arXiv subjects

Danqing Hu

Publications and source records attributed to Danqing Hu.

17 recordsLinked to original sources

arXiv2025

TraceCoder: Towards Traceable ICD Coding via Multi-Source Knowledge Integration

Automated International Classification of Diseases (ICD) coding assigns standardized diagnosis and procedure codes to clinical records, playing a critical role in healthcare systems. However, existing methods face challenges such as semantic gaps between clinical text and ICD codes, poor performance on rare and long-tail codes, and limited interpretability. To address these issues, we propose TraceCoder, a novel framework integrating multi-source external knowledge to enhance traceability and explainability in ICD coding. TraceCoder dynamically incorporates diverse knowledge sources, including UMLS, Wikipedia, and large language models (LLMs), to enrich code representations, bridge semantic gaps, and handle rare and ambiguous codes. It also introduces a hybrid attention mechanism to model interactions among labels, clinical context, and knowledge, improving long-tail code recognition and making predictions interpretable by grounding them in external evidence. Experiments on MIMIC-III-ICD9, MIMIC-IV-ICD9, and MIMIC-IV-ICD10 datasets demonstrate that TraceCoder achieves state-of-the-art performance, with ablation studies validating the effectiveness of its components. TraceCoder offers a scalable and robust solution for automated ICD coding, aligning with clinical needs for accuracy, interpretability, and reliability.

cs.CL↗

arXiv2025

TACL: Threshold-Adaptive Curriculum Learning Strategy for Enhancing Medical Text Understanding

Medical texts, particularly electronic medical records (EMRs), are a cornerstone of modern healthcare, capturing critical information about patient care, diagnoses, and treatments. These texts hold immense potential for advancing clinical decision-making and healthcare analytics. However, their unstructured nature, domain-specific language, and variability across contexts make automated understanding an intricate challenge. Despite the advancements in natural language processing, existing methods often treat all data as equally challenging, ignoring the inherent differences in complexity across clinical records. This oversight limits the ability of models to effectively generalize and perform well on rare or complex cases. In this paper, we present TACL (Threshold-Adaptive Curriculum Learning), a novel framework designed to address these challenges by rethinking how models interact with medical texts during training. Inspired by the principle of progressive learning, TACL dynamically adjusts the training process based on the complexity of individual samples. By categorizing data into difficulty levels and prioritizing simpler cases early in training, the model builds a strong foundation before tackling more complex records. By applying TACL to multilingual medical data, including English and Chinese clinical records, we observe significant improvements across diverse clinical tasks, including automatic ICD coding, readmission prediction and TCM syndrome differentiation. TACL not only enhances the performance of automated systems but also demonstrates the potential to unify approaches across disparate medical domains, paving the way for more accurate, scalable, and globally applicable medical text understanding solutions.

cs.CL↗

arXiv2024

Predicting Lung Cancer Patient Prognosis with Large Language Models

Prognosis prediction is crucial for determining optimal treatment plans for lung cancer patients. Traditionally, such predictions relied on models developed from retrospective patient data. Recently, large language models (LLMs) have gained attention for their ability to process and generate text based on extensive learned knowledge. In this study, we evaluate the potential of GPT-4o mini and GPT-3.5 in predicting the prognosis of lung cancer patients. We collected two prognosis datasets, i.e., survival and post-operative complication datasets, and designed multiple tasks to assess the models' performance comprehensively. Logistic regression models were also developed as baselines for comparison. The experimental results demonstrate that LLMs can achieve competitive, and in some tasks superior, performance in lung cancer prognosis prediction compared to data-driven logistic regression models despite not using additional patient data. These findings suggest that LLMs can be effective tools for prognosis prediction in lung cancer, particularly when patient data is limited or unavailable.

cs.CL↗

arXiv2024

The Power of Combining Data and Knowledge: GPT-4o is an Effective Interpreter of Machine Learning Models in Predicting Lymph Node Metastasis of Lung Cancer

