Knowledge graphs play a vital role in numerous artificial intelligence tasks, yet they frequently face the issue of incompleteness. In this study, we explore utilizing Large Language Models (LLM) for knowledge graph completion. We consider triples in knowledge graphs as text sequences and introduce an innovative framework called Knowledge Graph LLM (KG-LLM) to model these triples. Our technique employs entity and relation descriptions of a triple as prompts and utilizes the response for predictions. Experiments on various benchmark knowledge graphs demonstrate that our method attains state-of-the-art performance in tasks such as triple classification and relation prediction. We also find that fine-tuning relatively smaller models (e.g., LLaMA-7B, ChatGLM-6B) outperforms recent ChatGPT and GPT-4.

The scarcity of health care resources, such as ventilators, often leads to the unavoidable consequence of rationing, particularly during public health emergencies or in resource-constrained settings like pandemics. The absence of a universally accepted standard for resource allocation protocols results in governments relying on varying criteria and heuristic-based approaches, often yielding suboptimal and inequitable outcomes. This study addresses the societal challenge of fair and effective critical care resource allocation by leveraging deep reinforcement learning to optimize policy decisions. We propose a transformer-based deep Q-network that integrates individual patient disease progression and interaction effects among patients to enhance allocation decisions. Our method aims to improve both fairness and overall patient outcomes. Experiments using metrics such as normalized survival rates and interracial allocation rate differences demonstrate that our approach significantly reduces excess deaths and achieves more equitable resource allocation compared to severity- and comorbidity-based protocols currently in use. Our findings highlight the potential of deep reinforcement learning to address critical health care challenges.

Large AI models, connected to terminal devices via high-speed mobile communication networks, enable task collaboration and resource sharing, forming an intelligent framework for the Symbiotic Internet of Things (SIoT) paradigm in industry applications. Despite large language models hold significant potential for psychological intervention, their emotional interaction capabilities remain limited. This paper introduces a SIoT framework for psychological intervention and proposes an emotion-enhanced human-machine interaction architecture incorporating behavioral information captured by IoT devices. The system leverages ubiquitous sensing devices such as cameras and microphones, along with technologies like speech recognition and generation, as well as hyper-realistic digital humans, to create a natural interaction interface. Additionally, we introduce the end-of-utterance detection method and the behavior pattern control algorithm to facilitate smoother and more goal-oriented conversations. The proposed methods and prototype system have been validated through subjective and objective experiments, with results demonstrating their feasibility and suggesting that this approach could become one of the primary forms of humanmachine interaction for psychological intervention in the future.

Graph Neural Network (GNN) aggregates the neighborhood information into the node embedding and shows its powerful capability for graph representation learning in various application areas. However, most existing GNN variants aggregate the neighborhood information in a fixed non-injective fashion, which may map different graphs or nodes to the same embedding, detrimental to the model expressiveness. In this paper, we present a theoretical framework to improve the expressive power of GNN by taking both injectivity and continuity into account. Based on the framework, we develop injective and continuous expressive Graph Neural Network (iceGNN) that learns the graph and node representations in an injective and continuous fashion, so that it can map similar nodes or graphs to similar embeddings, and non-equivalent nodes or non-isomorphic graphs to different embeddings. We validate the proposed iceGNN model for graph classification and node classification on multiple benchmark datasets. The experimental results demonstrate that our model achieves state-of-the-art performances on most of the benchmarks.

Pupil dilation and eye movements are closely related to human emotional and cognitive processes. Visual stimulus, especially video clips, is widely used in computer-assisted experimental paradigms as emotional inducers. However, the level of attraction as a critical factor to such visual stimulus still needs comprehensive investigation. This article conducts a novel study of attraction assessment with pupil diameter and eye movements. We collected high temporal resolution ocular variation data from 50 subjects while they viewed a variety of emotional video stimuli. Besides, this article proposes two orthogonal-moment-based feature extraction methods for emotion classification, i.e., Legendre moment and Krawtchouk moment. The results of experiments show that our proposed feature sets achieve better classification performance compared with conventional time- or frequency-domain feature sets. The accuracy of predicting attraction level reached 87.4% and 91.0% when new features were used alone and combined with conventional features, respectively. Compared with the conventional features with an accuracy of 86.6%, our proposed features can improve the accuracy by 4.4%.This study conducts a ground investigation of quantitative attraction assessment for video clips, which might provide reference for paradigm design of affective computing research.

