arXiv subjects

Weixiang Shen

Publications and source records attributed to Weixiang Shen.

12 recordsLinked to original sources

arXiv2026

DDX-TRACE: A Benchmark for Medical Diagnostic Trajectories in VLMs

Medical diagnosis is not a single prediction from a fully specified vignette. It is a sequential workup: clinicians decide what evidence to obtain, revise a differential diagnosis, and stop when the diagnosis is sufficiently supported. Most medical AI benchmarks instead reveal the relevant context upfront and score only the final answer, making unsupported correct guesses, premature closure, inefficient workups, and poor uncertainty updating invisible. We introduce DDX-TRACE, a physician-adjudicated benchmark for multimodal neuroradiology that evaluates diagnostic trajectories under hidden evidence over 211 challenging cases. Each case begins with limited clinical history; models request imaging studies in free form, receive matched image bundles when available, update a probabilistic differential diagnosis after each turn, and stop with a localized final diagnosis. Evaluating state-of-the-art VLMs, we find that final diagnosis scores can substantially misrepresent workup quality: models may guess plausible diagnoses without essential evidence, request useful studies but misinterpret raw images, or acquire evidence inefficiently while updating uncertainty poorly. Controlled evidence variants isolate bottlenecks in planning, visual evidence extraction, and downstream differential reasoning. DDX-TRACE shifts medical AI evaluation from final answers to evidence-supported diagnostic trajectories.

cs.CV↗

arXiv2026

RealICU: Do LLM Agents Understand Long-Context ICU Data? A Benchmark Beyond Behavior Imitation

Intensive care units (ICU) generate long, dense and evolving streams of clinical information, where physicians must repeatedly reassess patient states under time pressure, underscoring a clear need for reliable AI decision support. Existing ICU benchmarks typically treat historical clinician actions as ground truth. However, these actions are made under incomplete information and limited temporal context of the underlying patient state, and may therefore be suboptimal, making it difficult to assess the true reasoning capabilities of AI systems. We introduce RealICU, a hindsight-annotated benchmark for evaluating large language models (LLMs) under realistic ICU conditions, where labels are created after senior physicians review the full patient trajectory. We formulate four physician-motivated tasks: assess Patient Status, Acute Problems, Recommended Actions, and Red Flag actions that risk unsafe outcomes. We partition each trajectory with 30-min windows and release two datasets: RealICU-Gold with 930-window annotations from 94 MIMIC-IV patients, and RealICU-Scale with 11,862 windows extended by Oracle, a physician-validated LLM hindsight labeler. Existing LLMs including memory-augmented ones performed poorly on RealICU, exposing two failure modes: a recall-safety tradeoff for clinical recommendations, and an anchoring bias to early interpretations of the patient. We further introduce ICU-Evo to study structured-memory agents that improves long-horizon reasoning but does not fully eliminate safety failures. Together, RealICU provides a clinically grounded testbed for measuring and improving AI sequential decision-support in high-stakes care. Project page: https://chengzhi-leo.github.io/RealICU-Bench/

cs.AI↗

arXiv2026

ProcessThinker: Enhancing Multi-modal Large Language Models Reasoning via Rollout-based Process Reward

Visual question answering increasingly requires multi-step reasoning. Recent post-training with reinforcement learning under verifiable rewards (RLVR) and Group Relative Policy Optimization (GRPO) can improve multimodal reasoning, but most approaches rely on sparse outcome-only rewards. As a result, they struggle to tell whether an incorrect answer comes from a small mistake late in the reasoning or from an unhelpful trajectory from the start. A common solution is to train a process reward model (PRM) for step-level supervision, but this typically requires large-scale high-quality chain-of-thought annotations and additional training cost. We propose ProcessThinker, a practical post-training pipeline that provides step-level process rewards without training an explicit PRM. ProcessThinker first rewrites reasoning traces into a step-tagged format for cold-start supervised fine-tuning, then applies GRPO with a standard format reward and our rollout-based process reward. Concretely, for each intermediate step, we sample multiple continuations from that step and use the empirical success rate (final-answer verification) as the step reward. This gives dense credit assignment and encourages reasoning steps that more reliably support a correct conclusion, helping reduce inconsistent or self-contradictory progress across steps -- a key issue in logical reasoning. Across four challenging video benchmarks (Video-MMMU, MMVU, VideoMathQA, and LongVideoBench), ProcessThinker consistently improves over the baseline model Qwen3-VL-8B-Instruct

cs.CL↗

arXiv2026

Evo-MedAgent: Beyond One-Shot Diagnosis with Agents That Remember, Reflect, and Improve

