搜索结果：Movies

Recommender systems are crucial for customizing user experiences across diverse sectors such as e-commerce and entertainment. Traditional correlation methods have been employed to forecast user preferences; however, they frequently prove inadequate when addressing intricate, high-dimensional datasets. Quantum computing presents a new approach for enhancing correlation computations, potentially resulting in more precise recommendations. This study seeks to address and compare the efficiency of classical and quantum correlation techniques in recommender systems utilizing four distinct datasets: Supermarket Sales, IMDB Top 250 movies, MovieLens 10k, and BigBasket products. The Item Recommendation and Quantum Correlation (IRQC) method, makes use of parameterized quantum circuits with rotation gates and entanglement. The experimental methodology comprised the utilization of both classical and quantum correlation approaches, evaluating their efficacy through critical metrics including mean absolute error (MAE) and root mean squared error (RMSE). The results demonstrated that quantum correlations consistently surpassed classical correlations across all datasets. The proposed Quantum Correlation approach obtains lower mean absolute errors of 0.99, 0.30, 0.90, and 0.92 in BigBasket, Supermarket Sales, IMDB Top 250 Movies, and MovieLens 10K datasets, respectively, than 1.20, 1.48, 1.10, and 1.00 with the classical methods. This study underlines the potential of quantum computing in machine learning applications, notably for boosting recommendation systems.

Non-invasive imaging-based assessment of blood flow plays a critical role in evaluating heart function and structure. Computed Tomography (CT) is a widely-used imaging modality that can robustly evaluate cardiovascular anatomy and function, but direct methods to estimate blood flow velocity from movies of contrast dynamics have not been developed. This study evaluates the impact of CT imaging parameters on Physics-Informed Neural Networks (PINN)-based flow estimation and proposes an improved PINN-based approach, SinoFlow, which uses sinogram data directly to estimate blood flow. We generated pulsatile flow fields in an idealized 2D vessel bifurcation using computational fluid dynamics and simulated CT scans with varying gantry rotation speeds, photon flux, and pulse mode imaging settings. We compared the performance of PINN-based flow estimation using reconstructed images (ImageFlow) to SinoFlow. SinoFlow significantly improved flow estimation performance by avoiding temporal inconsistency errors introduced by filtered backprojection. SinoFlow was robust across all tested gantry rotation speeds and consistently produced lower mean squared error and velocity errors than ImageFlow. Additionally, SinoFlow was less susceptible to noise in the sinogram and was compatible with pulsed-mode imaging and maintained higher accuracy with shorter pulse widths. This study demonstrates the potential of SinoFlow for CT-based flow estimation, providing a more promising approach for non-invasive blood flow assessment. The findings inform future applications of PINNs to CT images and provide an alternative which avoids limitations associated with image-based estimation.

Women's midlife is a distinct developmental stage marked by biological, psychological, and sociocultural transitions that may uniquely shape disordered eating risk. While sociocultural pressures from media sources (e.g., advertisements, movies, the Internet) to conform to appearance ideals have been linked to disordered eating attitudes primarily in younger women, less is known about these associations in middle-aged women. The purpose of the current study was twofold: first, to examine whether media pressure was associated with disordered eating attitudes in middle-aged women; second, to examine whether perceived stress or body appreciation moderated this relation. The sample consisted of 347 women ages 40-63 (Mage = 50.13, SD = 4.55) who were mothers of college women in the United States. Media pressure was positively correlated with disordered eating attitudes, measured by the Eating Attitudes Test-26 (r = .16, p = .002). The interaction between media pressure and perceived stress was significant (B = 0.08, p = .025, ΔR2 = .01); the association between media pressure and disordered eating attitudes was significant at high but not low perceived stress. The interaction between media pressure and body appreciation was also significant (B = -1.51, p = .002, ΔR2 = .03); the association was significant at low but not high body appreciation. These findings reveal that higher media pressure is associated with disordered eating attitudes in middle-aged women, especially in women with higher stress and women with lower body appreciation, suggesting that media pressure, perceived stress, and body appreciation may be useful intervention targets to reduce disordered eating attitudes in middle-aged women.