Graph neural network (GNN) is effective to model graphs for distributed representations of nodes and an entire graph. Recently, research on the expressive power of GNN attracted growing attention. A highly expressive GNN has the ability to generate discriminative graph representations. However, in the end-to-end training process for a certain graph learning task, an expressive GNN could generate graph representations overfitting the training data for the target task but losing information important for the model generalization, thus reducing the generalizability. In this paper, we propose Distribution Preserving GNN (DP-GNN), a GNN framework that can improve the generalizability of expressive GNN models by preserving several kinds of distribution information in graph representations and node representations. Besides the generalizability, by applying an expressive GNN backbone, DP-GNN can also have high expressive power. We evaluate the proposed DP-GNN framework on multiple benchmark datasets for graph classification tasks. The experimental results demonstrate that our model achieves state-of-the-art performances.

Image representation is a fundamental task in computer vision. However, most of the existing approaches for image representation ignore the relations between images and consider each input image independently. Intuitively, relations between images can help to understand the images and maintain model consistency over related images, leading to better explainability. In this paper, we consider modeling the image-level relations to generate more informative image representations, and propose ImageGCN, an end-to-end graph convolutional network framework for inductive multi-relational image modeling. We apply ImageGCN to chest X-ray images where rich relational information is available for disease identification. Unlike previous image representation models, ImageGCN learns the representation of an image using both its original pixel features and its relationship with other images. Besides learning informative representations for images, ImageGCN can also be used for object detection in a weakly supervised manner. The experimental results on 3 open-source x-ray datasets, ChestX-ray14, CheXpert and MIMIC-CXR demonstrate that ImageGCN can outperform respective baselines in both disease identification and localization tasks and can achieve comparable and often better results than the state-of-the-art methods.

Genetic pathways usually encode molecular mechanisms that can inform targeted interventions. It is often challenging for existing machine learning approaches to jointly model genetic pathways (higher-order features) and variants (atomic features), and present to clinicians interpretable models. In order to build more accurate and better interpretable machine learning models for genetic medicine, we introduce Pathway Augmented Nonnegative Tensor factorization for HighER-order feature learning (PANTHER). PANTHER selects informative genetic pathways that directly encode molecular mechanisms. We apply genetically motivated constrained tensor factorization to group pathways in a way that reflects molecular mechanism interactions. We then train a softmax classifier for disease types using the identified pathway groups. We evaluated PANTHER against multiple state-of-the-art constrained tensor/matrix factorization models, as well as group guided and Bayesian hierarchical models. PANTHER outperforms all state-of-the-art comparison models significantly (p

BACKGROUND: Genetic information is becoming more readily available and is increasingly being used to predict patient cancer types as well as their subtypes. Most classification methods thus far utilize somatic mutations as independent features for classification and are limited by study power. We aim to develop a novel method to effectively explore the landscape of genetic variants, including germline variants, and small insertions and deletions for cancer type prediction. RESULTS: We proposed DeepCues, a deep learning model that utilizes convolutional neural networks to unbiasedly derive features from raw cancer DNA sequencing data for disease classification and relevant gene discovery. Using raw whole-exome sequencing as features, germline variants and somatic mutations, including insertions and deletions, were interactively amalgamated for feature generation and cancer prediction. We applied DeepCues to a dataset from TCGA to classify seven different types of major cancers and obtained an overall accuracy of 77.6%. We compared DeepCues to conventional methods and demonstrated a significant overall improvement (p < 0.001). Strikingly, using DeepCues, the top 20 breast cancer relevant genes we have identified, had a 40% overlap with the top 20 known breast cancer driver genes. CONCLUSION: Our results support DeepCues as a novel method to improve the representational resolution of DNA sequencings and its power in deriving features from raw sequences for cancer type prediction, as well as discovering new cancer relevant genes.

Text classification is an important and classical problem in natural language processing. There have been a number of studies that applied convolutional neural networks (convolution on regular grid, e.g., sequence) to classification. However, only a limited number of studies have explored the more flexible graph convolutional neural networks (convolution on non-grid, e.g., arbitrary graph) for the task. In this work, we propose to use graph convolutional networks for text classification. We build a single text graph for a corpus based on word co-occurrence and document word relations, then learn a Text Graph Convolutional Network (Text GCN) for the corpus. Our Text GCN is initialized with one-hot representation for word and document, it then jointly learns the embeddings for both words and documents, as supervised by the known class labels for documents. Our experimental results on multiple benchmark datasets demonstrate that a vanilla Text GCN without any external word embeddings or knowledge outperforms state-of-the-art methods for text classification. On the other hand, Text GCN also learns predictive word and document embeddings. In addition, experimental results show that the improvement of Text GCN over state-of-the-art comparison methods become more prominent as we lower the percentage of training data, suggesting the robustness of Text GCN to less training data in text classification.

