bioRxiv [Preprint]. 2025 Aug 2:2025.07.31.668042. doi: 10.1101/2025.07.31.668042.

ABSTRACT

Amidst the fast-developing trend of RNA large language models with millions of parameters, we asked what would be the minimum required to rediscover the rules of RNA canonical base pairing, mainly the Watson-Crick-Franklin A:U, G:C and the wobble G:U base pairs (the secondary structure). Here, we conclude that it does not require much at all. It does not require knowing secondary structures; it does not require aligning the sequences; and it does not require many parameters. We selected a probabilistic model of palindromes (a stochastic context-free grammar or SCFG) with a total of just 21 parameters. Using standard deep learning techniques, we estimate its parameters by implementing the generative process in an automatic differentiation (autodiff) framework and applying stochastic gradient descent (SGD). We define and minimize a loss function that does not use any structural or alignment information. Trained on as few as fifty RNA sequences, the rules of RNA base pairing emerge after only a few iterations of SGD. Crucially, the sole inputs are RNA sequences. When optimizing for sequences corresponding to structured RNAs, SGD also yields the rules of RNA base-pair aggregation into helices. Trained on shuffled sequences, the system optimizes by avoiding base pairing altogether. Trained on messenger RNAs, it reveals interactions that are different from those of structural RNAs, and specific to each mRNA. Our results show that the emergence of canonical base-pairing can be attributed to sequence-level signals that are robust and detectable even without labeled structures or alignments, and with very few parameters. Autodiff algorithms for probabilistic models, such as, but not restricted to SCFGs, have significant potential as they allow these models to be incorporated into end- to-end RNA deep learning methods for discerning transcripts of different functionalities.

PMID:40766544 | PMC:PMC12324431 | DOI:10.1101/2025.07.31.668042

bioRxiv [Preprint]. 2025 Aug 3:2025.08.03.668353. doi: 10.1101/2025.08.03.668353.

ABSTRACT

Mapping protein sequence-function landscapes has either been limited to small steps (only few mutations) or to sequences similar to those already explored by evolution to maintain activity. Here, we overcome both limitations by applying deep-learning guided redesign to a natural protein tyrosine kinase to generate novel, functional sequences with highly combinatorial mutations. Using cell-free assays, we measure the activities and concentrations of 537 redesigned sequences, which differ from the wild-type by an average of 37 mutations while retaining activity in 85% of variants. These sequences sample 436 unique mutations at 76 different positions throughout the kinase domain. A simple regression model identifies key sequence determinants of function and predicts the function of unseen sequences. Our approach demonstrates how integrating deep-learning guided redesign, functional measurement at scale, and interpretable computational modelling enables functional exploration of highly combinatorial and sparse sequence-function landscapes at mutational scales not possible before.

PMID:40766444 | PMC:PMC12324526 | DOI:10.1101/2025.08.03.668353

Front Oncol. 2025 Jul 22;15:1630291. doi: 10.3389/fonc.2025.1630291. eCollection 2025.

ABSTRACT

INTRODUCTION: Schizophrenia is a severe psychological disorder that significantly impacts an individual's life and is characterized by abnormalities in perception, behavior, and cognition. Conventional Schizophrenia diagnosis techniques are time- consuming and prone to error. The study proposes a novel automated technique for diagnosing Schizophrenia based on electroencephalogram (EEG) sensor data, aiming to enhance interpretability and prediction performance.

METHODS: This research utilizes Deep Learning (DL) models, including the Deep Neural Network (DNN), Bi-Directional Long Short-Term Memory-Gated Recurrent Unit (BiLSTM- GRU), and BiLSTM with Attention, for the detection of Schizophrenia based on EEG data. During preprocessing, SMOTE is applied to address the class imbalance. Important EEG characteristics that influence model decisions are highlighted by the interpretable BiLSTM-Attention model using attention weights in conjunction with SHAP and LIME explainability tools. In addition to fine-tuning input dimensionality, F-test feature selection increases learning efficiency.

RESULTS: Through the integration of feature importance analysis and conventional performance measures, this study presents valuable insights into the discriminative neurophysiological patterns associated with Schizophrenia, advancing both diagnostic and neuroscientific expertise. The experiment's findings show that the BiLSTM with attention mechanism model provides and accuracy of 0.68%.

