Unified fair federated learning for digital healthcare.

Federated learning (FL) is a promising approach for healthcare institutions to train high-quality medical models collaboratively while protecting sensitive data privacy. However, FL models encounter fairness issues at diverse levels, leading to performance disparities across different subpopulations. To address this, we propose Federated Learning with Unified Fairness Objective (FedUFO), a unified framework consolidating diverse fairness levels within FL. By leveraging distributionally robust optimization and a unified uncertainty set, it ensures consistent performance across all subpopulations and enhances the overall efficacy of FL in healthcare and other domains while maintaining accuracy levels comparable with those of existing methods. Our model was validated by applying it to four digital healthcare tasks using real-world datasets in federated settings. Our collaborative machine learning paradigm not only promotes artificial intelligence in digital healthcare but also fosters social equity by embodying fairness.

Simulating doctors' thinking logic for chest X-ray report generation via Transformer-based Semantic Query learning.

Medical report generation can be treated as a process of doctors' observing, understanding, and describing images from different perspectives. Following this process, this paper innovatively proposes a Transformer-based Semantic Query learning paradigm (TranSQ). Briefly, this paradigm is to learn an intention embedding set and make a semantic query to the visual features, generate intent-compliant sentence candidates, and form a coherent report. We apply a bipartite matching mechanism during training to realize the dynamic correspondence between the intention embeddings and the sentences to induct medical concepts into the observation intentions. Experimental results on two major radiology reporting datasets (i.e., IU X-ray and MIMIC-CXR) demonstrate that our model outperforms state-of-the-art models regarding generation effectiveness and clinical efficacy. In addition, comprehensive ablation experiments fully validate the TranSQ model's innovation and interpretation. The code is available at https://github.com/zjukongming/TranSQ.

Operational and Intervention Effects of Targeted Tuina in Lumbar Intervertebral Disc Degeneration Model Rabbits.

Tui Na or massage therapy alleviates symptoms related to intervertebral disc degeneration (IDD). However, precise, repeatable, standardized instructions for Tuina manipulation are lacking. This study establishes IDD model rabbits induced by fibrous ring puncture, creates targeted Tuina stimulation protocols at the acupuncture points in the lumbar region, and describes in detail the operation methods and requirements of kneading, pointing, and flicking. New Zealand male white rabbits (n = 15) were selected and randomly divided into a blank group, a model group, and a Tuina group. The rabbits in the model group and the Tuina group were molded by fibrous ring puncture; the rabbits in the model group were only immobilized on the operating table without treatment. In contrast, the Tuina group used the "8N/10N, 30 cycles/min" prescription for kneading, pointing, and flicking to perform the intervention, using tactile sensory aids to monitor and regulate the intensity of the Tuina operation. Imaging diagnosis and pathological tests were used to assess the effect of Tuina in rabbits, and the results showed improved imaging features and significantly lowered pathology scores of lumbar disc degeneration in the Tuina group compared to the model group (P < 0.01). Targeted Tuina in the lumbar region may be beneficial in the alleviation of lumbar disc degeneration, but further verification is needed. By regularly performing Tuina and recording the mechanical information involved enables reproducible manipulation prescriptions and helps to observe the basic features of the underlying mechanism of Tuina for IDD.

Personalized Latent Structure Learning for Recommendation.

In recommender systems, users' behavior data are driven by the interactions of user-item latent factors. To improve recommendation effectiveness and robustness, recent advances focus on latent factor disentanglement via variational inference. Despite significant progress, uncovering the underlying interactions, i.e., dependencies of latent factors, remains largely neglected by the literature. To bridge the gap, we investigate the joint disentanglement of user-item latent factors and the dependencies between them, namely latent structure learning. We propose to analyze the problem from the causal perspective, where a latent structure should ideally reproduce observational interaction data, and satisfy the structure acyclicity and dependency constraints, i.e., causal prerequisites. We further identify the recommendation-specific challenges for latent structure learning, i.e., the subjective nature of users' minds and the inaccessibility of private/sensitive user factors causing universally learned latent structure to be suboptimal for individuals. To address these challenges, we propose the personalized latent structure learning framework for recommendation, namely PlanRec, which incorporates 1) differentiable Reconstruction, Dependency, and Acyclicity regularizations to satisfy the causal prerequisites; 2) Personalized Structure Learning (PSL) which personalizes the universally learned dependencies through probabilistic modeling; and 3) uncertainty estimation which explicitly measures the uncertainty of structure personalization, and adaptively balances personalization and shared knowledge for different users. We conduct extensive experiments on two public benchmark datasets from MovieLens and Amazon, and a large-scale industrial dataset from Alipay. Empirical studies validate that PlanRec discovers effective shared/personalized structures, and successfully balances shared knowledge and personalization via rational uncertainty estimation.