Tool-augmented large language model (LLM) agents can orchestrate specialist classifiers, segmentation models, and visual question-answering modules to interpret chest X-rays. However, these agents still solve each case in isolation: they fail to accumulate experience across cases, correct recurrent reasoning mistakes, or adapt their tool-use behavior without expensive reinforcement learning. While a radiologist naturally improves with every case, current agents remain static. In this work, we propose Evo-MedAgent, a self-evolving memory module that equips a medical agent with the capacity for inter-case learning at test time. Our memory comprises three complementary stores: (1)~\emph{Retrospective Clinical Episodes} that retrieve problem-solving experiences from similar past cases, (2)~an \emph{Adaptive Procedural Heuristics} bank curating priority-tagged diagnostic rules that evolves via reflection, much like a physician refining their internal criteria, and (3)~a \emph{Tool Reliability Controller} that tracks per-tool trustworthiness. On ChestAgentBench, Evo-MedAgent raises multiple-choice question (MCQ) accuracy from 0.68 to 0.79 on GPT-5-mini, and from 0.76 to 0.87 on Gemini-3 Flash. With a strong base model, evolving memory improves performance more effectively than orchestrating external tools on qualitative diagnostic tasks. Because Evo-MedAgent requires no training, its per-case overhead is bounded by one additional retrieval pass and a single reflection call, making it deployable on top of any frozen model.

cs.AI↗

arXiv2026

MedOpenClaw and MedFlowBench: Auditing Medical Agents in Full-Study Workflows

Medical imaging benchmarks often evaluate VLMs on pre-selected 2D images, slices, crops, or patches, making evaluation closer to visual recognition. Real clinical workflows impose a different burden: readers must search through complete studies, operate imaging software, navigate across slices and magnifications, and document visual evidence that can be audited. We argue that this evidence-producing workflow is a critical missing evaluation axis for medical imaging agents. To study it, we introduce MedFlowBench, a full-study benchmark for VLM agents, together with MedOpenClaw, a controlled and replayable runtime in which agents operate medical imaging viewers such as 3D Slicer and QuPath. In each episode, an agent inspects a complete radiology study or whole-slide pathology image, returns a task answer, and submits structured evidence, including key slices, coordinates, regions of interest, or lesion-state fields. This evidence is automatically checked against withheld masks, annotations, and labels. Across evaluated models, final answer-only scoring gives an overly optimistic picture: when answers must also be supported by correct evidence, performance drops substantially on complex workflows. We further find that adding image-analysis tools does not by itself solve the problem. Tools help when they make a complex procedure simple and reliable, but agents still struggle when they must choose inputs, manage viewer state, and verify intermediate outputs over multiple steps. MedFlowBench exposes whether medical imaging agents can produce auditable evidence from complete studies, rather than plausible answers from selected images.