DISCUSSION: The results show that the recommended approach is useful for Schizophrenia diagnosis.

PMID:40766336 | PMC:PMC12324168 | DOI:10.3389/fonc.2025.1630291

Res Sq [Preprint]. 2025 Jul 28:rs.3.rs-7108570. doi: 10.21203/rs.3.rs-7108570/v1.

ABSTRACT

High-throughput single-cell and spatial omics technologies have transformed biological research. Despite these advances, reliably identifying the molecular drivers and their interplays across biological levels and scales remains a significant challenge. Current experimental methods are limited by batch effects, the lack of simultaneous multi-modal measurements in individual cells, limited coverage of measured proteins, poor generalization to unseen conditions, and insufficient spatial context at a single-cell resolution. To overcome these challenges, we introduce scProSpatial, a unified, multi-modal, multi-scale deep learning framework designed to infer and impute high fidelity single-cell spatial proteomics from scRNA-seqs. Through comprehensive evaluations, scProSpatial accurately predicts spatial abundances of proteins in the absence of shared transcriptomics features, expands protein coverages by 50 times, and generalizes robustly to out-of-distribution scenarios. A case study in metastatic breast cancer further illustrates its utility, demonstrating scProSpatial's potential to drive cross-level, cross-scale multi-omics integration and analysis and reveal deeper insights into complex biological systems.

PMID:40766228 | PMC:PMC12324605 | DOI:10.21203/rs.3.rs-7108570/v1

Data Brief. 2025 Jul 20;62:111909. doi: 10.1016/j.dib.2025.111909. eCollection 2025 Oct.

ABSTRACT

This paper presents a measured spectrum strength dataset in the urban scenario with multiple radiation sources, aiming to address the limitation of open datasets for spectrum environment map (SEM) in realistic multi-source dynamic scenarios. The dataset was collected through high-precision measurements, covering the 30 MHz, 115 MHz, and 2 GHz frequency bands, with a spatial resolution of 1m×1 m. It includes spectrum strength or received signal strength (RSS) data in dBm for 80×105 grids. Each grid includes the information such as longitude, latitude, altitude, and time. The experiment utilizes three radiation sources with isotropic antennas and a mobile signal receiving system equipped with a spectrum analyzer and a GPS module. It collects data along a pre-defined path at a constant speed. The key feature of this dataset is its realistic representation of nonline ar characteristics of propagation channel in a multi-radiation source coexistence scenario. Its applications include the verification of spectrum map completion algorithms, wireless channel modelling, deep learning-driven signal prediction, and the optimization of Wi-Fi/cellular networks.

PMID:40766197 | PMC:PMC12319677 | DOI:10.1016/j.dib.2025.111909

Stem Cell Res Ther. 2025 Aug 5;16(1):426. doi: 10.1186/s13287-025-04507-y.

ABSTRACT

BACKGROUND: Extracellular Vesicles (EVs) derived from mesenchymal stem cells (MSCs) have gained recognition as promising therapeutic and drug delivery agents in regenerative medicine. However, their clinical application is limited by donor variability, low scalability, and inconsistent therapeutic quality. To overcome these challenges, a robust and standardized production platform is urgently needed.

METHODS: We developed a scalable biomanufacturing strategy by generating and expanding MSCs from extended pluripotent stem cells (EPSC) using a suspension bioreactor culture system. A fixed-bed bioreactor was integrated for automated, continuous expansion of iMSCs and downstream EV harvesting. EVs were isolated through a streamlined protocol and characterized for size, morphology, surface markers, and bioactivity. Therapeutic efficacy was assessed in a bleomycin-induced pulmonary fibrosis mouse model.

RESULTS: iMSC-derived EVs (iMSC-EVs) exhibited comparable characteristics to primary MSC-EVs, including a size distribution of 70-80 nm, cup-shaped morphology, and expression of canonical EV markers (CD63, CD81, TSG101). iMSCs were expanded for up to 20 days in 3D culture, yielding > 5 × 10⁸ cells per batch using a suspension bioreactor culture system and producing ~ 1.2 × 10¹³ EV particles/day in a fixed-bed bioreactor. In vivo, iMSC-EVs significantly reduced Ashcroft fibrosis scores and bronchoalveolar lavage fluid protein levels in bleomycin-injured lungs, with therapeutic efficacy comparable to primary MSC-EVs.

