Self-supervised monocular depth estimation serves as a key task in the development of endoscopic navigation systems. However, performance degradation persists due to uneven illumination inherent in endoscopic images, particularly in low-intensity regions. Existing low-light enhancement techniques fail to effectively guide the depth network. Furthermore, solutions from other fields, like autonomous driving, require well-lit images, making them unsuitable and increasing data collection burdens. To this end, we present DeLightMono - a novel self-supervised monocular depth estimation framework with illumination decoupling. Specifically, endoscopic images are represented by a designed illumination-reflectance-depth model, and are decomposed with auxiliary networks. Moreover, a self-supervised joint-optimizing framework with novel losses leveraging the decoupled components is proposed to mitigate the effects of uneven illumination on depth estimation. The effectiveness of the proposed methods was rigorously verified through extensive comparisons and an ablation study performed on two public datasets.

Decoupling domain-variant information (DVI) from domain-invariant information (DII) serves as a prominent strategy for mitigating domain shifts in the practical implementation of deep learning algorithms. However, in medical settings, concerns surrounding data collection and privacy often restrict access to both training and test data, hindering the empirical decoupling of information by existing methods. To tackle this issue, we propose an Adaptive Information Filter-driven Source-free Domain Adaptation (AIF-SFDA) algorithm, which leverages a frequency-based learnable information filter to autonomously decouple DVI and DII. Information Bottleneck (IB) and Self-supervision (SS) are incorporated to optimize the learnable frequency filter. The IB governs the information flow within the filter to diminish redundant DVI, while SS preserves DII in alignment with the specific task and image modality. Thus, the adaptive information filter can overcome domain shifts relying solely on target data. A series of experiments covering various medical image modalities and segmentation tasks were conducted to demonstrate the benefits of AIF-SFDA through comparisons with leading algorithms and ablation studies.

In recent years, the application value of monocular depth estimation under endoscopic environments has become increasingly prominent in the medical field. However, existing models exhibit low accuracy and limited generalizability in zero-shot scenarios, making them insufficient to meet clinical demands. To address this, this paper proposes a monocular depth estimation method adapted to endoscopic scenarios, constructing a self-supervised learning architecture that integrates a parameter self-estimation module with a depth extraction network. The core innovation of this framework focuses on the lightweight focal length estimation submodule within the parameter self-estimation module: on one hand, by employing a shared feature reuse strategy combined with a lightweight convolutional structure and residual connections, accurate focal length estimation is efficiently achieved; on the other hand, using an isolated training scheme, the focal length estimation is processed in separate warm-up and unfreezing stages, enhancing the main model's performance without affecting its convergence and significantly improving zero-shot training capabilities. Experimental results on the SCARED dataset and the Hamlyn zero-shot dataset demonstrate the superior performance of this framework, fully validating its effectiveness and practicality.

Multimodal representation, as an application framework of social computing, enables researchers to utilize multimodal data for more accurate predictive analysis of mortality rates. In lengthy clinical texts, there are numerous neutral words that do not directly reflect the patient's condition. However, the prevalence of frequently occurring neutral words diminishes the weights of key terms that are directly relevant to the patient's condition, resulting in an imbalance in the allocation of text feature weights. To address this issue, we propose multimodal mortality prediction-multilevel semantic hypergraph (MMP-MSH), a medical multimodal model based on a multilevel semantic hypergraph. Specifically, we approach clinical text in two ways. First, we employ a CNN to extract textual features directly. Second, the text processed into hypergraph is subjected to multilevel GCN to obtain global hypergraph information, which is then introduced into the model training process and combined with the features of each batch of clinical texts. We conducted experiments on the MIMIC-III dataset to evaluate the effectiveness of MMP-MSH in predicting mortality rates.

Cataract surgery, a widely performed operation worldwide, is incorporating semantic segmentation to advance computer-assisted intervention. However, the tissue appearance and illumination in cataract surgery often differ among clinical centers, intensifying the issue of domain shifts. While domain adaptation offers remedies to the shifts, the necessity for data centralization raises additional privacy concerns. To overcome these challenges, we propose a Multi-view Test-time Adaptation algorithm (MUTA) to segment cataract surgical scenes, which leverages multi-view learning to enhance model training within the source domain and model adaptation within the target domain. In the training phase, the segmentation model is equipped with multi-view decoders to boost its robustness against variations in cataract surgery. During the inference phase, test-time adaptation is implemented using multi-view knowledge distillation, enabling model updates in clinics without data centralization or privacy concerns. We conducted experiments in a simulated cross-center scenario using several cataract surgery datasets to evaluate the effectiveness of MUTA. Through comparisons and investigations, we have validated that MUTA effectively learns a robust source model and adapts the model to target data during the practical inference phase. Code and datasets are available at https://github.com/liamheng/CAI-algorithms.

The goal of the Knowledge Tracing (KT) task is to accurately predict a student’s aptitude in answering the next question based on their previous responses. Recent studies have shown promising results by employing pre-training models to capture general feature representations between questions and skills, and subsequently fine-tuning these models for the KT task. However, these methods still face challenges in accurately representing question difficulty and fail to consider the impact of feature fusion during pre-training. Additionally, existing models do not effectively harness the high-level semantic information available after pre-training during fine-tuning, resulting in an underutilization of their potential applications. To this end, this paper proposes Self-attentive Knowledge Tracing (Self-KT) with Feature Fusion Pre-training in the Online Education domain to address these challenges. Self-KT introduces a novel representation of question difficulty and innovatively implements dynamic feature fusion to obtain question embeddings. Furthermore, it enhances the self-attention mechanism by considering the influence of subsequent questions on the current question. We implemented Self-KT on multiple publicly available datasets, and the results demonstrated its significant superiority over the current state-of-the-art methods in knowledge tracing.

