PLoS One. 2025 Jun 4;20(6):e0313734. doi: 10.1371/journal.pone.0313734. eCollection 2025.

ABSTRACT

Malaria continues to be a severe health problem across the globe, especially within resource-limited areas which lack both skilled diagnostic personnel and diagnostic equipment. This study investigates the use of deep learning diagnosis for malaria through ConvNeXt models that incorporate transfer learning techniques with data augmentation methods for better model performance and transferability. A total number of 606276 thin blood smear images served as the final augmented dataset after the initial 27558 images underwent augmentation. The ConvNeXt Tiny model, version V1 Tiny, achieved an accuracy of 95.9%.; however, the upgraded V2 Tiny Remod version exceeded this benchmark, reaching 98.1% accuracy. The accuracy rate measured 61.4% for Swin Tiny, ResNet18 reached 62.6%, and ResNet50 obtained 81.4%. The combination of label smoothing with the AdamW optimiser produced a model which exhibited strong robustness as well as generalisability. The enhanced ConvNeXt V2 Tiny model combined with data augmentation, transfer learning techniques and explainability frameworks demonstrate a practical solution for malaria diagnosis that achieves high accuracy despite limitations of access to large datasets and microscopy expertise, often observed in resource-limited regions. The findings highlight the potential for real-time diagnostic applications in remote healthcare facilities and the viability of ConvNeXt models in enhancing malaria diagnosis globally.

PMID:40465684 | DOI:10.1371/journal.pone.0313734

Alzheimers Dement. 2025 Jun;21(6):e70326. doi: 10.1002/alz.70326.

ABSTRACT

INTRODUCTION: We aimed to examine the global impact of brain small vessel disease (SVD) on cognitive performance.

METHODS: In 892 participants from the Multi-Ethnic Study of Atherosclerosis (MESA), we derived perivascular spaces (PVS), white matter hyperintensities (WMH), microbleeds (MB), and white matter fractional anisotropy (FA) and trace (TR). Cognitive function was assessed with a comprehensive neuropsychological battery.

RESULTS: A composite SVD measure was constructed as a linear combination of basal ganglia PVS, thalamus PVS, periventricular WMH, subcortical WMH, and white matter FA and TR, and exhibited associations with worse global and domain-specific cognitive performance. Additionally, SVD mediated the effect of age and cardiovascular disease risk on global cognitive function, both directly and through smaller gray matter (GM) volume.

DISCUSSION: Integrating multiple individual SVD endophenotypes may more accurately reflect the neurobiology of SVD and capture its global impact on cognition. SVD mediates the effects of age and cardiovascular disease risk on cognition through both atrophy-related and non-atrophy-related pathways.

HIGHLIGHTS: Associations between individual magnetic resonance imaging (MRI) markers of brain small vessel disease and cognitive outcomes might not fully capture the global impact of small vessel disease on cognition. We modeled small vessel disease as a latent construct, integrating multiple MRI endophenotypes in strategic brain regions. The small vessel disease construct was associated with worse global and domain-specific cognitive performance. The small vessel disease construct exhibited mediating effects in the relationships of aging and cardiovascular disease risk with cognition through pathways that both involve and are independent of brain atrophy. Integrating information from multiple relevant imaging endophenotypes could open new avenues in small vessel disease research, broadening our understanding of its risk factors and clinical correlates.

PMID:40465677 | DOI:10.1002/alz.70326

PLoS One. 2025 Jun 4;20(6):e0324294. doi: 10.1371/journal.pone.0324294. eCollection 2025.

ABSTRACT

As industrialization and the development of smart cities progress, effective waste collection, classification, and management have become increasingly vital. Recycling processes depend on accurately identifying and restoring waste materials to their original states, essential for reducing pollution and promoting environmental sustainability. In recent years, deep learning (DL) techniques have been applied strategically to enhance waste management processes, including capturing, classifying, composting, and disposing of waste. In light of the current context, the study presents an innovative waste classification model that utilizes a tailored DenseNet201 architecture coupled with an integrated Squeeze and Excitation (SE) attention mechanism and the fusion of parallel Convolutional Neural Network (CNN) branches. The integration of SE attention enables squeezing the irrelevant features and excites the important ones and the fusion of parallel CNN branches enhances the extraction of intricate, deeper, and more distinguishable features from waste data. The evaluation of the model across four publicly available datasets, along with three additional datasets to enhance waste diversity and the model's reliability, and the incorporation of Grad-CAM to visualize and interpret the model's focus areas for transparent decision-making, confirms its effectiveness in improving waste management practices. Furthermore, this model's successful deployment in a web-based sorting system marks a tangible stride in translating theoretical advancements into on-the-ground implementation, promising heightened efficiency and scalability in waste management practices. This work presents a precise solution for adaptable waste classification, heralding a paradigm shift in global waste disposal norms.