MOTIVATION: Hypertension is a heterogeneous syndrome in need of improved subtyping using phenotypic and genetic measurements with the goal of identifying subtypes of patients who share similar pathophysiologic mechanisms and may respond more uniformly to targeted treatments. Existing machine learning approaches often face challenges in integrating phenotype and genotype information and presenting to clinicians an interpretable model. We aim to provide informed patient stratification based on phenotype and genotype features. RESULTS: In this article, we present a hybrid non-negative matrix factorization (HNMF) method to integrate phenotype and genotype information for patient stratification. HNMF simultaneously approximates the phenotypic and genetic feature matrices using different appropriate loss functions, and generates patient subtypes, phenotypic groups and genetic groups. Unlike previous methods, HNMF approximates phenotypic matrix under Frobenius loss, and genetic matrix under Kullback-Leibler (KL) loss. We propose an alternating projected gradient method to solve the approximation problem. Simulation shows HNMF converges fast and accurately to the true factor matrices. On a real-world clinical dataset, we used the patient factor matrix as features and examined the association of these features with indices of cardiac mechanics. We compared HNMF with six different models using phenotype or genotype features alone, with or without NMF, or using joint NMF with only one type of loss We also compared HNMF with 3 recently published methods for integrative clustering analysis, including iClusterBayes, Bayesian joint analysis and JIVE. HNMF significantly outperforms all comparison models. HNMF also reveals intuitive phenotype-genotype interactions that characterize cardiac abnormalities. AVAILABILITY AND IMPLEMENTATION: Our code is publicly available on github at https://github.com/yuanluo/hnmf. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Bioinformatics (2018) doi: 10.1093/bioinformatics/bty804 In the abstract, the availability and implementation section has been updated to include the link for the code, as follows: Our code is publicly available on github at https://github.com/yuanluo/hnmf.

Traditional Chinese Medicine (TCM) has been developed for several thousand years and plays a significant role in health care for Chinese people. This paper studies the problem of classifying TCM clinical records into 5 main disease categories in TCM. We explored a number of state-of-the-art deep learning models and found that the recent Bidirectional Encoder Representations from Transformers can achieve better results than other deep learning models and other state-of-the-art methods. We further utilized an unlabeled clinical corpus to fine-tune the BERT language model before training the text classifier. The method only uses Chinese characters in clinical text as input without preprocessing or feature engineering. We evaluated deep learning models and traditional text classifiers on a benchmark data set. Our method achieves a state-of-the-art accuracy 89.39% ± 0.35%, Macro F1 score 88.64% ± 0.40% and Micro F1 score 89.39% ± 0.35%. We also visualized attention weights in our method, which can reveal indicative characters in clinical text.

Privacy is a major concern in sharing human subject data to researchers for secondary analyses. A simple binary consent (opt-in or not) may significantly reduce the amount of sharable data, since many patients might only be concerned about a few sensitive medical conditions rather than the entire medical records. We propose event-level privacy protection, and develop a feature ablation method to protect event-level privacy in electronic medical records. Using a list of 13 sensitive diagnoses, we evaluate the feasibility and the efficacy of the proposed method. As feature ablation progresses, the identifiability of a sensitive medical condition decreases with varying speeds on different diseases. We find that these sensitive diagnoses can be divided into three categories: (1) five diseases have fast declining identifiability (AUC below 0.6 with less than 400 features excluded); (2) seven diseases with progressively declining identifiability (AUC below 0.7 with between 200 and 700 features excluded); and (3) one disease with slowly declining identifiability (AUC above 0.7 with 1,000 features excluded). The fact that the majority (12 out of 13) of the sensitive diseases fall into the first two categories suggests the potential of the proposed feature ablation method as a solution for event-level record privacy protection.

High dimensional biomedical datasets contain thousands of features which can be used in molecular diagnosis of disease, however, such datasets contain many irrelevant or weak correlation features which influence the predictive accuracy of diagnosis. Without a feature selection algorithm, it is difficult for the existing classification techniques to accurately identify patterns in the features. The purpose of feature selection is to not only identify a feature subset from an original set of features [without reducing the predictive accuracy of classification algorithm] but also reduce the computation overhead in data mining. In this paper, we present our improved shuffled frog leaping algorithm which introduces a chaos memory weight factor, an absolute balance group strategy, and an adaptive transfer factor. Our proposed approach explores the space of possible subsets to obtain the set of features that maximizes the predictive accuracy and minimizes irrelevant features in high-dimensional biomedical data. To evaluate the effectiveness of our proposed method, we have employed the K-nearest neighbor method with a comparative analysis in which we compare our proposed approach with genetic algorithms, particle swarm optimization, and the shuffled frog leaping algorithm. Experimental results show that our improved algorithm achieves improvements in the identification of relevant subsets and in classification accuracy.