CONCLUSIONS: This study establishes a scalable and standardized platform for producing high-quality iMSC-EVs using bioreactor-based systems. Our approach addresses key limitations in traditional EV production and sets the stage for AI-integrated, fully automated, GMP-compliant manufacturing of therapeutic EVs suitable for clinical translation.

PMID:40765003 | DOI:10.1186/s13287-025-04507-y

Front Pharmacol. 2025 Jul 22;16:1606914. doi: 10.3389/fphar.2025.1606914. eCollection 2025.

ABSTRACT

INTRODUCTION: Prostate cancer (PCa) remains a significant global health challenge despite advancements in treatment strategies. There is a need to explore the molecular heterogeneity of PCa to facilitate the development of personalized treatment approaches. This study investigates the molecular heterogeneity of PCa by combining genomic and transcriptomic data using a systems biology approach.

METHODS: By utilising whole-genome sequencing and differentially expressed genes from "The Cancer Genome Atlas Prostate Adenocarcinoma (TCGA-PRAD)" patient samples, we identified 357 intersecting genes. From protein-protein interaction network analysis, 22 hub genes were identified as critical regulators associated with PCa prognosis and pathogenesis. Furthermore, these hub genes were subjected to functional and pathway enrichment analysis via gene ontology (GO) and the Kyoto Encyclopaedia of Genes and Genomes (KEGG).

RESULTS: Notably, the PI3K/Akt signalling pathway was significantly enriched with eight of these hub genes, with significant clinical relevance. Dipeptidyl Peptidase 4 (DPP4) emerged as a promising therapeutic target. Further, in vitro assays were performed to investigate and validate the molecular role of DPP4 through pharmacological inhibition using Linagliptin, a selective DPP4 inhibitor. Inhibition of DPP4 led to the induction of apoptosis, G1/S phase cell cycle arrest, and significant suppression of cell proliferation and migration in PC3 and DU145 cell lines.

DISCUSSION: These experiments revealed novel downstream regulatory effects of DPP4, demonstrating that its inhibition results in the transcriptional downregulation of FGF17, PDGFRA, COL4A1, and COL9A2, thereby contributing to the inactivation of the PI3K/Akt signaling pathway. Collectively, these findings highlight DPP4 as a potential therapeutic target for the treatment of PCa.

PMID:40766751 | PMC:PMC12321844 | DOI:10.3389/fphar.2025.1606914

bioRxiv [Preprint]. 2025 Jul 31:2025.07.29.667408. doi: 10.1101/2025.07.29.667408.

ABSTRACT

Mass spectrometry (MS)-based proteomics remains technically demanding and prohibitively expensive for many large-scale or routine applications, with per-sample costs of hundreds of dollars or more. To democratize access to proteomics and facilitate its integration into more high-throughput multi-omic studies, we present a robust analytical framework for achieving in-depth, quantitative proteome profiling at a cost of approximately $10 per sample, termed the "$10 proteome." Using the Thermo Fisher Orbitrap Astral and Bruker timsTOF Ultra 2 mass spectrometers, we evaluated performance across sample inputs ranging from 200 pg to 100 ng and active gradient lengths from 5 to 60 minutes. Proteome coverage saturated within the low-nanogram input range, with ∼8000 protein groups quantified from as little as 10 ng of input and nearly 6000 protein groups from 200 pg. With already demonstrated low-cost one-pot sample preparation workflows that are appropriate for this sample input range, standardized MS acquisition settings, and high-throughput nanoLC operated at ∼10 min per sample, the $10 proteome becomes feasible. This study establishes a practical, scalable, and cost-effective foundation for global proteome profiling, paving the way for routine, large-scale applications in systems biology, clinical research and beyond.

PMID:40766599 | PMC:PMC12324313 | DOI:10.1101/2025.07.29.667408

Res Sq [Preprint]. 2025 Jul 29:rs.3.rs-7077811. doi: 10.21203/rs.3.rs-7077811/v1.