PMID:40465648 | DOI:10.1371/journal.pone.0324294

Proc Natl Acad Sci U S A. 2025 Jun 10;122(23):e2420158122. doi: 10.1073/pnas.2420158122. Epub 2025 Jun 4.

ABSTRACT

Characterizing the conformational ensemble of biomolecular systems is key to understand their functions. Cryoelectron microscopy (cryo-EM) captures two-dimensional snapshots of biomolecular ensembles, giving in principle access to thermodynamics. However, these images are very noisy and show projections of the molecule in unknown orientations, making it very difficult to identify the biomolecule's conformation in each individual image. Here, we introduce cryo-EM simulation-based inference (cryoSBI) to infer the conformations of biomolecules and the uncertainties associated with the inference from individual cryo-EM images. CryoSBI builds on simulation-based inference, a merger of physics-based simulations and probabilistic deep learning, allowing us to use Bayesian inference even when likelihoods are too expensive to calculate. We begin with an ensemble of conformations, templates from experiments, and molecular modeling, serving as structural hypotheses. We train a neural network approximating the Bayesian posterior using simulated images from these templates and then use it to accurately infer the conformation of the biomolecule from each experimental image. Training is only done once on simulations, and after that, it takes just a few milliseconds to make inference on an image, making cryoSBI suitable for arbitrarily large datasets and direct analysis on micrographs. CryoSBI eliminates the need to estimate particle pose and imaging parameters, significantly enhancing the computational speed compared to explicit likelihood methods. Importantly, we obtain interpretable machine learning models by integrating physics-based approaches with deep neural networks, ensuring that our results are transparent and reliable. We illustrate and benchmark cryoSBI on synthetic data and showcase its promise on experimental single-particle cryo-EM data.

PMID:40465628 | DOI:10.1073/pnas.2420158122

PLoS One. 2025 Jun 4;20(6):e0324513. doi: 10.1371/journal.pone.0324513. eCollection 2025.

ABSTRACT

SARS-CoV-2 infection has been associated with increased autoimmune disease risk. Past studies have not aligned regarding the most prevalent autoimmune diseases after infection, however. Furthermore, the relationship between infection severity and new autoimmune disease risk has not been well examined. We used RECOVER's electronic health record (EHR) networks, N3C, PCORnet, and PEDSnet, to estimate types and frequency of autoimmune diseases arising after SARS-CoV-2 infection and assessed how infection severity related to autoimmune disease risk. We identified patients of any age with SARS-CoV-2 infection between April 1, 2020 and April 1, 2021, and assigned them to a World Health Organization COVID-19 severity category for adults or the PEDSnet acute COVID-19 illness severity classification system for children (<age 21). We collected baseline covariates from the EHR in the year pre-index infection date and followed patients for 2 years for new autoimmune disease, defined as ≥ 2 new ICD-9, ICD-10, or SNOMED codes in the same concept set, starting >30 days after SARS-CoV-2 infection index date and occurring ≥1 day apart. We calculated overall and infection severity-stratified incidence ratesper 1000 person-years for all autoimmune diseases. With least severe COVID-19 severity as reference, survival analyses examined incident autoimmune disease risk. The most common new-onset autoimmune diseases in all networks were thyroid disease, psoriasis/psoriatic arthritis, and inflammatory bowel disease. Among adults, inflammatory arthritis was the most common, and Sjögren's disease also had high incidence. Incident type 1 diabetes and hematological autoimmune diseases were specifically found in children. Across networks, after adjustment, patients with highest COVID-19 severity had highest risk for new autoimmune disease vs. those with least severe disease (N3C: adjusted Hazard Ratio, (aHR) 1.47 (95%CI 1.33-1.66); PCORnet aHR 1.14 (95%CI 1.02-1.26); PEDSnet: aHR 3.14 (95%CI 2.42-4.07)]. Overall, severe acute COVID-19 was most strongly associated with autoimmune disease risk in three EHR networks.