High-dimensional biomedical datasets contain thousands of features used in molecular disease diagnosis, however many irrelevant or weak correlation features influence the predictive accuracy. Feature selection algorithms enable classification techniques to accurately identify patterns in the features and find a feature subset from an original set of features without reducing the predictive classification accuracy while reducing the computational overhead in data mining. In this paper we present an improved shuffled frog leaping algorithm (ISFLA) which explores the space of possible subsets to obtain the set of features that maximizes the predictive accuracy and minimizes irrelevant features in high-dimensional biomedical data. Evaluation employs the K-nearest neighbour approach and a comparative analysis with a genetic algorithm, particle swarm optimization and the shuffled frog leaping algorithm shows that our improved algorithm achieves improvements in the identification of relevant subsets and in classification accuracy.

In computer-aided diagnosis, a Bayesian classifier that can give the class membership probabilities should be more favorable than classifiers that only give a class assertion. In Bayesian classification, an important and critical step is the probability distribution estimation for each class. Existing methods usually estimate the probability distribution in the whole sample space where the original distribution may be too complex to model. In this paper, we propose a probability distribution estimation method based on local probabilistic model assumption. In our method, the estimation of global probability for a certain point is transformed to the computation of local distribution in a small region, where the local distribution is supposed to be simpler and can be assumed as a simpler probabilistic model. By this method, we implement the Bayesian classifiers based on several local probabilistic model assumptions, and experiments with these classifier have been conducted on several real-word biological and medical datasets; the experimental results demonstrate the efficacy of the proposed method for probabilistic classification in medical diagnosis.

In this paper we propose a biometric solution for individual identification based on electroencephalography with classification using local probability centers. In our study, the electroencephalography signals of a subject are recorded from only one active channel Cz with eyes closed and without any external stimulations. The original signals are preprocessed by Haar wavelet transformation; then a number of features are extracted from the preprocessed signals; and then a classifier with local probability centers are employed to assign the signals to the right person according to the features extracted. By this method we have achieved an average identification accuracy of 96.21% for a dataset of 11 subjects' electroencephalography patterns and the high F-measure values of different persons has shown that this method performed robustly and effectively for various subjects. In addition, we have studied the variation of recognition accuracy with the time length of electroencephalography sessions and found that a longer electroencephalography session is usually more effective for individual identification. These results are in agreement to the previous research and show the evidence that the electroencephalography carries identity information and a longer electroencephalography session usually carries more identity information. With the simple implementation and good performance, we consider our proposed approach to be suitable for development and implementation in a ‘unimodal’ biometric identification system or may be combined with other biometric methods to form a ‘multimodal’ biometric identification system.

The k-nearest neighbor method generates predictions for a particular instance from its neighborhood. It is a simple but effective supervised method for classification. However, the traditional k-nearest neighbor algorithm using the majority voting rule for the class label usually loses a part of useful information in the neighborhood. This paper tries to learn from the neighborhood for more useful information for classification and proposes an improved version of k-nearest neighbor method by heuristically organizing the local distribution characteristics. Different from the traditional methods, the proposed method considers the neighborhood of a query sample from the perspective of local distribution and learns from the neighborhood for local distribution characteristics for classification. We analyze the impact of local distribution characteristics on classification and heuristically develop a formulation to estimate the membership degree, which indicates the level of membership of a query sample to each class; then the query sample is classified to the class which has the highest membership degree with respect to the query sample. Experiments have been conducted on several real data sets; the results support the conclusion that the proposed method is superior to the traditional voting k-nearest neighbor method and comparable with or better than several state-of-the-art methods in terms of classification performance and robustness.

Accounting for a patient's emotional state is integral in medical care. Positive emotions have a significant influence on mental and physical health. Much research has been carried out on the impact of emotion on the development and course of different illnesses. Emotion and significant contexts have an important role in helping people cope with depression, but lack of support undermines coping. In this paper, we present an emotion-focused e-health monitoring support system in assisting them to cope with their illness. In the system proposed, a user could monitor his emotional states in the absence of a doctor and regulate his emotions through the system. To provide personalized health care services to the user anywhere and anytime, the system should convert low-level multimodal context (including physiological signal, user profile and environment information) to high-level context. The objective of this research is establishing an emotion-focused e-health system for health care services.