Lymph node metastasis (LNM) is a crucial factor in determining the initial treatment for patients with lung cancer, yet accurate preoperative diagnosis of LNM remains challenging. Recently, large language models (LLMs) have garnered significant attention due to their remarkable text generation capabilities. Leveraging the extensive medical knowledge learned from vast corpora, LLMs can estimate probabilities for clinical problems, though their performance has historically been inferior to data-driven machine learning models. In this paper, we propose a novel ensemble method that combines the medical knowledge acquired by LLMs with the latent patterns identified by machine learning models to enhance LNM prediction performance. Initially, we developed machine learning models using patient data. We then designed a prompt template to integrate the patient data with the predicted probability from the machine learning model. Subsequently, we instructed GPT-4o, the most advanced LLM developed by OpenAI, to estimate the likelihood of LNM based on patient data and then adjust the estimate using the machine learning output. Finally, we collected three outputs from the GPT-4o using the same prompt and ensembled these results as the final prediction. Using the proposed method, our models achieved an AUC value of 0.778 and an AP value of 0.426 for LNM prediction, significantly improving predictive performance compared to baseline machine learning models. The experimental results indicate that GPT-4o can effectively leverage its medical knowledge and the probabilities predicted by machine learning models to achieve more accurate LNM predictions. These findings demonstrate that LLMs can perform well in clinical risk prediction tasks, offering a new paradigm for integrating medical knowledge and patient data in clinical predictions.

cs.CL↗

arXiv2024

The current status of large language models in summarizing radiology report impressions

Large language models (LLMs) like ChatGPT show excellent capabilities in various natural language processing tasks, especially for text generation. The effectiveness of LLMs in summarizing radiology report impressions remains unclear. In this study, we explore the capability of eight LLMs on the radiology report impression summarization. Three types of radiology reports, i.e., CT, PET-CT, and Ultrasound reports, are collected from Peking University Cancer Hospital and Institute. We use the report findings to construct the zero-shot, one-shot, and three-shot prompts with complete example reports to generate the impressions. Besides the automatic quantitative evaluation metrics, we define five human evaluation metrics, i.e., completeness, correctness, conciseness, verisimilitude, and replaceability, to evaluate the semantics of the generated impressions. Two thoracic surgeons (ZSY and LB) and one radiologist (LQ) compare the generated impressions with the reference impressions and score each impression under the five human evaluation metrics. Experimental results show that there is a gap between the generated impressions and reference impressions. Although the LLMs achieve comparable performance in completeness and correctness, the conciseness and verisimilitude scores are not very high. Using few-shot prompts can improve the LLMs' performance in conciseness and verisimilitude, but the clinicians still think the LLMs can not replace the radiologists in summarizing the radiology impressions.

cs.CL↗

arXiv2023

Pseudogap Behavior in the Local Spinon Spectrum of Power-Law Diverging Multichannel Kondo Model

Motivated by the emergence of higher-order van Hove singularities (VHS) with power-law divergent density of states (DOS) ($ρ_c(ω)=ρ_0/|ω|^{r}$, $0<r<1$) in materials, we investigate a multichannel Kondo model involving conduction electrons near the higher-order van Hove filling. This model considers $M$ channel and $N$ spin degrees of freedom. Employing a renormalization group analysis and dynamical large-$N$ approach, our results reveal a crossover from a non-Fermi liquid to pseudogap behavior in the spectral properties of the local impurity at the overscreened fixed point. We precisely determine the conditions under which the crossover occurs, either by tuning the exponent $r$ or the ratio $κ=M/N$ to a critical value. This pseudogap phase of spinon exhibits distinct physical properties that could have an impact on the properties of real systems. The results of this study provide novel insights into the non-Fermi liquid and pseudogap behaviors observed in strongly correlated systems and offer a playground to study the interplay between higher-order van Hove singularities and multichannel Kondo physics.

cond-mat.str-el↗

arXiv2023

Zero-shot information extraction from radiological reports using ChatGPT

Electronic health records contain an enormous amount of valuable information, but many are recorded in free text. Information extraction is the strategy to transform the sequence of characters into structured data, which can be employed for secondary analysis. However, the traditional information extraction components, such as named entity recognition and relation extraction, require annotated data to optimize the model parameters, which has become one of the major bottlenecks in building information extraction systems. With the large language models achieving good performances on various downstream NLP tasks without parameter tuning, it becomes possible to use large language models for zero-shot information extraction. In this study, we aim to explore whether the most popular large language model, ChatGPT, can extract useful information from the radiological reports. We first design the prompt template for the interested information in the CT reports. Then, we generate the prompts by combining the prompt template with the CT reports as the inputs of ChatGPT to obtain the responses. A post-processing module is developed to transform the responses into structured extraction results. We conducted the experiments with 847 CT reports collected from Peking University Cancer Hospital. The experimental results indicate that ChatGPT can achieve competitive performances for some extraction tasks compared with the baseline information extraction system, but some limitations need to be further improved.