PMID:40465573 | DOI:10.1371/journal.pone.0324513

PLoS Comput Biol. 2025 Jun 4;21(6):e1013082. doi: 10.1371/journal.pcbi.1013082. eCollection 2025 Jun.

ABSTRACT

Eastern equine encephalitis virus (EEEV) is an arthropod-borne, positive-sense RNA alphavirus posing a substantial threat to public health. Unlike similar viruses such as SARS-CoV-2, EEEV replicates efficiently in neurons, producing progeny viral particles as soon as 3-4 hours post-infection. EEEV infection, which can cause severe encephalitis with a human mortality rate surpassing 30%, has no licensed, targeted therapies, leaving patients to rely on supportive care. Although the general characteristics of EEEV infection within the host cell are well-studied, it remains unclear how these interactions lead to rapid production of progeny viral particles, limiting development of antiviral therapies. Here, we present a novel rule-based model that describes attachment, entry, uncoating, replication, assembly, and export of both infectious virions and virus-like particles within mammalian cells. Additionally, it quantitatively characterizes host ribosome activity in EEEV replication via a model parameter defining ribosome density on viral RNA. To calibrate the model, we performed experiments to quantify viral RNA, protein, and infectious particle production during acute infection. We used Bayesian inference to calibrate the model, discovering in the process that an additional constraint was required to ensure consistency with previous experimental observations of a high ratio between the amounts of full-length positive-sense viral genome and negative-sense template strand. Overall, the model recapitulates the experimental data and predicts that EEEV rapidly concentrates host ribosomes densely on viral RNA. Dense packing of host ribosomes was determined to be critical to establishing the characteristic positive to negative RNA strand ratio because of its role in governing the kinetics of transcription. Sensitivity analysis identified viral transcription as the critical step for infectious particle production, making it a potential target for future therapeutic development.

PMID:40465541 | DOI:10.1371/journal.pcbi.1013082

J Drugs Dermatol. 2025 Jun 1;24(6):617-620. doi: 10.36849/JDD.8590.

ABSTRACT

BACKGROUND: Periorificial dermatitis (POD) is a facial skin rash often found in the oral commissures and nasolabial folds, and around the eyes. Treatment options include topical metronidazole and azelaic acid, and oral tetracycline-class antibiotics. While the broad-spectrum antibiotics are efficacious, they can lead to adverse gastro-intestinal (GI) symptoms, negatively impact gut flora, and lead to antibiotic resistance. Narrow-spectrum tetracyclines, such as sarecycline, have a low potential for promoting bacterial resistance and GI issues.

OBJECTIVE: The main objective of this study is to demonstrate sarecycline&rsquo;s efficacy and safety in treating POD. It was hypothesized that subjects with POD given a 4-week course of sarecycline would have improvement in their POD and tolerate the medication well.

METHODS: This was a single-center, prospective, pilot study to evaluate the efficacy, safety, and tolerability of sarecycline for the treatment of POD with once-daily dosing over 4 consecutive weeks. Subjects were evaluated using the PODSI score at weeks 0, 2, and 4.

RESULTS: All 9 subjects who completed the study had shown improvement in POD with no reported drug-related adverse events. All subjects were female, and the mean age was 41 years old.

CONCLUSION: Sarecycline may be an appropriate novel treatment option for POD and should be explored further in a larger population study capturing this data. Furthermore, there is a need for more large-scale clinical studies evaluating treatment options for POD, with a focus on the impacts of antibiotic resistance and its implications on public health.

CITATION: Swenson K, Graber E. A single-center pilot study to evaluate the efficacy, safety, and tolerability of sarecycline for treating periorificial dermatitis. J Drugs Dermatol. 2025;24(6):617-620. doi:10.36849/JDD.8590.

PMID:40465502 | DOI:10.36849/JDD.8590

Clin Transl Sci. 2025 Jun;18(6):e70255. doi: 10.1111/cts.70255.