ABSTRACT

Multiple mechanisms of immunity must be coordinated to defend against a comprehensive range of pathogens; however, the mechanisms by which broad-spectrum antipathogens act remain largely elusive. Here, we employed systems biology approaches to understand the organization of human immune cells at the single-cell level, as well as their reorganization in response to K21, a silane derivative effective against viral, bacterial, and fungal infections. K21 induced pro-inflammatory pathways in M1 and M2c macrophages without altering cytokine secretion, decreased a specific subtype of M1 macrophages and CXCL4-induced M2-like macrophages, and improved mitochondrial health by enhancing mitochondrial recycling via mitophagy. Similar treatment of the in vivo model organism C. elegans induced mitophagy and extended lifespan, suggesting evolutionary conservation of mechanism. Our work demonstrates that a drug that remodels mitochondria and metabolism can shape the immune cell repertoire, which could aid the development of more effective antimicrobials and prevent the emergence of drug-resistant pathogens.

PMID:40766249 | PMC:PMC12324592 | DOI:10.21203/rs.3.rs-7077811/v1

Blood Vessel Thromb Hemost. 2025 May 26;2(3):100077. doi: 10.1016/j.bvth.2025.100077. eCollection 2025 Aug.

ABSTRACT

We developed a platform to measure the oxygen-dependent mechanical properties and oxygen saturation of individual irreversibly sickled cells (ISCs). We identified and measured ISCs from a cohort of 10 individuals with sickle cell disease. ISCs were found to have an average shear surface modulus 20 times that of nonsickled cells and a sixth that of red blood cells (RBCs) with detectable hemoglobin polymer. We found that the number of ISCs was significantly reduced at 53 mm Hg oxygen compared with ≥91 mm Hg oxygen, suggesting that these RBCs can still form polymer under hypoxia. We also found that the fraction of ISCs present in a blood sample had a negative correlation with donor fetal hemoglobin (HbF) fraction, suggesting that HbF could play a role in mitigating occurrence of ISCs.

PMID:40765910 | PMC:PMC12320412 | DOI:10.1016/j.bvth.2025.100077

Front Plant Sci. 2025 Jul 22;16:1624335. doi: 10.3389/fpls.2025.1624335. eCollection 2025.

ABSTRACT

INTRODUCTION: Areca catechu is a widely cultivated palm species with significant economic and medicinal value. However, A. catechu is a tropical plant that is particularly susceptible to low temperatures.

METHODS: This study integrates physiological profiling with transcriptomic sequencing to systematically investigate the cold-response mechanisms of A. catechu.

RESULTS: Multivariate variance analysis revealed that peroxidase (POD) activity and chlorophyll content are significant biomarkers strongly correlated with cold tolerance. A comprehensive investigation into the temporal expression of genes in response to 24 hours of cold stress was conducted, using RNA-seq analysis. This analysis yielded a substantial number of differentially expressed genes (DEGs), amounting to 20,870, which were found to be subject to temporal regulation. KEGG pathway enrichment analysis revealed substantial activation in three metabolic pathways: phytohormone signaling, alkaloid biosynthesis (tropane/piperidine/pyridine), and flavonoid biosynthesis. The application of Weighted Gene Co-expression Network Analysis (WGCNA), in conjunction with a dynamic tree-cutting algorithm, resulted in the identification of 25 co-expression modules. Eigenvector centrality analysis identified six hub genes responsive to cold stress: ZMYND15, ABHD17B, ATL8, WNK5, XTH3 and TPS. The findings of this study delineate three key aspects: (1) temporal dynamics of cold-responsive physiological processes, (2) pathway-level characterization of DEG enrichment patterns, and (3) genetic determinants underlying cold stress adaptation.

DISCUSSION: These findings clarify the time series and core physiological indicators of A. catechu during various physiological processes, identify pivotal genes associated with cold stress, and provide a gene-to-phenotype framework for optimizing cold-resilient cultivation protocols and molecular marker-assisted breeding strategies.

