Infrared image generation is essential in scenarios with low illumination or complex environments, but the scarcity of aligned visible–infrared data and the lack of physical realism in generated results remain key challenges. However, existing generative models often overlook the thermodynamic principles underlying infrared imaging, resulting in synthetic images that are visually plausible yet physically inaccurate. In this paper, we propose Infrared Physics-guided Latent Diffusion (IPLD), a novel framework that integrates physics-guided modeling into a latent diffusion process for high-fidelity synthesis of infrared images. Central to IPLD is the Temperature–Emissivity–Environmental Radiance (TeR) decomposition, which decomposes thermal signals into temperature, emissivity, and environmental radiance components, governed by the laws of blackbody radiation. To enhance the environmental radiance modeling, we introduce Environmental Radiance Map Estimation (ERME), a hybrid local–global estimation mechanism that preserves both spatial detail and thermal consistency. Furthermore, a novel Skip Connection Diffusion Transformer (SCDT) is proposed to strengthen and balance semantic structure and fine-grained details during image reconstruction. Extensive experiments on public datasets demonstrate that IPLD outperforms state-of-the-art Generative Adversarial Network (GAN)-based and diffusion-based methods, achieving superior results in Structural Similarity Index Measure (SSIM), Peak Signal-to-Noise Ratio (PSNR), Learned Perceptual Image Patch Similarity (LPIPS), and Fréchet Inception Distance (FID) metrics. Ablation studies validate the complementary value of TeR, ERME, and SCDT in improving radiative realism. Our approach establishes a new paradigm for physically grounded image translation, offering enhanced generalization and reliability for downstream perception tasks such as target detection.

Deep learning advances neural decoding in functional magnetic resonance imaging (fMRI) tasks with convolution and attention-based methods. However, these methods struggle with capturing global spatiotemporal information due to high dimensionality, noise and inter-individual difference of fMRI, which also increase computational complexity and prior bias. To this end, a novel causal state space model, Causal fMRI-Mamba, is proposed for neural decoding and task state mapping. It effectively captures global spatiotemporal information via eliminating local redundancies and capturing long-distance dependencies. Meanwhile, a causal representation framework is designed to extract invariant high-order features and disentangle related causal features, enhancing model performance. Furthermore, a dense connection module is expanded to prevent significant causal information loss in hidden states of inter layers. On the HCP brain task state classification task, Causal fMRI-Mamba achieves better performance and generalization than comparison methods.

In practical medical diagnosis, small sample sizes in gene expression profiling data can lead to overfitting. Addressing this, we leverage the potential of the Conditional Wasserstein Generative Adversarial Network with Gradient Penalty (CWGAN-GP) to amplify data volumes. However, it lacks control over the locations of generated samples and struggles to get a better balance between discriminators and generators during training. To overcome these hurdles, we propose the Improved CWGAN-GP, implementing two critical improvements. The first involves the adoption of a data segmentation strategy based on sample influence scores. By calculating the influence score for each sample, we prioritize samples at decision boundaries and outside the distributions as the training set, thus yielding more explicit decision boundaries. The second enhancement is that a depth feature constraint based on the Pearson correlation coefficient is proposed. Here, an encoder extracts the deep features, applying a constraint between the noise and deep features guided by the Pearson correlation coefficient. This strategy navigates the model closer to a Nash equilibrium. Empirical evaluations conducted on six publicly available gene expression profiling datasets validate our approach, demonstrating that it not only generates higher quality samples but also showcases superior stability compared to existing methods.

The utilization of both constrained and unconstrained-based optimization for solving constrained multi-objective optimization problems (CMOPs) has become prevalent among recently proposed constrained multiobjective evolutionary algorithms (CMOEAs). However, the constrained-based optimization which adopted by many CMOEAs typically gives priority to feasible solutions over infeasible ones regardless of their objective values, potentially leading to degraded performance due to the elimination of promising infeasible solutions with strong convergence and diversity. Furthermore, many existing CMOEAs have difficulty in maintaining diversity while focusing on feasibility, thereby hindering their ability to effectively address CMOPs characterized by complex feasible regions. To tackle these challenges, a constraint-Pareto dominance relationship is proposed in this paper to evaluate solutions based on both objectives and feasibility, to improve the optimization potential by reduce the elimination probability of promising infeasible solutions. A diversity enhancement strategy is also designed to enable simultaneously focus on both diversity and feasibility, thus effectively ensuring the diversity of the feasible solutions obtained. Empirical results from benchmark suites and real-world problems demonstrate that our proposed algorithm surpasses state-of-the-art CMOEAs.