Major gaps limit the comprehensive understanding of the associations between screen time and cognitive development in preschool years. The objectives of this study were to examine (1) screen time patterns and (2) the associations between screen time patterns and cognitive development in preschool children. This cross-sectional study used baseline data, collected between October 2022 and December 2023 and analyzed in 2025, from the TECHnology and Development in Early Childhood project. Participants were 359 preschool children (aged 3, 3.5, or 4 years ± 2 weeks) and their parents from Western Canada. Screen time patterns (type: show/movie/video, electronic game, communication; content: parent-defined educational or entertainment, researcher-defined educational or entertainment; device: TV, mobile [tablet/smartphone]; context: co-use with an adult, evening) were parent reported using a 2-week online daily diary. Cognitive development (language, response inhibition, working memory, self-control) was assessed through 4 established short games administered by researchers during a recorded virtual meeting. Regression models adjusted for covariates were conducted. Based on 4,774 days of data, children had an average of 77 min/day of screen time. The most prevalent type and content of screen time across devices were show/movie/video (69 min/day) and entertainment (parent defined; 55 min/day) screen time, respectively. Higher levels of show/movie/video, mobile device, and parent-defined entertainment screen time (10 min/day) were consistently associated with less advanced executive function scores (response inhibition: B= -0.45 to -0.62; working memory: B= -0.02 to -0.03; self-control: B= -0.86, -1.05). For language, engaging in versus not engaging in communication screen time (B=1.71; p=0.037) and co-use with an adult (B=1.88; p=0.039) was associated with more advanced scores, only when nontypical and typical days were included. Entertainment shows/movies/videos were common among preschool children. Some patterns of screen time appeared to be more detrimental to cognitive development than others. Findings can help inform more nuanced screen time recommendations and interventions that aim to minimize harm and maximize the benefits of this societal reality.

In this paper, we chart an emerging academic terrain: cultural evolution of the arts, which is a theory-driven exploration of artistic dynamics, often done with large datasets of music, literature, movies, paintings, or games. This field has grown at the intersection of cultural evolution theory and several academic fields: computational humanities, anthropology, network science, and others, and poses interesting challenges for each of them. What constitutes artistic transmission in the first place? Is it possible to find recurring patterns in artistic history - and how much data is needed for that? What makes the evolution of the arts different from the evolution of other forms of knowledge? We discuss all these problems in this paper. Additionally, we perform a bibliometric analysis of this field and explore a co-citation network of the works on artistic evolution. Finally, we highlight major challenges for this field in the future, as the arts are rapidly evolving in the digital age.

High-speed atomic force microscopy (HS-AFM) enables direct visualization of protein dynamics under near-physiological conditions, yet its intrinsic limitation to surface topography prevents atomic-level structural characterization. We present AFM-Fold, a generative AI-based framework that reconstructs three-dimensional protein conformations directly from AFM images. AFM-Fold combines a group-equivariant convolutional neural network, which extracts low-dimensional collective variables (CVs) from AFM images, with a guided diffusion process that generates conformations consistent with the inferred CVs. Using pseudo-AFM images of Adenylate kinase, AFM-Fold accurately reproduced not only the open and closed conformations, but also a continuous range of intermediate conformations spanning the open-closed transition. Application to 159 experimental HS-AFM frames of the flagellar protein FlhAC further demonstrated that AFM-Fold yields conformations more consistent with experimental images than rigid-body fitting of the crystal structure, and captures time-correlated domain motions that reflect underlying conformational dynamics. AFM-Fold enables rapid, physically plausible structure estimation from individual AFM images, typically within one minute per frame, without relying on molecular dynamics simulations. This unified and computationally efficient pipeline opens a route to high-throughput structural analysis of HS-AFM movies.

People can experience the same event yet form distinct memories shaped by individual interpretations. Prior research shows that multivariate activity patterns in the Default Mode Network (DMN) are correlated across individuals during shared experiences, suggesting a role in representing high-level event features. However, it remains unclear whether these shared neural patterns reflect similarity in subsequent memory content. Here, we examined whether memory similarity correlates with intersubject spatial patterns in the DMN using a pre-existing dataset. Twenty-four individuals watched and recounted two cartoon movies during fMRI scanning. Using topic modeling, we transformed verbal recall into vectors of latent topics to quantify memory similarity across participants. We found that greater similarity in recalled content was associated with stronger shared activation patterns at encoding and retrieval, particularly in the posterior medial, medial prefrontal and anterior temporal cortices. These findings highlight the utility of natural language processing tools in linking memory representations to brain activity and underscore the DMN's role in encoding, interpreting, and recalling complex event features.

