Medical question-answering (QA) is a critical task for evaluating how effectively large language models (LLMs) encode clinical knowledge and assessing their potential applications in medicine. Despite showing promise on multiple-choice tests, LLMs frequently struggle with open-ended medical questions, producing responses with dangerous hallucinations or lacking comprehensive coverage of critical aspects. Existing approaches attempt to address these challenges through domain-specific fine-tuning, but this proves resource-intensive and difficult to scale across models. To improve the comprehensiveness and factuality of medical responses, we propose a novel approach utilizing structured medical reasoning. Our method guides LLMs through an seven-step cognitive process inspired by clinical diagnosis, enabling more accurate and complete answers without additional training. Experiments on the MedLFQA benchmark demonstrate that our approach achieves the highest Factuality Score of 85.8, surpassing fine-tuned models. Notably, this improvement transfers to smaller models, highlighting the method's efficiency and scalability. Our code and datasets are available.

Large language models (LLMs) have shown promise for generative and knowledge-intensive tasks including question-answering (QA) tasks. However, the practical deployment still faces challenges, notably the issue of "hallucination", where models generate plausible-sounding but unfaithful or nonsensical information. This issue becomes particularly critical in the medical domain due to the uncommon professional concepts and potential social risks involved. This paper analyses the phenomenon of hallucination in medical generative QA systems using widely adopted LLMs and datasets. Our investigation centers on the identification and comprehension of common problematic answers, with a specific emphasis on hallucination. To tackle this challenge, we present an interactive self-reflection methodology that incorporates knowledge acquisition and answer generation. Through this feedback process, our approach steadily enhances the factuality, consistency, and entailment of the generated answers. Consequently, we harness the interactivity and multitasking ability of LLMs and produce progressively more precise and accurate answers. Experimental results on both automatic and human evaluation demonstrate the superiority of our approach in hallucination reduction compared to baselines.

And now, an algorithm can predict not only whether they carry a genetic mutation, but which genes were mutated. The study, published Monday in Nature Medicine, is the latest from a Boston-based company called FDNA, one of a few organizations creating software that can help physicians diagnose genetic syndromes based just on a face -- and may serve an important validation of the company's technology, said Yaron Gurovich, the company's chief technology officer. "We went for this high-impact journal to prove beyond any doubt that this technology is good, it performs as we say, we can stand behind it, and now it opens a lot of doors to publish more," he said. The study itself is a collection of experiments testing how the results of algorithms -- FDNA refers to them as DeepGestalt -- stack up against clinicians' diagnoses. In one of the experiments, DeepGestalt's performance was better than random chance when picking which of five genetic mutations might be causing a condition called Noonan syndrome.

A facial recognition scan could become part of a standard medical checkup in the not-too-distant future. Researchers have shown how algorithms can help identify facial characteristics linked to genetic disorders, potentially speeding up clinical diagnoses. In a study published this month in the journal Nature Medicine, US company FDNA published new tests of their software, DeepGestalt. Just like regular facial recognition software, the company trained their algorithms by analyzing a dataset of faces. FDNA collected more than 17,000 images covering 200 different syndromes using a smartphone app it developed named Face2Gene.

The photograph is cropped close on the face of four-year-old Yael, who is smiling and looking as healthy as can be. But a computer analysis of her features says something's not right. She has MR XL Bain Type, the computer predicts--a very rare syndrome that causes a wide range of health problems. It turned out that the computer was right. Yael is one of thousands of children who have contributed to the development of an artificial intelligence system called DeepGestalt that can identify rare genetic disorders based on facial features alone.

Rare disorders often show up in someone's appearance. Individuals with Noonan syndrome--a genetic condition that inhibits the body's growth and development--can have wide-set eyes, for example, and those with Bain type intellectual disability--caused by a mutated gene on the X chromosome--sport almond-shaped eyes and small chins (see above). Now, researchers have trained artificial intelligence to recognize these features, paving the way for early--and cheap--diagnoses. Scientists built a computer program, DeepGestalt, and trained it on a publicly available data set of more than 17,000 photos of patients affected by more than 200 rare disorders. The program then used deep learning to recognize which patterns of markers were linked to hundreds of different genetic syndromes.