Attribute-aware interpretation learning for thyroid ultrasound diagnosis.

Thyroid nodule diagnosis from ultrasound images is a critical computer-aided diagnosis task. Previous works tried to imitate the doctor's diagnosis logic by considering the key attributes to improve the diagnosis performance and explaining the conclusion. However, their clinical feasibilities are still ambiguous because of the ignorance of the correlation between attribute features and global characteristics, as well as the lack of clinical effectiveness evaluation of result interpretations. Following the common logic of ultrasonic investigation, we design a novel Attribute-Aware Interpretation Learning (AAIL) model, consisting of attribute properties discovery module and attribute-global feature fusion module. Adequate result interpretation ensures reliability and transparency of diagnostic conclusions, including the visualization of attribute features and the relationship between attributes and the global feature. Extensive experiments on a practical dataset demonstrate the model's effectiveness, and an innovative human-computer collaborative experiment demonstrates the auxiliary diagnostic ability of the interpretations that can benefit professional doctors.

Debiased Graph Neural Networks With Agnostic Label Selection Bias.

Most existing graph neural networks (GNNs) are proposed without considering the selection bias in data, i.e., the inconsistent distribution between the training set with the test set. In reality, the test data are not even available during the training process, making selection bias agnostic. Training GNNs with biased selected nodes leads to significant parameter estimation bias and greatly impacts the generalization ability on test nodes. In this article, we first present an experimental investigation, which clearly shows that the selection bias drastically hinders the generalization ability of GNNs, and theoretically proves that the selection bias will cause the biased estimation on GNN parameters. Then to remove the bias in GNN estimation, we propose a novel debiased GNNs (DGNN) with a differentiated decorrelation regularizer. The differentiated decorrelation regularizer estimates a sample weight for each labeled node such that the spurious correlation of learned embeddings could be eliminated. We analyze the regularizer in causal view and it motivates us to differentiate the weights of the variables based on their contribution to the confounding bias. Then, these sample weights are used for reweighting GNNs to eliminate the estimation bias, and thus, help to improve the stability of prediction on unknown test nodes. Comprehensive experiments are conducted on several challenging graph datasets with two kinds of label selection biases. The results well verify that our proposed model outperforms the state-of-the-art methods and DGNN is a flexible framework to enhance existing GNNs.

Underlying mechanism of the photodynamic activity of hematoporphyrin­induced apoptosis in U87 glioma cells.

Photodynamic therapy (PDT) is a relatively novel type of tumor therapy method with low toxicity and limited side­effects. The aim of the present study was to investigate the underlying mechanism and potential microRNAs (miRNAs) involved in the treatment of glioma by PDT with hematoporphyrin, a clinical photosensitizer. The photodynamic activity of hematoporphyrin on the cell viability and apoptosis of gliomas was investigated by MTT, and flow cytometry and fluorescence microscopy, respectively. Alterations in singlet oxygen and mitochondrial membrane potential were detected. The differentially expressed miRNAs and proteins were evaluated by miRNA gene chip and apoptosis­associated protein chip, respectively. The results demonstrated that cell viability significantly decreased with hematoporphyrin concentration. PDT with hematoporphyrin significantly increased cell apoptosis at a later stage, induced the content of reactive oxygen species (ROS) and decreased the mitochondrial membrane potential, indicating that PDT with hematoporphyrin inhibited cell growth via induction of radical oxygen, decreased the mitochondrial membrane potential and induced apoptosis. The upregulated miRNAs, including hsa­miR­7641, hsa­miR­9500, hsa­miR­4459, hsa­miR­21­5p, hsa­miR­663a and hsa­miR­205­5p may be important in PDT­induced cell apoptosis in glioma. Transporter 1, ATP binding cassette subfamily B member­ and nuclear factor­κB­mediated apoptosis signaling pathways were the most significant pathways. Thus, the current study presents PDT as a potential therapeutic approach for the treatment of malignant glioma, and identified miRNAs for the molecular design and development of a third­generation photosensitizer (PS).