Author name cluster

Mei Liu

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

8 papers

2 author rows

EAAI Journal 2025 Journal Article

A network with enhanced ability to interact scale features with channel features

Peng Su
Huizi Han
Mei Liu
Siqun Ma
Jiasheng Chen

Road damage must be accurately and promptly detected because road hazards have the potential to cause catastrophic traffic accidents. A Multiscale channel Shuffle Fusion-You Only Look Once(MSF-YOLO) series algorithm with multiscale cross-channel interaction data is suggested as a solution to this issue. It solves the problem of continuous change of road damage scale during vehicle traveling by enhancing the expressiveness of scale features and improves the network detection performance by enhancing the inter-channel interaction features so that the network learns the original input features as well as the input features after channel interaction with almost no additional computational cost. The experimental results show that Multiscale channel Shuffle Fusion-You Only Look Once-small (MSF-YOLO-s) accuracy on the Global Road Damage Detection Challenge (GRDDC) 2020 dataset is improved by 5. 8% to reach 65. 9%, while MSF-YOLO-s computation amount and number of parameters are reduced by 20. 9% and 8. 7%, respectively, compared to baseline. While experimental verification on the Common Objects in Context (COCO) 2017 datasets confirms its good generalizability. Ultimately, the model is implemented throughout the entire vehicle to anticipate fractures and enhance driving safety.

Details DOI

JBHI Journal 2025 Journal Article

Fine-Grained Temporal Site Monitoring in EGD Streams via Visual Time-Aware Embedding and Vision-Text Asymmetric Coworking

Fang Peng
Hongkuan Shi
Shiquan He
Qiang Hu
Ting Li
Fan Huang
Xinxia Feng
Mei Liu

Esophagogastroduodenoscopy (EGD) requires inspecting plentiful upper gastrointestinal (UGI) sites completely for a precise cancer screening. Automated temporal site monitoring for EGD assistance is thus of high demand, yet often fails if directly applying the existing methods of online action detection. The key challenges are two-fold: 1) the global camera motion dominates, invalidating the temporal patterns derived from the object optical flows, and 2) the UGI sites are fine-grained, yielding highly homogenized appearances. In this paper, we propose an EGD-customized model, powered by two novel designs, i. e. , Visual Time-aware Embedding plus Vision-text Asymmetric Coworking (VTE+VAC), for real-time accurate fine-grained UGI site monitoring. Concretely, VTE learns visual embeddings by differentiating frames via classification losses, and meanwhile by reordering the sampled time-agnostic frames to be temporally coherent via a ranking loss. Such joint objective encourages VTE to capture the sequential relation without resorting to the inapplicable object optical flows, and thus to provide the time-aware frame-wise embeddings. In the subsequent analysis, VAC uses a temporal sliding window, and extracts vision-text multimodal knowledge from each frame and its corresponding textualized prediction via the learned VTE and a frozen BERT. The text embeddings help provide more representative cues, but also may cause misdirection due to prediction errors. Thus, VAC randomly drops or replaces historical predictions to increase the error tolerance to avoid collapsing onto the last few predictions. Qualitative and quantitative experiments demonstrate that the proposed method achieves superior performance compared to other state-of-the-art methods, with an average F1-score improvement of at least 7. 66%.

Details DOI

AAAI Conference 2025 Conference Paper

MonoBox: Tightness-Free Box-Supervised Polyp Segmentation Using Monotonicity Constraint

Qiang Hu
Zhenyu Yi
Ying Zhou
Fan Huang
Mei Liu
Qiang Li
Zhiwei Wang

We propose MonoBox, an innovative box-supervised segmentation method constrained by monotonicity to liberate its training from the user-unfriendly box-tightness assumption. In contrast to conventional box-supervised segmentation, where the box edges must precisely touch the target boundaries, MonoBox leverages imprecisely-annotated boxes to achieve robust pixel-wise segmentation. The 'linchpin' is that, within the noisy zones around box edges, MonoBox discards the traditional misguiding multiple-instance learning loss, and instead optimizes a carefully-designed objective, termed monotonicity constraint. Along directions transitioning from the foreground to background, this new constraint steers responses to adhere to a trend of monotonically decreasing values. Consequently, the originally unreliable learning within the noisy zones is transformed into a correct and effective monotonicity optimization. Moreover, an adaptive label correction is introduced, enabling MonoBox to enhance the tightness of box annotations using predicted masks from the previous epoch and dynamically shrink the noisy zones as training progresses. We verify MonoBox in the box-supervised segmentation task of polyps, where satisfying box-tightness is challenging due to the vague boundaries between the polyp and normal tissues. Experiments on both public synthetic and in-house real noisy datasets demonstrate that MonoBox exceeds other anti-noise state-of-the-arts by improving Dice by at least 5.5% and 3.3%, respectively.

PDF Details DOI

ICRA Conference 2024 Conference Paper

Robust and Remote Center of Cyclic Motion Control for Redundant Robots with Partially Unknown Structure

Long Jin 0001
Kun Liu
Mei Liu

Remote center of motion (RCM) describes a robot with a rod-like end-effector operating through a hole in the interface separating the internal space from the external space. Considering that the control of RCM may be influenced by perturbations (noises) and that the end-effector is frequently replaced to complete different tasks, the structural information related to the robot manipulator and its rod-like end-effector may contain errors. This paper proposes an acceleration-level remote center of cyclic motion (ARC 2 M) control scheme, which takes into account the cyclic motion index and the physical limitations of robot manipulators to achieve repetitive motion planning and RCM control at the acceleration level. Additionally, a parameter calculation method is proposed to compute unknown parameters of the end-effector under the influence of noise. Kalman filter and a neural dynamics-based method are employed to address noises effects, and related theoretical analyses are given. To validate the proposed ARC 2 M scheme, simulations and physical experiments are carried out. The source code is available at https://github.com/LongJin-lab/ARCM.