Abstract Feature selection can be seen as a multi-objective task, where the goal is to select a subset of features that exhibit minimal correlation among themselves while maximizing their correlation with the target label. Multi-objective particle swarm optimization algorithm (MOPSO) has been extensively utilized for feature selection and has achieved good performance. However, most MOPSO-based feature selection methods are random and lack knowledge guidance in the initialization process, ignoring certain valuable prior information in the feature data, which may lead to the generated initial population being far from the true Pareto front (PF) and influence the population’s rate of convergence. Additionally, MOPSO has a propensity to become stuck in local optima during the later iterations. In this paper, a novel feature selection method (fMOPSO-FS) is proposed. Firstly, with the aim of improving the initial solution quality and fostering the interpretability of the selected features, a novel initialization strategy that incorporates prior information during the initialization process of the particle swarm is proposed. Furthermore, an adaptive hybrid mutation strategy is proposed to avoid the particle swarm from getting stuck in local optima and to further leverage prior information. The experimental results demonstrate the superior performance of the proposed algorithm compared to the comparison algorithms. It yields a superior feature subset on nine UCI benchmark datasets and six gene expression profile datasets.

With the surge of biomedical data science, more and more AI techniques are employed to discover knowledge, unveil latent data behavior, generate new insight, and seek optimal strategies in decision making.Different AI methods have been proposed and developed in almost all different biomedical data science fields that range from drug discovery, electronic medical records (EMRs) data automation, single-cell RNA sequencing, early disease diagnosis, COVID research, and healthcare analytics.The AI methods and systems also generate a massive amount of data or big data that not only bring unpreceded progress in biomedical fields but also new challenges for AI.One of the key challenges should be the explainability of AI in biomedical data science problem-solving.It refers to that an AI method or system should not only bring good results but also have good interpretability, i.e., let users know why this way is the optimal one rather than the others.The existing AI methods employed in biomedical data science generally lack good explainability and may not create trustworthiness and transparency in usage well.For example, a deep learning model may bring good accuracy in disease diagnosis by analyzing corresponding bioimages, but it can be hard to explain well about the setting of thousands of parameters in the model.It can be possible that some small perturbations of the parameters may generate totally different learning results and challenge the robustness and stability of the deep learning model.Since AI models cannot explain themselves well, it is likely to encounter a high risk to make an incorrect decision making and decrease its trustworthiness and reliability, even if it has the advantage in accuracy, speed, or complicate data relationship revealing.On the other hand, the AI interpretation issue has been raised almost ten years ago in some subfield of biomedical data science such as bioinformatics.For instance, bioinformaticians found that the gene markers or network markers recommended from an AI disease diagnosis system may not explain themselves, i.e., the identified markers not only cannot apply themselves well in clinical practice, but also those markers that do well in the clinical practice may not be recommended from the AI system [1].

Deep learning models are more capable of handling large and complex datasets that generally appear in the insurance industry than traditional machine learning models. In this study, transfer learning was employed to build and optimize a simulated automobile damage assessment system. Several classic deep learning methods were applied to extract features from original and augmented automobile damage images. Then, traditional machine learning and cross-validation techniques were applied to train and validate the system. The proposed deep learning model demonstrated advantages over traditional machine learning models regarding features extraction and accuracy. Deep learning approaches fused with logistic regression and support vector machine were found performing as well as those with artificial neural networks under two simulated scenarios. With the proposed method, automobile damage images can be evaluated for insurance adjustment purposes automatically, based on the acquired input. Hence, insurers can automate the claim and adjustment process, thereby achieving cost and time savings.

BACKGROUND: MicroRNAs are a class of important small noncoding RNAs, which have been reported to be involved in the processes of tumorigenesis and development by targeting a few genes. Existing studies show that the imbalance between cell proliferation and apoptosis is closely related to the initiation and development of cancers. However, the impact of miRNAs on this imbalance has not been studied systematically. RESULTS: In this study, we first construct a cell fate miRNA-gene regulatory network. Then, we propose a systematical method for calculating the global impact of miRNAs on cell fate genes based on the shortest path. Results on breast cancer and liver cancer datasets show that most of the cell fate genes are perturbed by the differentially expressed miRNAs. Most of the top-identified miRNAs are verified in the Human MicroRNA Disease Database (HMDD) and are related to breast and liver cancers. Function analysis shows that the top 20 miRNAs regulate multiple cell fate related function modules and interact tightly based on their functional similarity. Furthermore, more than half of them can promote sensitivity or induce resistance to some anti-cancer drugs. Besides, survival analysis demonstrates that the top-ranked miRNAs are significantly related to the overall survival time in the breast and liver cancers group. CONCLUSION: In sum, this study can help to systematically study the important role of miRNAs on proliferation and apoptosis and thereby uncover the key miRNAs during the process of tumorigenesis. Furthermore, the results of this study will contribute to the development of clinical therapy based miRNAs for cancers.

