Yinghan Zhao

1 Introduction The escalating and increasingly globalized impacts of human activities on the natural environment have ushered in the Anthropocene (Waters et al. 2016). Recent global assessments document multiple, concurrent signals of planetary stress, including a ~1.1°C rise in global mean surface temperature (Wang et al. 2022), the threat of extinction facing approximately one million species (Diaz et al. 2019), and the degradation of up to 40% of global land (UNCCD 2025).

Abstract Chronic ocular graft-versus-host disease (coGVHD) after allogeneic hematopoietic stem cell transplantation (allo-HSCT) may lead to irreversible ocular surface damage and even vision loss. Current management of coGVHD faces challenges, with frequent missed or misdiagnosed cases. This study aimed to leverage a multimodal large language model (MLLM) to develop an early warning and diagnostic system for coGVHD.

Large Reasoning Models (LRMs) excel at complex reasoning tasks, but their efficiency is often hampered by overly verbose outputs. Prior steering methods attempt to address this issue by applying a single, global vector to hidden representations—an approach grounded in the restrictive linear representation hypothesis.