semantic similarity by providing richer contextual evidence. These findings provide empirical evidence that retrieval augmentation enhances both the factual reliability and semantic quality of LLM-generated answers in MedQA. The proposed framework offers a practical and scalable approach to mitigating hallucinations without requiring task-specific fine-tuning, highlighting the potential of retrieval-augmented approaches for trustworthy artificial intelligence (AI)-assisted healthcare applications.
Background This study aimed to generate medical qualification exam questions and their corresponding answers from real-world electronic health records (EHRs) with large language models (LLMs), and to compare their output to that of human medical experts. Methods Utilizing a multicenter bidirectional anonymized database China Elderly Comorbidity Medical Database (CECMed), a total of 8 LLMs: ERNIE 4, ChatGLM 4, Doubao, Hunyuan, Spark 4, Qwen, Llama 3, and Mistral were tasked with generating open-ended questions and answers based on a subset of sampled admission reports. LLMs generated the medical question and answer through few-shot prompting. An independent expert panel scored the AI-generated outputs based on multiple criteria, including coherence, sufficiency of key information, information correctness, factual consistency, evidence of statement, and professionalism, using 5-point Likert scales. Results For question generation, ERNIE 4 achieved the highest cumulative score (16.47). Human experts surpassed LLMs in sufficiency of key information (3.67) but lagged in information correctness (3.63 vs. LLMs' 4.03–4.57). The information correctness of ERNIE was significantly higher than the human's [0.93 (0.62, 1.24), p < 0.01]. For answer generation, humans led overall (14.49), while Doubao outperformed the other LLMs in coherence (3.57), factual consistency (3.60), and professionalism (3.53). The coherence of human's was significantly better than that of 8 LLMs, especially outperformed Llama [0.8 (0.37, 1.23), p < 0.01] and Mistral [0.87 (0.45, 1.28), p < 0.01]. Conclusions Conventional medical education requires clinicians to formulate questions and answers based on prototypes from EHRs, which is heuristic and time-consuming. This study shows that mainstream LLMs could generate questions and answers with real-world EHRs at levels close to clinicians. Although current LLMs performed dissatisfactorily in some aspects, medical students and interns may find LLMs a useful auxiliary tool to support their learning. Clinical Trial Registration https://clinicaltrials.gov/study/NCT06316544, identifier: NCT06316544.
Yunqi Zhu, Wen Tang, Huayu Yang et al.· Frontiers in Digital Health· 0 citations
This entry-level tutorial aims to equip healthcare professionals with the tools necessary to effectively integrate LLMs into clinical practice, ensuring that these powerful technologies are applied in a safe, reliable, and impactful manner.
Qiao Jin, Nicholas Wan, Robert Leaman et al.· Nature Protocols· 1 citation