cs.CV↗

arXiv2026

SWE Context Bench: A Benchmark for Context Learning in Coding

Large language models are increasingly used as coding agents for software engineering tasks. Current benchmarks mainly evaluate whether the agent can correctly solve the request or fix the bugs. They largely treat tasks as independent and do not assess whether agents can reuse previous experience across related problems. As a result, the efficiency gains from reusing the previous experience remains difficult to measure. We introduce SWE-ContextBench, a benchmark designed to explicitly evaluate context understanding and retrieval in coding agents. SWE-ContextBench consists of 1,100 base tasks with another 376 related tasks derived from real dependency and reference relationships among GitHub issues and pull requests. SWE-ContextBench groups base tasks and related tasks with shared context across 51 unique repositories and 9 programming languages. The benchmark evaluates how accurately and efficiently agents solve related issues when prior cases are available in context. Using SWE-ContextBench, we study the behavior of multiple coding agents across varying context reuse settings and retrieval strategies. Our results show that accurately summarized and retrieved previous experience can significantly improve resolution accuracy and reduce runtime and token cost, particularly on harder tasks. In contrast, unfiltered or incorrectly selected context provides limited or negative benefits. These findings highlight the importance of context management and retrieval accuracy, and position SWE-ContextBench as a principled benchmark for studying context learning in coding agents.

cs.SE↗

arXiv2025

Experimental Methods, Health Indicators, and Diagnostic Strategies for Retired Lithium-ion Batteries: A Comprehensive Review

Reliable health assessment of retired lithium-ion batteries is essential for safe and economically viable second-life deployment, yet remains difficult due to sparse measurements, incomplete historical records, heterogeneous chemistries, and limited or noisy battery health labels. Conventional laboratory diagnostics, such as full charge-discharge cycling, pulse tests, Electrochemical Impedance Spectroscopy (EIS) measurements, and thermal characterization, provide accurate degradation information but are too time-consuming, equipment-intensive, or condition-sensitive to be applied at scale during retirement-stage sorting, leaving real-world datasets fragmented and inconsistent. This review synthesizes recent advances that address these constraints through physical health indicators, experiment testing methods, data-generation and augmentation techniques, and a spectrum of learning-based modeling routes spanning supervised, semi-supervised, weakly supervised, and unsupervised paradigms. We highlight how minimal-test features, synthetic data, domain-invariant representations, and uncertainty-aware prediction enable robust inference under limited or approximate labels and across mixed chemistries and operating histories. A comparative evaluation further reveals trade-offs in accuracy, interpretability, scalability, and computational burden. Looking forward, progress toward physically constrained generative models, cross-chemistry generalization, calibrated uncertainty estimation, and standardized benchmarks will be crucial for building reliable, scalable, and deployment-ready health prediction tools tailored to the realities of retired-battery applications.

eess.SP↗

arXiv2025

How Far Have Medical Vision-Language Models Come? A Comprehensive Benchmarking Study

Vision-Language Models (VLMs) trained on web-scale corpora excel at natural image tasks and are increasingly repurposed for healthcare; however, their competence in medical tasks remains underexplored. We present a comprehensive evaluation of open-source general-purpose and medically specialised VLMs, ranging from 3B to 72B parameters, across eight benchmarks: MedXpert, OmniMedVQA, PMC-VQA, PathVQA, MMMU, SLAKE, and VQA-RAD. To observe model performance across different aspects, we first separate it into understanding and reasoning components. Three salient findings emerge. First, large general-purpose models already match or surpass medical-specific counterparts on several benchmarks, demonstrating strong zero-shot transfer from natural to medical images. Second, reasoning performance is consistently lower than understanding, highlighting a critical barrier to safe decision support. Third, performance varies widely across benchmarks, reflecting differences in task design, annotation quality, and knowledge demands. No model yet reaches the reliability threshold for clinical deployment, underscoring the need for stronger multimodal alignment and more rigorous, fine-grained evaluation protocols.

cs.CV↗

arXiv2024

Active Cell Balancing for Extended Operational Time of Lithium-Ion Battery Systems in Energy Storage Applications

Cell inconsistency within a lithium-ion battery system poses a significant challenge in maximizing the system operational time. This study presents an optimization-driven active balancing method to minimize the effects of cell inconsistency on the system operational time while simultaneously satisfying the system output power demand and prolonging the system operational time in energy storage applications. The proposed method utilizes a fractional order model to forecast the terminal voltage dynamics of each cell within a battery system, enhanced with a particle-swarm-optimisation-genetic algorithm for precise parameter identification. It is implemented under two distinct cell-level balancing topologies: independent cell balancing and differential cell balancing. Subsequently, the current distribution for each topology is determined by resolving two optimization control problems constrained by the battery's operational specifications and power demands. The effectiveness of the proposed method is validated by extensive experiments based on the two balancing topologies. The results demonstrate that the proposed method increases the operational time by 3.2%.