cs.CL↗

arXiv2023

Auxiliary-Bath Numerical Renormalization Group Method and Successive Collective Screening in Multi-Impurity Kondo Systems

We propose an auxiliary-bath algorithm for the numerical renormalization group (NRG) method to solve multi-impurity models with shared electron baths. The method allows us to disentangle the electron baths into independent Wilson chains to perform standard NRG procedures beyond the widely adopted independent bath approximation. Its application to the 2-impurity model immediately reproduces the well-known even- and odd-parity channels. For 3-impurity Kondo models, we find collective screening of cluster degrees of freedom depending on impurity configurations and clarify the false prediction of a non-Fermi liquid ground state for the $C_3$ symmetric case in previous literature due to improper treatment of disentanglement. Our work highlights the importance of nonlocal spatial correlations due to shared baths and reveals a generic picture of successive collective screening for entropy depletion that is crucial in real correlated systems. Our method greatly expands the applicability of the NRG and opens an avenue for its further development.

cond-mat.str-el↗

arXiv2022

Evolution of topological end states in the one-dimensional Kondo-Heisenberg model with site modulation

We investigate the interplay of the topological and Kondo effects in a one-dimensional Kondo-Heisenberg model with nontrivial conduction band using the density matrix renormalization group method. By analyzing the density profile, the local hybridization, and the spin/charge gap, we find that the Kondo effect can be destructed at the edges of the chain by the topological end state below a finite critical Kondo coupling $J_{K}^{c}$. We construct a phase diagram characterizing the transition of the end states.

cond-mat.str-el↗

arXiv2022

Continuous ferromagnetic quantum phase transition on an anisotropic Kondo lattice

Motivated by the recent discovery of ferromagnetic quantum criticality in the heavy fermion compound CeRh$_6$Ge$_4$, we develop a numerical algorithm of infinite projected entangled pair states for the anisotropic ferromagnetic Kondo-Heisenberg model in two dimensions and study the ferromagnetic quantum phase transitions with varying magnetic and hopping anisotropy. Our calculations reveal a continuous ferromagnetic quantum phase transition in the large anisotropic region and first-order quantum phase transitions for smaller anisotropy. Our results highlight the importance of magnetic anisotropy on ferromagnetic quantum criticality in Kondo lattice systems and provide a possible explanation for the experimental observation in CeRh$_6$Ge$_4$ with a quasi-one-dimensional magnetic structure. Our work opens the avenue for future studies of the rich Kondo lattice physics using state-of-the-art tensor network approaches.

cond-mat.str-el↗

arXiv2022

Learning disentangled representation for classical models

Finding disentangled representation plays a predominant role in the success of modern deep learning applications, but the results lack a straightforward explanation. Here we apply the information bottleneck method and its $β$-VAE implementation to find the disentangled low-dimensional representation of classical models. For the Ising model, our results reveal a deep connection between the disentangled features and the physical order parameters, and the widely-used Bernoulli decoder is found to be learning a mean-field Hamiltonian at fixed temperature. This analogy motivates us to extend the application of $β$-VAE to more complex classical models with non-binary variables using different decoder neural network and propose a modified architecture $β^2$-VAE to enforce thermal fluctuations in generated samples. Our work provides a way to design novel physics-informed algorithm that can yield learned features in potential correspondence with real physical properties.

cond-mat.str-el↗

arXiv2021

Schwinger boson approach for the dynamical mean-field theory of the Kondo lattice

We apply the dynamical large-$N$ Schwinger boson technique as an impurity solver for the dynamical mean-field theory (DMFT) calculations of the Kondo lattice model. Our approach captures the hybridization physics through the DMFT self-consistency that is missing in the pure Schwinger boson calculations with independent electron baths. The resulting thermodynamic and transport properties are in qualitative agreement with more rigorous calculations and give the correct crossover behavior over a wide temperature range from the local moment regime to the Fermi liquid. Our method may be further extended to combine with the density functional theory for efficient material calculations.