ABSTRACT

Pharmacogenetics is a promising strategy to facilitate individualized care for patients with Major Depressive Disorder (MDD). Research is ongoing to identify the optimal genetic markers for predicting outcomes to antidepressant therapies. The primary aim of this systematic review was to summarize antidepressant pharmacogenetic studies to enhance understanding of the genes, variants, datatypes/methodologies, and outcomes investigated in the context of MDD. The secondary aim was to identify clinical genetic panels indicated for antidepressant prescribing and summarize their genes and variants. Screening of N = 5793 articles yielded N = 390 for inclusion, largely comprising adult (≥ 18 years) populations. Top-studied variants identified in the search were discussed and compared with those represented on the N = 34 clinical genetic panels that were identified. Summarization of articles revealed sources of heterogeneity across studies and low rates of replicability of pharmacogenetic associations. Heterogeneity was present in outcome definitions, treatment regimens, and differential inclusion of mediating variables in analyses. Efficacy outcomes (i.e., response, remission) were studied at greater frequency than adverse-event outcomes. Studies that used advanced analytical approaches, such as machine learning, to integrate variants with complimentary biological datatypes were fewer in number but achieved higher rates of significant associations with treatment outcomes than candidate variant approaches. As large biological datasets become more prevalent, machine learning will be an increasingly valuable tool for parsing the complexity of antidepressant response. This review provides valuable context and considerations surrounding pharmacogenetic associations in MDD which will help inform future research and translation efforts for guiding antidepressant care.

PMID:40465332 | DOI:10.1111/cts.70255

JAMA Surg. 2025 Jun 4. doi: 10.1001/jamasurg.2025.1503. Online ahead of print.

ABSTRACT

IMPORTANCE: Pharmacogenetics can improve medication-related outcomes by optimizing efficacy and minimizing adverse effects. It is unknown whether the presence of drug-gene interactions (DGIs) at the time of surgery results in adverse outcomes in the postoperative setting.

OBJECTIVE: To determine the association of active DGIs on postsurgical outcomes following vascular surgery procedures.

DESIGN, SETTING, AND PARTICIPANTS: This was a retrospective cohort study of Veterans Affairs (VA) hospital patients participating in the Million Veteran Program who had a vascular procedure documented in the VA Surgical Quality Improvement Program (VASQIP) from January 1, 2011, to December 31, 2022. Data analysis was performed from June 1, 2023, to October 31, 2024.

EXPOSURE: Receipt of drugs impacted by pharmacogenetic variants 30 days prior to and up to 7 days following the vascular surgery procedure.

MAIN OUTCOMES AND MEASURES: Clinical outcomes collected as part of VASQIP, including length of stay (LOS), 30-day readmission, composite of myocardial infarction, stroke, and myocardial injury after noncardiac surgery, and 30-day postoperative death.

RESULTS: Among 10 098 patients (mean [SD] age, 68.8 [8.3] years; 1581 [15.7%] Black [self-reported]; 9884 [97.9%] male), 5020 (49.7%) had a DGI. The most common DGIs included proton pump inhibitors with CYP2C19, statins with SLCO1B1, and clopidogrel with CYP2C19. Compared with 0 DGIs, the presence of 1, 2, or 3 or more DGIs was associated with a longer median (IQR) LOS: with 0 DGIs, 3 (1-6) days vs 1 DGI, 3 (1-7) days (adjusted incidence rate ratio [IRR], 1.12; 95% CI, 1.10-1.14; P < .001); 2 DGIs, 3 (1-7) days (adjusted IRR, 1.22; 95% CI, 1.19-1.25; P < .001); and 3 or more DGIs, 4 (2-8) days (adjusted IRR, 1.40; 95% CI, 1.35-1.44; P < .001). The 30-day readmission rate, which was 17.4% among those with 0 DGIs, was not significantly different in those with 1 DGI (17.6%; adjusted odds ratio [aOR], 1.01; 95% CI, 0.90-1.14; P = .84) but was significantly higher in those with 2 DGIs (21.2%; aOR, 1.26; 95% CI, 1.08-1.47; P = .004) and 3 or more DGIs (25.1%; aOR, 1.61; 95% CI, 1.30-1.99; P < .001). The risk of the composite outcome, which was 3.5% in those with 0 DGIs, was not significantly different in those with 1 DGI (4.1%; aOR, 1.15; 95% CI, 0.91-1.45; P = .24) but was significantly higher in those with 2 DGIs (5.7%; aOR, 1.62; 95% CI, 1.22-2.15; P = .001) and those with 3 or more DGIs (5.5%; aOR, 1.60; 95% CI, 1.04-2.36; P = .02).