Details

JBHI Journal 2022 Journal Article

Dynamic Sepsis Prediction for Intensive Care Unit Patients Using XGBoost-Based Model With Novel Time-Dependent Features

Shuhui Liu
Bo Fu
Wen Wang
Mei Liu
Xin Sun

Sepsis is a systemic inflammatory response caused by pathogens such as bacteria. Because its pathogenesis is not clear, the clinical manifestations of patients vary greatly, and the alarming incidence and mortality pose a great threat to patients and medical systems, especially in the ICU (Intensive Care Unit). The traditional judgment criteria have the problem of low specificity. Artificial intelligence models could greatly improve the accuracy of sepsis prediction and judgment. Based on the XGBoost machine learning framework taking demographic, vital signs, laboratory tests and medical intervention data as input, this paper proposes a novel model for dynamically predicting sepsis and assessing risk. To realize the model, two methods for feature construction are introduced. For the observed time-series data of vital signs and laboratory tests, the time-dependent method performs to construct the time-dependent characteristics after the statistical screening. For the clinical intervention data, the statistical counting method is applied to construct count-dependent characteristics. Moreover, a new objective function is proposed for the XGBoost framework, and the first-order and second-order gradients of the objective function are also given for model training. Compared with the state-of-the-art methods at present, the proposed model has the best performance, with AUROC improved by 5. 4% on the MIMIC-III dataset and 2. 1% on PhysioNet Challenge 2019 dataset. The data processing and training methods of this model can be conveniently applied in different electronic health record systems and has a wide application prospect.

Details DOI

AIIM Journal 2017 Journal Article

Identification of adverse drug-drug interactions through causal association rule discovery from spontaneous adverse event reports

Ruichu Cai
Mei Liu
Yong Hu
Brittany L. Melton
Michael E. Matheny
Hua Xu
Lian Duan
Lemuel R. Waitman

Objective Drug-drug interaction (DDI) is of serious concern, causing over 30% of all adverse drug reactions and resulting in significant morbidity and mortality. Early discovery of adverse DDI is critical to prevent patient harm. Spontaneous reporting systems have been a major resource for drug safety surveillance that routinely collects adverse event reports from patients and healthcare professionals. In this study, we present a novel approach to discover DDIs from the Food and Drug Administration’s adverse event reporting system. Methods Data-driven discovery of DDI is an extremely challenging task because higher-order associations require analysis of all combinations of drugs and adverse events and accurate estimate of the relationships between drug combinations and adverse event require cause-and-effect inference. To efficiently identify causal relationships, we introduce the causal concept into association rule mining by developing a method called Causal Association Rule Discovery (CARD). The properties of V-structures in Bayesian Networks are utilized in the search for causal associations. To demonstrate feasibility, CARD is compared to the traditional association rule mining (AR) method in DDI identification. Results Based on physician evaluation of 100 randomly selected higher-order associations generated by CARD and AR, CARD is demonstrated to be more accurate in identifying known drug interactions compared to AR, 20% vs. 10% respectively. Moreover, CARD yielded a lower number of drug combinations that are unknown to interact, i. e. , 50% for CARD and 79% for AR. Conclusion Evaluation analysis demonstrated that CARD is more likely to identify true causal drug variables and associations to adverse event.

Details DOI

JBHI Journal 2013 Journal Article

Adverse Drug Effect Detection

Lian Duan
M. Khoshneshin
W. N. Street
Mei Liu

Large collections of electronic patient records provide abundant but under-explored information on the real-world use of medicines. Although they are maintained for patient administration, they provide a broad range of clinical information for data analysis. One growing interest is drug safety signal detection from these longitudinal observational data. In this paper, we proposed two novel algorithms-a likelihood ratio model and a Bayesian network model-for adverse drug effect discovery. Although the performance of these two algorithms is comparable to the state-of-the-art algorithm, Bayesian confidence propagation neural network, the combination of three works better due to their diversity in solutions. Since the actual adverse drug effects on a given dataset cannot be absolutely determined, we make use of the simulated observational medical outcomes partnership (OMOP) dataset constructed with the predefined adverse drug effects to evaluate our methods. Experimental results show the usefulness of the proposed pattern discovery method on the simulated OMOP dataset by improving the standard baseline algorithm-chi-square-by 23. 83%.

Details DOI

LORI Conference 2011 Conference Paper

Efficient Action Extraction with Many-to-Many Relationship between Actions and Features

Jianfeng Du
Yong Hu 0002
Charles X. Ling
Ming Fan
Mei Liu

Real-world problems often call for efficient methods to discovery actionable knowledge on which business can directly act [3]. Some works for discovering actionable knowledge [3, 5] view actions as behaviors which render a state of an instance into a preferred state, where a state is represented by feature values of the instance and whether a state is preferred is determined by a classifier. Actions usually have many-to-many relations with features of an instance. That is, an action may affect multiple features of an instance, and vise versa, a feature may be influenced by multiple actions. This type of many-to-many relationships between actions and features is prevalent in real-world applications. However, most existing works [3, 5] only deal with one-to-one relationship and ignore manyto- many relationship between actions and features. In these works, an action is treated as a behavior with a fixed execution cost. Restricting to a one-to-one relationship between actions and features may not yield an action set (i. e. a set of actions) with the minimal total execution cost. Moreover, one-to-one relationship is simply a special case of many-to-many relationship, and hence the latter will be applicable to more real-world problems. Therefore we aim to extract action sets from a classifier for which the total execution cost is minimal based on many-to-many relationship between actions and features.

Details