PMID:40765861 | PMC:PMC12321864 | DOI:10.3389/fpls.2025.1624335

Int J Biol Sci. 2025 Jul 11;21(10):4586-4603. doi: 10.7150/ijbs.109973. eCollection 2025.

ABSTRACT

Bladder cancer poses severe threats to human health due to its aggressive nature and resistance to drug treatment; however, the underlying mechanisms have not been fully investigated. In the present study, we identify SBSPON (Somatomedin B and Thrombospondin Type 1 Domain Containing) as a novel tumor suppressor. The expression of SBSPON was downregulated in bladder cancer and correlated with poor overall survival. SBSPON suppressed the proliferation and migration ability of bladder cancer cells in vitro, and inhibited tumor growth of bladder cancer cells in vivo. Genetic ablation of Sbspon in mice significantly accelerated the progression of N-butyl-N-(4-hydroxybutyl)-nitrosamine (BBN) induced bladder cancer. Mechanistically, SBSPON is a novel HSPA5 binding glycoprotein. SBSPON functioned through binding to HSPA5 and inhibiting its membrane translocation, resulting in an inactivated AKT signaling pathway. More importantly, SBSPON inhibited the cisplatin resistance of bladder cancer cells by reducing the inhibitory effect of HSPA5 on apoptosis. In summary, the novel glycoprotein SBSPON functions as a tumor suppressor and inhibits resistance to cisplatin in bladder cancer. This may provide novel therapeutic strategies for bladder cancer treatment.

PMID:40765821 | PMC:PMC12320502 | DOI:10.7150/ijbs.109973

Int J Biol Sci. 2025 Jul 11;21(10):4683-4700. doi: 10.7150/ijbs.113805. eCollection 2025.

ABSTRACT

Colorectal cancer (CRC) metastasis continues to account for a substantial proportion of cancer-related deaths worldwide. Calcium-binding protein S100A4 is a known executor of CRC metastasis. S100A4 has been correlated to metastasis formation in the past, and therefore pharmaceutical intervention reduces the metastatic phenotype. Herein, a high-throughput screen (HTS) of 105,600 compounds from the EMBL screening library using an S100A4 promoter-driven luciferase construct transfected into HCT116 cells identified novel compounds for S100A4 transcriptional inhibition. The most promising inhibitors identified were tested for S100A4 transcriptional inhibition, their impact on wound healing, migration, proliferation and viability of cancer cells. Subsequently, the leading candidate E12 was tested in vivo in a xenograft mouse model (HCT116/CMVp- Luc). After several testing rounds, E12 a 2-(4-fluorobenzenesulfonamido)benzamide-based compound showed the strongest inhibition of S100A4 expression at mRNA (EC50 < 1 µM; 48 h) and protein level and concomitant restriction of metastatic abilities in two CRC cell lines with a tolerable viability reduction. In vivo, a reduction in metastasis formation was demonstrated, displayed by reduced overall bioluminescence of tumors and human satellite DNA in the liver of treated mice. This study exhibited E12's promising potential for S100A4 targeted metastasis inhibition therapy to improve the outcome of metastasized CRC patients.

PMID:40765817 | PMC:PMC12320493 | DOI:10.7150/ijbs.113805

J Drug Target. 2025 Aug 5:1-29. doi: 10.1080/1061186X.2025.2544786. Online ahead of print.

ABSTRACT

Vascular diseases such as atherosclerosis, aneurysms, and peripheral arterial disease remain leading causes of morbidity and mortality, with current treatments primarily managing symptoms rather than addressing underlying molecular drivers. Gene therapy offers a promising avenue for targeted intervention, and recent advances in multi-omics approaches-including genomics, transcriptomics, and epigenetics-are enhancing the precision and efficacy of these therapies. High-throughput sequencing and integrative omics analyses have facilitated the identification of causal genes, non-coding RNAs, and epigenetic regulators involved in vascular pathology. This review examines how multi-omics frameworks inform gene therapy design, from genomic editing of cardiovascular disease loci to transcriptome-guided RNA therapies and epigenetic modulation of disease states. We highlight emerging applications such as CRISPR-based interventions, RNA therapeutics, and individualized precision medicine strategies. Additionally, we address analytical challenges, implementation hurdles, and ethical considerations in translating multi-omics-driven gene therapies into clinical practice. By integrating systems biology and advanced computational methods, the convergence of multi-omics and gene therapy holds transformative potential for vascular medicine, offering new avenues for disease modification and patient-specific therapeutic solutions.

