Search for: All records

The ground state of the three-proton emitter 17Na has been identified. Energy correlations between the protons and the 14O residue indicate that its decay is sequential initiated by the emission of a single proton to the ground state of 16Ne followed by simultaneous emission of two protons to the ground state of 14O, and the derived total decay energy is 4.45(3) MeV. The angular distribution of the proton from the first step indicates that this emission is d-wave and only consistent with a spin parity of 5=2þ. This assignment is consistent with state-of-the-art Gamow shell-model calculations. As the ground state of 17C is 3=2þ, the assigned spin parity of its mirror 17Na indicates that the pair presents a new case of ground-state mirror symmetry violation (GMSV). Unlike other GMSV cases, which arise solely from either a Coulomb induced shift or deformation effects, the 17Na=17C case is generated by differences in the many-body correlations. 

Artificial Intelligence (AI) has demonstrated significant potential in healthcare, particularly in disease diagnosis and treatment planning. Recent progress in Medical Large Vision-Language Models (Med-LVLMs) has opened up new possibilities for interactive diagnostic tools. However, these models often suffer from factual hallucination, which can lead to incorrect diagnoses. Fine-tuning and retrieval-augmented generation (RAG) have emerged as methods to address these issues. However, the amount of high-quality data and distribution shifts between training data and deployment data limit the application of fine-tuning methods. Although RAG is lightweight and effective, existing RAG-based approaches are not sufficiently general to different medical domains and can potentially cause misalignment issues, both between modalities and between the model and the ground truth. In this paper, we propose a versatile multimodal RAG system, MMed-RAG, designed to enhance the factuality of Med-LVLMs. Our approach introduces a domain-aware retrieval mechanism, an adaptive retrieved contexts selection, and a provable RAG-based preference fine-tuning strategy. These innovations make the RAG process sufficiently general and reliable, significantly improving alignment when introducing retrieved contexts. Experimental results across five medical datasets (involving radiology, ophthalmology, pathology) on medical VQA and report generation demonstrate that MMed-RAG can achieve an average improvement of 43.8% in factual accuracy in the factual accuracy of Med-LVLMs. 

The recent emergence of Medical Large Vision Language Models (Med-LVLMs) has enhanced medical diagnosis. However, current Med-LVLMs frequently encounter factual issues, often generating responses that do not align with established medical facts. Retrieval-Augmented Generation (RAG), which utilizes external knowledge, can improve the factual accuracy of these models but introduces two major challenges. First, limited retrieved contexts might not cover all necessary information, while excessive retrieval can introduce irrelevant and inaccurate references, interfering with the model’s generation. Second, in cases where the model originally responds correctly, applying RAG can lead to an over-reliance on retrieved contexts, resulting in incorrect answers. To address these issues, we propose RULE, which consists of two components. First, we introduce a provably effective strategy for controlling factuality risk through the calibrated selection of the number of retrieved contexts. Second, based on samples where over-reliance on retrieved contexts led to errors, we curate a preference dataset to fine-tune the model, balancing its dependence on inherent knowledge and retrieved contexts for generation. We demonstrate the effectiveness of RAFE on three medical VQA datasets, achieving an average improvement of 20.8% in factual accuracy. 

Abstract Land cover data are commonly used to model the terrestrial carbon (C) sink, yet these data have wide margins of error that significantly alter estimates of global C storage. Here we demonstrate this data vulnerability in grasslands, which are critical to C cycling but whose estimated distribution has varied by >50 million km²(3.5–42% of the Earth’s terrestrial surface). Comparing multiple high-resolution land cover products with expertly annotated grassland data from six continents, we show sources of mapping error and discuss C implications based on 2023 United Nations (UN) FAO estimates. Past misidentification arose from inconsistent definitions on grassland identity and classification flaws especially relating to woody plant cover. Correcting these errors adjusted grassland coverage to 22.8% of the terrestrial land base (30.1 million km²), elevating UN projections of soil C stocks to 155.02 Pg (0–30 cm depth). These findings underscore the challenges of biome mapping for ecosystem accounting and policy, when lacking field-validated remotely sensed data.