Patterns最新文献

Describing gene interactions in cells is challenging due to their complexity and the limited microscopic detail available. Boolean networks offer a powerful, coarse-grained approach to modeling these dynamics using binary agents and their interactions. In this context, attractors-stable states of the system-are associated with biological phenotypes, making their identification biologically important. However, traditional computing struggles with the exponential growth of the state space in such models. Here, we present a novel quantum search algorithm for identifying attractors in synchronous Boolean networks, specifically designed for use on quantum computers. The algorithm iteratively suppresses known attractor basins, increasing the probability of detecting new ones. Unlike classical methods, it guarantees the discovery of a new attractor in each run. Early tests demonstrate strong resilience to noise on current NISQ (noisy intermediate-scale quantum) devices, marking a promising advance toward practical quantum-enhanced biological modeling.

Traditionally, to improve the segmentation performance of models, most approaches prefer to use more complex modules. This is not suitable for the medical field, especially for mobile medical devices, where computationally loaded models are not suitable for real clinical environments due to computational resource constraints. Recently, state-space models, represented by Mamba, have become a strong competitor to traditional convolutional neural networks and transformers. In this paper, we deeply explore the key elements of parameter influence in Mamba and propose an UltraLight Vision Mamba UNet (UltraLight VM-UNet) based on this. Specifically, we propose a method for processing features in parallel Vision Mamba, named the PVM Layer, which achieves competitive performance with the lowest computational complexity while keeping the overall number of processing channels constant. We conducted segmentation experiments on three public datasets of skin lesions and showed that UltraLight VM-UNet exhibits competitive performance with only 0.049M parameters and 0.060 GFLOPs.

Blood glucose (BG) in hospitalized patients is influenced by numerous clinical factors, including medications not traditionally associated with glycemic control. To better characterize these effects, we analyzed electronic health record data from 97,281 inpatient encounters (2014-2022), capturing 3,009,686 point-of-care BG measurements. We extracted over 300 variables-medications, labs, and socio-demographics-and used Lasso, ridge, and elastic net regression for predictive modeling, alongside propensity score matching (PSM) for causal inference. While Lasso reduced multicollinearity, it often assigned implausible coefficient directions. In contrast, PSM yielded clinically consistent and interpretable estimates, identifying 55 variables significantly associated with BG changes, without shrinking coefficients to zero of known BG-modulating drugs. Findings were validated in a 2022-2024 test set of 27,847 encounters. This work highlights the value of causal inference in observational EHR analysis and identifies both established and under-recognized (e.g., cholecalciferol) medication effects on BG, offering insights that inform safer inpatient glycemic management.

This article presents a holistic research agenda to address the significant environmental impact of information and communication technology (ICT), which accounts for 2.1%-3.9% of global greenhouse gas emissions. It proposes several research thrusts to achieve sustainable computing: accurate carbon accounting models, life cycle design strategies for hardware, efficient use of renewable energy, and integrated design and management strategies for next-generation hardware and software systems. If successful, the research would flatten and reverse growth trajectories for computing power and carbon, especially for rapidly growing applications like artificial intelligence. The research takes a holistic approach because strategies that reduce operational carbon may increase embodied carbon, and vice versa. Achieving these goals will require interdisciplinary collaboration between computer scientists, electrical engineers, environmental scientists, and economists.

Cytometry data, including flow and mass cytometry, are widely used in immunological studies such as cancer immunotherapy and vaccine trials. These data provide rich insights into immune cell dynamics and their relationship to clinical outcomes. However, traditional analyses based on summary statistics may overlook critical single-cell information. To address this, we introduce cytoGPNet, a novel method for predicting individual-level outcomes from cytometry data. cytoGPNet addresses four key challenges: (1) accommodating varying numbers of cells per sample, (2) analyzing longitudinal cytometry data to capture temporal patterns, (3) maintaining robustness despite limited sample sizes, and (4) ensuring interpretability for biomarker discovery. We apply cytoGPNet across multiple immunological studies with diverse designs and show that it consistently outperforms existing methods in predictive accuracy. Importantly, cytoGPNet also offers interpretable insights at multiple levels, enhancing our understanding of immune responses. These results highlight cytoGPNet's potential to advance cytometry-based analysis in immunological research.

Radiographic images play a critical role in disease diagnosis, but accurately interpreting them requires considerable expertise and workload. Recent research has advanced artificial intelligence-based medical image analysis, but such advancements remain limited in real clinical practice where multistage diagnosis is required for fine-grained diseases. This study proposes a screening-to-subtyping (S2S) AI paradigm specifically designed for accurate radiological diagnosis of fine-grained diseases, encompassing the entire diagnostic process from initial screening to final subtyping. The S2S framework integrates information from multiple diagnostic phases, radiological viewpoints, lesion dimensions, and imaging modalities to address complex diagnostic challenges. Evaluation using a large-scale, multi-center radiography dataset of fine-grained thoracic cancer subtypes demonstrates the system's robust performance. Furthermore, this investigation offers novel insights into human-AI collaboration for diagnosing intricate fine-grained pathologies. Our results highlight the substantial clinical potential of S2S AI across varied healthcare environments and disease entities, facilitating deeper integration of artificial intelligence in radiological diagnostics.

There are intensive debates about whether heterogeneous networks promote prosocial behaviors such as fairness and cooperation. Theoretical models predict that network heterogeneity plays a positive role, but this prediction has not been validated by experiments. We reconcile this debate by conducting experiments with two-stage ultimatum games on networks. In the first stage, we identify responders with strong fairness preferences, referred to as leaders. In the second stage, when leaders occupy high-degree nodes in a heterogeneous network, their ability to motivate fairness among neighboring proposers is amplified, and collective fairness is facilitated. We propose an evolutionary game model and an agent-based simulation framework that capture the microscopic mechanisms underlying the networked experiments. Our experiments, model, and simulations suggest that network reciprocity is achievable but requires coordinated interactions between different prosocial inclinations of individuals and social network structures.

Interpretability is essential for building trust in health artificial intelligence (AI), but ensuring trustworthiness requires addressing broader ethical concerns, such as fairness, privacy, and reliability. This opinion article discusses the multilayered role of interpretability and transparency in addressing these concerns by highlighting their fundamental contribution to the responsible adoption and regulation of health AI.

Existing protein machine learning representations typically model either the sequence or structure distribution, with the other modality implicit. Here, we characterize an embedding of the joint distribution of protein sequence and structure by compressing the latent space of the protein folding model ESMFold. This provides mechanistic interpretability insights, as well as a flexible compressed representation. We term these CHEAP (compressed hourglass embedding adaptations of proteins) embeddings. In continuous compression schemes, the ESMFold latent space can be reduced by factors of 128 $\times$ along the channel and 8 $\times$ along the length while retaining structure information at <2 Å scale accuracy and performing competitively on protein function and localization benchmarks. In discrete compression schemes, we construct a tokenized all-atom structure vocabulary that retains high reconstruction accuracy, thus introducing a tokenized representation of an all-atom structure that can be obtained from the sequence alone. CHEAP democratizes representations captured by large models and can enable flexible downstream applications such as generation, search, and prediction.

Women have been instrumental in shaping data science from its earliest days. This opinion highlights both the achievements and the ongoing challenges faced by women in the field, emphasizing that a wide range of perspectives and backgrounds among data scientists is essential to drive innovation and improve research quality.