eess.SY↗

arXiv2024

Towards Accurate and Efficient Sorting of Retired Lithium-ion Batteries: A Data Driven Based Electrode Aging Assessment Approach

Retired batteries (RBs) for second-life applications offer promising economic and environmental benefits. However, accurate and efficient sorting of RBs with discrepant characteristics persists as a pressing challenge. In this study, we introduce a data driven based electrode aging assessment approach to address this concern. To this end, a number of 15 feature points are extracted from battery open circuit voltage (OCV) curves to capture their characteristics at different levels of aging, and a convolutional neural network with an optimized structure and minimized input size is established to relocate the relative positions of these OCV feature points. Next, a rapid estimation algorithm is proposed to identify the three electrode aging parameters (EAPs) which best reconstruct the 15 OCV feature points over the entire usable capacity range. Utilizing the three EAPs as sorting indices, we employ an adaptive affinity propagation algorithm to cluster RBs without the need for pre-determining the clustering number. Unlike conventional sorting methods based solely on battery capacity, the proposed method provides profound insights into electrode aging behaviors, minimizes the need for constant-current charging data, and supports module/pack-level tests for the simultaneous processing of high volumes of RBs.

eess.SY↗

arXiv2024

Recent Advancements in Battery State of Power Estimation Technology: A Comprehensive Overview and Error Source Analysis

Accurate state of power (SOP) estimation is of great importance for lithium-ion batteries in safety-critical and power-intensive applications for electric vehicles. This review article delves deeply into the entire development flow of current SOP estimation technology, offering a systematic breakdown of all key aspects with their recent advancements. First, we review the design of battery safe operation area, summarizing diverse limitation factors and furnishing a profound comprehension of battery safety across a broad operational scale. Second, we illustrate the unique discharge and charge characteristics of various peak operation modes, such as constant current, constant voltage, constant current-constant voltage, and constant power, and explore their impacts on battery peak power performance. Third, we extensively survey the aspects of battery modelling and algorithm development in current SOP estimation technology, highlighting their technical contributions and specific considerations. Fourth, we present an in-depth dissection of all error sources to unveil their propagation pathways, providing insightful analysis into how each type of error impacts the SOP estimation performance. Finally, the technical challenges and complexities inherent in this field of research are addressed, suggesting potential directions for future development. Our goal is to inspire further efforts towards developing more accurate and intelligent SOP estimation technology for next-generation battery management systems.

eess.SY↗

arXiv2024

Recent Advances in Model-Based Fault Diagnosis for Lithium-Ion Batteries: A Comprehensive Review

Lithium-ion batteries (LIBs) have found wide applications in a variety of fields such as electrified transportation, stationary storage and portable electronics devices. A battery management system (BMS) is critical to ensure the reliability, efficiency and longevity of LIBs. Recent research has witnessed the emergence of model-based fault diagnosis methods in advanced BMSs. This paper provides a comprehensive review on the model-based fault diagnosis methods for LIBs. First, the widely explored battery models in the existing literature are classified into physics-based electrochemical models and electrical equivalent circuit models. Second, a general state-space representation that describes electrical dynamics of a faulty battery is presented. The formulation of the state vectors and the identification of the parameter matrices are then elaborated. Third, the fault mechanisms of both battery faults (incl. overcharege/overdischarge faults, connection faults, short circuit faults) and sensor faults (incl. voltage sensor faults and current sensor faults) are discussed. Furthermore, different types of modeling uncertainties, such as modeling errors and measurement noises, aging effects, measurement outliers, are elaborated. An emphasis is then placed on the observer design (incl. online state observers and offline state observers). The algorithm implementation of typical state observers for battery fault diagnosis is also put forward. Finally, discussion and outlook are offered to envision some possible future research directions.

eess.SY↗