Alexithymia has been associated with deficits in social cognition, although findings are inconsistent and often limited by methodological constraints. This study aimed to clarify this relationship using ecologically valid and traditional standardized measures across multiple social-cognitive domains. A total of 163 adults from the general population completed a series of measures, including the Toronto Alexithymia Scale (TAS-20), Questionnaire of Cognitive and Affective Empathy (QCAE), Reading the Mind in the Eyes Test (RMET), Movies for the Assessment of Social Cognition (MASC), and Amsterdam Dynamic Facial Expression Set-Bath Intensity Variations (ADFES-BIV). Results of hierarchical regression analyses revealed that alexithymia facets significantly predicted performance on affective and cognitive empathy (QCAE), and Theory of Mind (MASC total and "No ToM" scores). The only exceptions were affective Theory of Mind (RMET) and recognition of others' emotions (ADFES-BIV), for which none of the alexithymia facets emerged as significant predictors. The findings suggest that alexithymia is associated with poorer performance in cognitive and affective empathy and contextual Theory of Mind, whereas no significant association emerged for emotion recognition. The results suggest that integrating dynamic and context-rich tasks may be useful for detecting subtle social-cognitive difficulties in individuals with alexithymic traits.

Sociability is central for humans. Visual information ranging from low-level physical features (e.g., luminance) to mid-level semantic information (e.g., face recognition) and high-level social inference (e.g., emotional valence of social interactions) is constantly sampled for navigating the social world. In this study, we utilized large-scale eye tracking during natural vision for mapping how different levels of visual information guide the perception of socially relevant features (social vision) simultaneously. In three experiments, participants (N = 166) watched full-length films and short movie clips with varying social content (total duration: 193 minutes) during eye tracking. To model the association between perceptual features and spatiotemporal gaze parameters (gaze position, gaze synchronization, pupil size and blinking), we extracted 39 stimulus features from the movies, including low-level audiovisual features (e.g., luminance, motion), presence and location of mid-level semantic categories (e.g., faces, objects), and high-level social information (e.g., body movements, pleasantness). Integrative analysis techniques with cross-validation were developed to simultaneously associate the perceptual features with the gaze behavior. Pupil size was modulated by luminance, scene cuts, and emotional arousal while gaze position was most accurately predicted by a combination of the presence of human faces, local motion, and entropy. Faces and eyes were prioritized over other semantic categories, and blinking rate decreased during periods of attentional engagement. Altogether, the results show that human social vision is primarily guided by low-level physical features and mid-level semantic categories, while high-level social features such as emotional arousal primarily modulate pupillary responses.

Visual assessment of membrane motion is essential for managing EXCOR VAD, but accuracy depends on observer experience. We evaluated a high-speed image AI model to support healthcare providers. Patients on EXCOR Pediatric admitted to the University of Tokyo Hospital (May 2022-May 2024) were included. Membrane images were obtained from patients and a manually filled pump at bench. An image recognition model was trained to estimate membrane position. Experienced physicians (N = 11) and inexperienced physicians (N = 11) assessed pump status in a sample dataset (N = 12) with and without AI assistance. A total of 142 movies from five patients were collected (98 training, 45 validation), plus 1100 bench images for training. Model accuracy was 0.91, with AUROCs of 0.99 ("fill") and 0.96 ("empty"). Among experienced physicians, accuracy significantly improved with AI assistance from 0.83 (0.67-0.88) to 0.92 (0.92-1.0) (median (IQR); p = 0.016). Among inexperienced physicians, accuracy also significantly improved with AI assistance from 0.67 (0.5-0.75) to 0.83 (0.75-0.92) (p = 0.049). A high-speed image AI can facilitate the visual assessment of EXCOR VAD by healthcare providers.