cond-mat.str-el↗

arXiv2020

Energy-scale cascade and correspondence between Mott and Kondo lattice physics

We propose an energy-scale correspondence between the Mott physics and the Kondo lattice physics and construct a tentative phase diagram of their correlated electrons with two characteristic energy scales $ω^*$ and $Ω$ marking the upper boundary of a low-energy regime with well-developed long-range coherence and the lower boundary of localized spectral weight in the energy space, respectively. In between, there exists a crossover region with emergent but damped quasiparticle excitations. We argue that the presence of two separate energy scales is a generic property of correlated electrons on a lattice and reflects an intrinsic two-stage process of the quasiparticle dynamics to build up the lattice coherence. For the Hubbard model, they correspond to the kink and waterfall structures on the dispersion, while for the periodic Anderson model, they are associated with the indirect and direct hybridization gaps. Our work reveals a deep connection between the Mott and Kondo lattice physics and provides a basic ingredient for the study of many-body systems.

cond-mat.str-el↗

arXiv2020

Effective classical correspondence of the Mott transition

We derive an effective classical model to describe the Mott transition of the half-filled one-band Hubbard model in the framework of the dynamical mean-field theory with hybridization expansion of the continuous time quantum Monte Carlo. We find a simple two-body interaction of exponential form and reveal a classical correspondence of the Mott transition driven by a logarithmically divergent interaction length. Our work provides an alternative angle to view the Mott physics and suggests a renewed possibility to extend the application of the quantum-to-classical mapping in understanding condensed matter physics

cond-mat.str-el↗

arXiv2019

Hybridization fluctuations in the half-filled periodic Anderson model

Motivated by recent photoemission and pump-probe experiments, we report determinant Quantum Monte Carlo simulations of hybridization fluctuations in the half-filled periodic Anderson model. A tentative phase diagram is constructed based solely on hybridization fluctuation spectra and reveals a crossover regime between an unhybridized selective Mott state and a fully hybridized Kondo insulating state. This intermediate phase exhibits nonlocal hybridization fluctuations and consequentially the so-called "band bending" and a direct hybridization gap as observed in angle-resolved photoemission spectroscopy and optical conductivity. This connects the band bending with the nonlocal hybridization fluctuations as proposed in latest ultrafast optical pump-probe experiment. The Kondo insulating state is only established at lower temperatures with the development of sufficiently strong inter-site hybridization correlations. Our work suggests a unified picture for interpreting recent photoemission, pump-probe, and optical observations and provides numerical evidences for the importance of hybridization fluctuations in heavy fermion physics.

cond-mat.str-el↗

arXiv2017

Hybridization oscillation in the one-dimensional Kondo-Heisenberg model with Kondo holes

We use the density matrix renormalization group method to study the properties of the one-dimensional Kondo-Heisenberg model doped with Kondo holes. We find that the perturbation of the Kondo holes to the local hybridization exhibits spatial oscillation pattern and its amplitude decays exponentially with distance away from the Kondo hole sites. The hybridization oscillation is correlated with both the charge density oscillation of the conduction electrons and the oscillation in the correlation function of the Heisenberg spins. In particular, we find that the oscillation wavelength for intermediate Kondo couplings is given by the Fermi wavevector of the large Fermi surface even before it is formed. This suggests that heavy electrons responsible for the oscillation are already present in this regime and start to accumulate around the to-be-formed large Fermi surface in the Brillouin zone.

cond-mat.str-el↗

arXiv2017

Symmetry-enforced heavy-fermion physics in the quadruple-perovskite CaCu3Ir4O12

Heavy-fermion materials are mostly rare-earth or actinide intermetallics with very few exceptions in d-electron systems. The physical mechanism for these d-electron heavy fermion systems remains unclear. Here by studying the quadruple-perovskite CaCu3Ir4O12, we propose a symmetry-based mechanism that may enforce heavy-fermion physics in d-electron systems. We show that electron hoppings between neighboring Cu 3d-orbitals are strictly prohibited by the crystal symmetry, so that Cu 3d-electrons can only become delocalized through hybridization with other more itinerant bands, resembling that in typical heavy-fermion rare-earth intermetallics. This provides a useful way to enforce heavy-fermion physics in d-electron systems and may help future design of new heavy-fermion materials.

cond-mat.str-el↗