PMID:40765035 | DOI:10.1080/1061186X.2025.2544786

Mol Syst Biol. 2025 Aug 5. doi: 10.1038/s44320-025-00134-0. Online ahead of print.

ABSTRACT

Generating longitudinal and multi-layered big biological data is crucial for effectively implementing artificial intelligence (AI) and systems biology approaches in characterising whole-body biological functions in health and complex disease states. Big biological data consists of multi-omics, clinical, wearable device, and imaging data, and information on diet, drugs, toxins, and other environmental factors. Given the significant advancements in omics technologies, human metabologenomics, and computational capabilities, several multi-omics studies are underway. Here, we first review the recent application of AI and systems biology in integrating and interpreting multi-omics data, highlighting their contributions to the creation of digital twins and the discovery of novel biomarkers and drug targets. Next, we review the multi-omics datasets generated worldwide to reveal interactions across multiple biological layers of information over time, which enhance precision health and medicine. Finally, we address the need to incorporate big biological data into clinical practice, supporting the development of a clinical decision support system essential for AI-driven hospitals and creating the foundation for an AI and systems biology-based healthcare model.

PMID:40764831 | DOI:10.1038/s44320-025-00134-0

Hepat Med. 2025 Aug 1;17:61-73. doi: 10.2147/HMER.S527644. eCollection 2025.

ABSTRACT

PURPOSE: Metabolic dysfunction-associated fatty liver disease (MAFLD) is considered a major global health concern. Considering the preliminary trend of hepatoprotective function of Liv.52 DS, the present study was conducted to explore its role in MAFLD.

PATIENTS AND METHODS: This randomized, double-blind, placebo-controlled, prospective, multicenter study was performed at four tertiary care hospitals in India. A total of 52 randomized subjects were administered either Liv.52 DS or placebo tablets twice daily for six months. Liver Stiffness Measurement (LSM) and Controlled Attenuated Parameter (CAP) values were compared at baseline and 6 months. After completion of the study, data from 47 subjects were available for analysis (31 in the Liv.52 DS group and 16 in the placebo group).

RESULTS: The mean LSM score, was reduced from 7.3 to 6.0 (Change From Baseline = 17.5%) in the active group with statistically significance (p = 0.007) compared to placebo group with LSM score reduction from 7.5 to 6.9 (CFB = 7.29%). A shift in the mean value from fibrosis (>6.0 kPa) to almost no significant fibrosis (<6.0 kPa), as per the Indian National Association for the Study of the Liver (INASL) cutoff, was achieved in the Liv.52 DS Group. Improvement was also observed in CAP values with Liv.52 DS, where 71% of the subjects showed an overall improvement in steatosis grade. The other liver markers like alanine transaminase (ALT) and aspartate aminotransferase (AST) were within the normal range. There were no cases of nephrotoxicity (common concern for herbal formulation), and no drug-related adverse events were reported.

CONCLUSION: A significant improvement in LSM and improvement in CAP was observed after 6 months of treatment with Liv.52 DS using fibroscan. This suggests that Liv.52 DS should be further explored for its potential role in the treatment of unmet medical needs in MAFLD patients.

PMID:40765845 | PMC:PMC12324062 | DOI:10.2147/HMER.S527644

Cureus. 2025 Jul 6;17(7):e87376. doi: 10.7759/cureus.87376. eCollection 2025 Jul.