Existing movie scene segmentation methods struggle with long-range temporal dependencies, attention degradation, and rigid multimodal fusion, often ignoring the inherent hierarchical structure of movies. To address these limitations, this paper proposes a Hierarchical Attentive Transformer (HATrans). First, to capture multi-granularity semantic consistency and discriminative scene boundaries, we adopt a shot-to-segment hierarchical encoding pipeline that explicitly models structural priors. Second, to mitigate attention degradation when processing extremely long shot sequences, we introduce a hierarchical masked attention mechanism combined with a temporal position-aware bias, which restricts irrelevant connections and enhances structural sensitivity. Third, to overcome the inflexibility of conventional multimodal integration, we propose a label-guided attention fusion module that leverages semantic category priors to dynamically weight visual, audio, and subtitle features based on varying semantic contexts. Experimental results on MovieNet-42 K suggest that HATrans achieves competitive performance, outperforming several baselines including CMTS.

Objective: Physicians inevitably face illness; yet, occupying the role of patient poses distinct psychological and professional challenges. To elucidate the unique challenges and strengths physicians may experience as patients, this study examines how popular media portrays physicians, highlighting common themes and their implications for clinical practice and medical education. Methods: Literature was reviewed on the unique experiences of physicians in the role of patient. Searches were conducted on Google and ChatGPT using the terms movies + physician as patients, television + physician as patients, and popular media + physician as patients. Additional examples were drawn from the authors' media knowledge base. Retrieved results were reviewed for depictions that exemplify the challenges and strengths unique to physician patients. Scenes from widely recognized films and television shows were discussed to illustrate key themes. Results: Six recurring themes emerged: (1) shame and loss of professional identity, (2) interference in one's own care through self-diagnosis, (3) fear of burdening colleagues, (4) difficulty relinquishing control, (5) curbside consultations with blurred boundaries, and (6) health literacy as a strength. These narratives reflect the physician health literature and offer resonant vignettes of the tensions physician patients face. Conclusion: Physicians who become patients balance vulnerability with unique strengths, yet their professional identity often complicates care. Clinicians should anticipate these dynamics, set clear boundaries, and normalize help-seeking to ensure safe and dignified treatment. Media-based narratives can serve as powerful teaching tools, fostering empathy and preparing clinicians to navigate the complexities of caring for or assuming the role of physician patients. Prim Care Companion CNS Disord 2026;28(3):26m04196. Author affiliations are listed at the end of this article.

Tracking and lineage tracing are widely needed tasks in biological image analysis. For cells that grow and divide, tracking is challenging because cells change in number, shape, and size throughout a recording. Longer intervals between images make tracking more difficult. Consequently, tracking has to be performed between consecutive or temporally close images, which leads to exponentially decreasing tracking accuracy and high sensitivity to error rates. For budding yeast, this challenge is further heightened by the similarity of cells in colonies, their dense packing, asymmetric cell divisions, and movement due to colony growth. A related task, lineage tracing, is similarly challenging without fluorescent markers since a new daughter cell can be surrounded by multiple potential mother cells. Here, we present neural networks for budding yeast tracking and lineage tracing, named LYN-track and LYN-trace, respectively, which leverage fine geometric features of cells and their neighborhoods. To train and test the algorithms, we recorded and annotated budding and fission yeast timelapse microscopy movies (78 852 frame-to-frame tracklets, 2512 images), which we make available. On these and existing datasets, our neural network-based methods demonstrate robust, above state-of-the-art performance. Both tools are integrated into graphical user interfaces (GUIs) and can be retrained with custom data.

Most brain-behavior mapping studies rely on resting-state functional connectivity (FC), but this approach has known accuracy limits and can be outperformed by movie-watching FC. Here, we present a novel deep neural network framework to predict cognitive scores and sex from FC during naturalistic movie viewing, and examine how movie content and its ability to synchronize brain activity across individuals relate to prediction performance. We show that FC from movie-watching generally outperforms resting-state FC - even when compared to five times more temporal data - with sensory and higher-order brain networks emerging as the most important for prediction. Using both static and sliding-window dynamic FC approaches, we find that higher cognitive prediction accuracy is positively associated with greater inter-subject synchrony and the duration of human faces and voices in the movies; these effects were not found for sex prediction. This work underscores the promise of naturalistic movie viewing as a powerful tool for probing individual differences in the brain and revealing neural underpinnings of human behavior.