The biological data science is characterized by a massive amount of data from heterogeneous sources.How to decipher complex relationships among heterogeneous datasets remains an urgent challenge.Although traditional model-driven methods still play an important role in analyzing all kinds of data, it lacks capabilities to exploit the huge amount of available data or even big data to discover knowledge, predict data behaviors, and decipher complex relationships among data.Therefore, data-driven becomes the theme of biological data science for its capabilities in listening to data, interacting with data, and extracting knowledge from data.Modern artificial intelligence will dominate biological data science for its unpreceded learning capabilities to process complex data.Compared to traditional AI techniques (e.g.automated reasoning), machine learning and deep learning are the core to enable machines with intelligence.A deep learning machine has much more complicate learning topologies, which may change dynamically for the sake of learning, besides at least the same complicate-level learning mechanism as traditional machine learning models such as support vector machines.Deep learning is good at discovering latent complex relationships among data and handling big data well.More importantly, deep learning merges feature extraction and prediction (e.g.classification) in a single learning procedure and makes feature extraction more adaptive and compatible with prediction.The scRNAseq data, SNP, interactome or even clinical data usually need very different but complicate feature extrac-

BACKGROUND: Although high-throughput microarray based molecular diagnostic technologies show a great promise in cancer diagnosis, it is still far from a clinical application due to its low and instable sensitivities and specificities in cancer molecular pattern recognition. In fact, high-dimensional and heterogeneous tumor profiles challenge current machine learning methodologies for its small number of samples and large or even huge number of variables (genes). This naturally calls for the use of an effective feature selection in microarray data classification. METHODS: We propose a novel feature selection method: multi-resolution independent component analysis (MICA) for large-scale gene expression data. This method overcomes the weak points of the widely used transform-based feature selection methods such as principal component analysis (PCA), independent component analysis (ICA), and nonnegative matrix factorization (NMF) by avoiding their global feature-selection mechanism. In addition to demonstrating the effectiveness of the multi-resolution independent component analysis in meaningful biomarker discovery, we present a multi-resolution independent component analysis based support vector machines (MICA-SVM) and linear discriminant analysis (MICA-LDA) to attain high-performance classifications in low-dimensional spaces. RESULTS: We have demonstrated the superiority and stability of our algorithms by performing comprehensive experimental comparisons with nine state-of-the-art algorithms on six high-dimensional heterogeneous profiles under cross validations. Our classification algorithms, especially, MICA-SVM, not only accomplish clinical or near-clinical level sensitivities and specificities, but also show strong performance stability over its peers in classification. Software that implements the major algorithm and data sets on which this paper focuses are freely available at https://sites.google.com/site/heyaumapbc2011/. CONCLUSIONS: This work suggests a new direction to accelerate microarray technologies into a clinical routine through building a high-performance classifier to attain clinical-level sensitivities and specificities by treating an input profile as a 'profile-biomarker'. The multi-resolution data analysis based redundant global feature suppressing and effective local feature extraction also have a positive impact on large scale 'omics' data mining.

BACKGROUND: As a novel cancer diagnostic paradigm, mass spectroscopic serum proteomic pattern diagnostics was reported superior to the conventional serologic cancer biomarkers. However, its clinical use is not fully validated yet. An important factor to prevent this young technology to become a mainstream cancer diagnostic paradigm is that robustly identifying cancer molecular patterns from high-dimensional protein expression data is still a challenge in machine learning and oncology research. As a well-established dimension reduction technique, PCA is widely integrated in pattern recognition analysis to discover cancer molecular patterns. However, its global feature selection mechanism prevents it from capturing local features. This may lead to difficulty in achieving high-performance proteomic pattern discovery, because only features interpreting global data behavior are used to train a learning machine. METHODS: In this study, we develop a nonnegative principal component analysis algorithm and present a nonnegative principal component analysis based support vector machine algorithm with sparse coding to conduct a high-performance proteomic pattern classification. Moreover, we also propose a nonnegative principal component analysis based filter-wrapper biomarker capturing algorithm for mass spectral serum profiles. RESULTS: We demonstrate the superiority of the proposed algorithm by comparison with six peer algorithms on four benchmark datasets. Moreover, we illustrate that nonnegative principal component analysis can be effectively used to capture meaningful biomarkers. CONCLUSION: Our analysis suggests that nonnegative principal component analysis effectively conduct local feature selection for mass spectral profiles and contribute to improving sensitivities and specificities in the following classification, and meaningful biomarker discovery.