ABSTRACT

Selective androgen receptor modulators (SARMs) causing drug-induced liver injury is a rare, albeit inadequately described, potentially serious side effect for those in the fitness industry looking to maximize muscle growth, strength gain, and fat loss as quickly as possible. We present a case of a patient with drug-induced liver injury after starting Stenabolic, a newer SARM. We report a case of a 40-year-old male who presented with vague gastrointestinal symptoms. Before the presentation, he was relatively healthy but taking multiple over-the-counter supplements. Although he had been taking most of these supplements for a long time without notable side effects, he had recently started taking Stenabolic, a performance-enhancing drug under the SARM category. Laboratory and imaging studies confirmed hepatocellular injury. After ruling out infectious and autoimmune etiology, it was thought that the likely source was Stenabolic. The patient was treated with supportive care and was advised to discontinue Stenabolic. Upon discharge, he began to show clinical improvement. Although there is limited research about Stenabolic, other agents in the SARM class have been implicated in similar patterns of liver injury. Its structural and pharmacologic similarities to anabolic steroids raise concern for hepatotoxicity through an idiosyncratic immune-mediated mechanism. This case highlights the potential hepatotoxicity of performance-enhancing supplements like Stenabolic. With the growing popularity of SARMs and limited regulation, healthcare providers should maintain a high index of suspicion for supplement-induced liver injury. Further research is needed to clarify the safety and mechanisms of these agents. Until then, their use should be discouraged.

PMID:40765588 | PMC:PMC12324147 | DOI:10.7759/cureus.87376

Osteoporos Int. 2025 Aug 6. doi: 10.1007/s00198-025-07604-z. Online ahead of print.

ABSTRACT

Deep learning is a machine learning method that mimics neural networks to build decision-making models. Recent advances in computing power and algorithms have enhanced deep learning's potential for vertebral fracture diagnosis in medical imaging. The application of deep learning in vertebral fracture diagnosis, including the identification of vertebrae and classification of vertebral fracture types, might significantly reduce the workload of radiologists and orthopedic surgeons as well as greatly improve the accuracy of vertebral fracture diagnosis. In this narrative review, we will summarize the application of deep learning models in the diagnosis of vertebral fractures.

PMID:40764417 | DOI:10.1007/s00198-025-07604-z

NPJ Digit Med. 2025 Aug 6;8(1):506. doi: 10.1038/s41746-025-01882-x.

ABSTRACT

We developed a rapid scanning optical microscope, termed "BlurryScope", that leverages continuous image acquisition and deep learning to provide a cost-effective and compact solution for automated inspection and analysis of tissue sections. This device offers comparable speed to commercial digital pathology scanners, but at a significantly lower price point and smaller size/weight. Using BlurryScope, we implemented automated classification of human epidermal growth factor receptor 2 (HER2) scores on motion-blurred images of immunohistochemically (IHC) stained breast tissue sections, achieving concordant results with those obtained from a high-end digital scanning microscope. Using a test set of 284 unique patient cores, we achieved testing accuracies of 79.3% and 89.7% for 4-class (0, 1+, 2+, 3+) and 2-class (0/1+, 2+/3+) HER2 classification, respectively. BlurryScope automates the entire workflow, from image scanning to stitching and cropping, as well as HER2 score classification.

PMID:40764388 | DOI:10.1038/s41746-025-01882-x

Sci Rep. 2025 Aug 5;15(1):28569. doi: 10.1038/s41598-025-14016-w.

ABSTRACT

Speech is one of the most efficient methods of communication among humans, inspiring advancements in machine speech processing under Natural Language Processing (NLP). This field aims to enable computers to analyze, comprehend, and generate human language naturally. Speech processing, as a subset of artificial intelligence, is rapidly expanding due to its applications in emotion recognition, human-computer interaction, and sentiment analysis. This study introduces a novel algorithm for emotion recognition from speech using deep learning techniques. The proposed model achieves up to a 15% improvement compared to state-of-the-art deep learning methods in speech emotion recognition. It employs advanced supervised learning algorithms and deep neural network architectures, including Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) with Long Short-Term Memory (LSTM) units. These models are trained on labeled datasets to accurately classify emotions such as happiness, sadness, anger, fear, surprise, and neutrality. The research highlights the system's real-time application potential, such as analyzing audience emotional responses during live television broadcasts. By leveraging advancements in deep learning, the model achieves high accuracy in understanding and predicting emotional states, offering valuable insights into user behavior. This approach contributes to diverse domains, including media analysis, customer feedback systems, and human-machine interaction, showcasing the transformative potential of combining speech processing with neural networks.

PMID:40764384 | DOI:10.1038/s41598-025-14016-w