MitoGraph is a widely used tool for the automated segmentation of mitochondrial networks in three-dimensional (3D) fluorescence microscopy. However, the emergence of advanced live-cell microscopes such as lattice light-sheet microscopy (LLSM) has produced massive four-dimensional (4D, 3D+time) datasets that highlight a critical bottleneck: current CPU-based implementations are computationally prohibitive, often requiring days or weeks to process. To address this limitation, we developed MitoGraph-GPU, a Python-based GPU implementation that accelerates the dominant filtering steps by vectorizing Hessian/eigenvalue and vesselness computations using CuPy, and streamlines network processing with faster skeletonization and topology analysis. Tested across budding yeast and human lung organoid datasets, MitoGraph-GPU achieves up to 11× speedup in yeast cells and 30× speedup in per-frame segmentation of lung cells. Segmentation fidelity is preserved, with  ~99.9% agreement in maximum intensity projections of segmented images, and minimal differences in downstream measurements. Critically, this throughput enables practical analysis of large 4D datasets : in an LLSM organoid use case (10 movies, 60 frames, ~ 50 cells per movie), total processing time decreases from ~ 500 h on CPU to ~ 20 h on GPU (25× faster). By producing accurate mitochondrial surfaces and skeletons, MitoGraph-GPU can serve as an efficient segmentation module for downstream mitochondrial tracking and analyses, enabling scalable high-throughput 4D mitochondrial phenotyping.

Comprehensive wiring diagrams from electron microscopy (EM) are a powerful tool to understand the inner workings of the brain. The retina is an easily accessible part of the brain that performs complex visual computations. Its thin, layered structure offers a unique opportunity to decipher neural cell types and map their connectivity. A major obstacle has been the limited size of existing retinal EM datasets, which could not resolve rare cell types and neurons with large dendritic arbors. Here, we describe Eyewire II, a large-scale EM dataset covering nearly 1 mm 2 of the adult mouse retina - roughly 10-100 times larger than previous retinal EM volumes. Human proofreading of an automated reconstruction has so far yielded more than 8, 000 bipolar cells, 13, 000 amacrine cells, and 4, 000 retinal ganglion cells. Automated detection is complete for synaptic ribbons and in progress for conventional synapses. Prior to EM imaging, visual responses to diverse stimuli - including natural movies - were recorded in a subset of neurons using two-photon Ca 2+ imaging, enabling direct alignment of morphological and functional cell type identity. To enable high-throughput, automatic cell typing, we devised a human-in-the-loop approach that combines deep learning with human expert annotations. As a proof-of-principle, we show that morphological features of reconstructed bipolar cells are sufficient to recover all 15 known bipolar cell types with regular, non-overlapping mosaics. Together, these data and tools establish Eyewire II as a shared resource for the field of retina research. Already now, more than 30 laboratories worldwide are contributing proofreading, expert annotations, and software tools, advancing Eyewire II towards a complete cell type catalog and synaptic wiring diagram of a mammalian retina.

High-speed atomic force microscopy (HS-AFM) movies have contributed significantly to our understanding of biomolecular dynamics, because these experiments have resolved conformational changes at submolecular resolution in aqueous environments. However, the resolution of HS-AFM images and AFM images in general critically depend on the sharpness of the tip. Focused electron beam-induced deposition (FEBID) in a scanning electron microscope (SEM) yields AFM tips with high aspect ratio and small tip radius, formed by deposition of metal-organic precursors such as ferrocene, producing amorphous carbon deposits, permissive for the reproducible acquisition of high-resolution images. Although FEBID tips are essential for HS-AFM image quality, it is common to rely on commercial cantilevers, because of limited SEM access or perceived technical barriers. Here, we showcase the use of a commercial benchtop SEM for the fabrication of FEBID tips for high-resolution AFM. To facilitate the fabrication process, we developed a cantilever and sample holder that minimize drift and mechanical vibrations during SEM operation and enable efficient and reproducible fabrication of sharp AFM tips. The shape of the FEBID tips was evaluated as a function of beam defocus, deposition time, acceleration voltage and beam current. We describe an optimized workflow for FEBID tip fabrication (eight tips can be fabricated within 1-2 h) and demonstrate high-resolution and long-term stable HS-AFM imaging of annexin V and plasmid DNA using the FEBID tips fabricated in this study. We anticipate that this protocol may be useful for a wide range of AFM practitioners and facilitate the reproducible acquisition of high-resolution AFM data.