Medical imaging provides many valuable clues involving anatomical structure and pathological characteristics. However, image degradation is a common issue in clinical practice, which can adversely impact the observation and diagnosis by physicians and algorithms. Although extensive enhancement models have been developed, these models require a well pre-training before deployment, while failing to take advantage of the potential value of inference data after deployment. In this paper, we raise an algorithm for source-free unsupervised domain adaptive medical image enhancement (SAME), which adapts and optimizes enhancement models using test data in the inference phase. A structure-preserving enhancement network is first constructed to learn a robust source model from synthesized training data. Then a teacher-student model is initialized with the source model and conducts source-free unsupervised domain adaptation (SFUDA) by knowledge distillation with the test data. Additionally, a pseudo-label picker is developed to boost the knowledge distillation of enhancement tasks. Experiments were implemented on ten datasets from three medical image modalities to validate the advantage of the proposed algorithm, and setting analysis and ablation studies were also carried out to interpret the effectiveness of SAME. The remarkable enhancement performance and benefits for downstream tasks demonstrate the potential and generalizability of SAME. The code is available at https://github.com/liamheng/Annotation-free-Medical-Image-Enhancement.

In the Intensive Care Unit (ICU), mortality prediction tasks primarily rely on clinical records consisting of patients’ clinical time series data and physicians’ diagnostic opinions. However, due to the irregularity of time series data and clinical records, existing medical multimodal models focus on generating complete and regular data to address this issue, neglecting the impact of generated modality data on multimodal feature fusion. Inaccurate generation of modality data may lead to noise. To overcome this problem, we propose a novel medical multimodal model named MedCT-BERT in this paper. Specifically, the existing models can address the irregularities in time series data but are limited to handling fixed time series data. We optimize their hyperparameters for better performance. By doing so, our model can process and generate complete regular imputed values in parallel for time series data with varying missing values and sequence lengths. To iteratively refine the generated imputed values, we introduce a time series feature correlation information to reduce noise in multimodal data fusion. We conduct experiments on the MIMIC-III dataset with MedCT-BERT and validate the effectiveness of the model in mortality prediction tasks.

Dropout prediction is an important task due to the high attrition rate commonly found on the massive open online courses (MOOC) platforms. Researchers usually use neural networks to establish various prediction models based on the behavioral features of student data. However, the existing methods ignore the periodic feature of learning behaviors and the influence of learning time distribution information on the prediction results, resulting in the potential association relationship between the input data is not learned by the model. Thus, after in-depth analysis of MOOC learners' behavior data, this paper proposes the concept of periodic feature, and found that different learning time has different effects on the prediction results. Based on the gained insights, we propose a hybrid neural network model (CGDC-LSTM) to model and to predict users' dropout behavior. CGDC-LSTM utilizes Convolutional Neural Network (CNN) to maintain the local correlation of students' behavior, and uses a module combining Group Convolution and Dilated Causal Convolution to fit the periodic feature of students, and combines Long Short-Term Memory Network (LSTM) into the model to extract the learning time distribution information to capture the influence of different learning periods on the results. Experimental results on the KDD Cup 2015 dataset demonstrate that the proposed model shows better prediction performance compared to baseline methods.

In image-text multimodal image classification, the fusion is usually between text features and image features. Such fusion assumes image features from multiple network layers are unitedly fused with text features synchronously. In fact, these image features have different interactions with text features. These interactions have mutual interference in conventional fusion. And since the low-level image features are semantically weak, the fusion between them and text features is not as effective as that between high-level image features and text features. To solve problems above, the paper proposed a framework of multi-level independent fusion between text features and different-level image features. In this framework, the fusions between text features and multi-level image features are conducted asynchronously, where the fusions are independent from each other. Moreover, to improve the fusion efficiency when text features are fused with low-level image features, our method complement semantic information for low-level image features with Twin Pyramid (TP) module which can propagate semantic information top down to them. Substantial experiments on MIMIC-CXR data sets demonstrate that the multi-level independent fusion can effectively concatenate the image features and text features and outperforms the traditional methods.

With the continuous expansion and deepening of smart city construction, research has focused on establishing the best method for evaluating the level of smart city development. Because of the direct influence of index weight on the evaluation results, determining index weights for smart city evaluation is a challenging and strategically important problem. Generally, previous research has used the expert scoring method or the entropy method to determine the weights of indexes in the evaluation of smart cities. However, the number of samples used by these methods is far fewer than optimal, and these methods have provided low accuracy and poor performance. In addition, expert scoring leads to results that are strongly subjective, and it fails to consider the conflict between indexes that always accompanies the process of determining weights using traditional methods. To address these issues, we proposed CV-CRITIC, an index weight determination method based on the CRITIC method that can be used effectively in smart city evaluation. Use of an objective weighting method such as CRITIC can reduce the effect of subjective factors. The CV-CRITIC method used an improved weighted algorithm based on objective data that considered the influence of two factors on the index weights: the information contained in the index itself, and the conflict between indicators. In addition, we introduced the Coefficient of Variance method into the index weight determination process to analyze the important differences between the two factors, and to determine the combined weights. The comparative results of two sets of experiments verified the advantages of our proposed method in terms of objectivity, simplicity and accuracy.