CONCLUSIONS AND RELEVANCE: The findings suggest that patients with DGIs at the time of vascular surgery have increased risk of cardiovascular morbidity, increased readmission, and longer LOS. Further work is needed to determine which DGIs contribute to these outcomes and whether preoperative pharmacogenetic testing has the potential to mitigate these risks.

PMID:40465239 | DOI:10.1001/jamasurg.2025.1503

J Community Genet. 2025 Jun 4. doi: 10.1007/s12687-025-00807-1. Online ahead of print.

ABSTRACT

Recent reports confirm that cystic fibrosis (CF) is a global disease. In Asian populations, both the spectrum of cystic fibrosis transmembrane conductance regulator (CFTR) gene mutations and the clinical course differ from those observed in Western populations. Although the recognition of CF is increasing in South Asia, comprehensive data from Southeast Asian countries remain sparse. The underdiagnosis of CF in Southeast Asia is attributed to limited awareness among healthcare professionals and restricted access to sweat chloride testing. Until 2021, CF had not been documented in the indigenous population of Vietnam. This study presents the first three confirmed cases of CF in native Vietnamese individuals. Additionally, a literature review of CF cases reported across Southeast Asia was conducted to provide insights into its prevalence and variations in CFTR mutation profiles within the region. A total of 50 cases were identified, distributed across Malaysia (30 cases), Thailand (8), the Philippines (6), Vietnam (5), and Indonesia (1), revealing a mutation spectrum distinct from that observed in Caucasian populations. The most common mutations included p.Phe508del and p.Ile1295PhefsX32, each found in 11.5% of cases. These findings highlight the need for increased clinical awareness, expanded access to sweat chloride testing, and the establishment of CF centers and regional CF registries to better understand and manage CF in Southeast Asia.

PMID:40465100 | DOI:10.1007/s12687-025-00807-1

Small. 2025 Jun 4:e2502739. doi: 10.1002/smll.202502739. Online ahead of print.

ABSTRACT

Triboelectric nanogenerators (TENGs) are an emerging technology that harvests abundant vibrational energy present in ambient environment. TENGs typically rely on polymer contact interfaces, which, while ideal for wearable and flexible applications, limit their applicability in industry settings, where high-temperature plant equipment generates plentiful and wasted vibrational energy. In this study, a biocompatible PDMS-hydrogel nanocomposite TENG is fabricated, containing nanoparticles of ZnAl-layered double hydroxide (LDH). This device demonstrates a maximum power density of 110 µW cm-2, and nanocomposite-based TENG shows exceptional stability in terms of output voltage up to 200 °C, making it suitable for harvesting waste vibrational energy from high-temperature industrial equipment. The fabricated TENG demonstrates its potential for structural health monitoring by exhibiting distinct energy spectral changes under different wave input excitations (sinusoidal, square, and triangular) at the same frequency, signifying its potential for vibration analysis of industrial machines. With its high-temperature functionality, the device remains applicable for wearable energy harvesting and human motion monitoring, ideal for monitoring in high-temperature environments. Here, this is demonstrated via a deep learning model for classification of human motions using the TENG voltage waveforms. The combination of high-temperature stability and wearable motion monitoring enables future industrial energy harvesting and extreme environment personnel monitoring.

PMID:40465357 | DOI:10.1002/smll.202502739

JMIR Med Inform. 2025 Jun 4;13:e68138. doi: 10.2196/68138.

ABSTRACT

BACKGROUND: The growing availability of electronic health records (EHRs) presents an opportunity to enhance patient care by uncovering hidden health risks and improving informed decisions through advanced deep learning methods. However, modeling EHR sequential data, that is, patient trajectories, is challenging due to the evolving relationships between diagnoses and treatments over time. Significant progress has been achieved using transformers and self-supervised learning. While BERT-inspired models using masked language modeling (MLM) capture EHR context, they often struggle with the complex temporal dynamics of disease progression and interventions.

OBJECTIVE: This study aims to improve the modeling of EHR sequences by addressing the limitations of traditional transformer-based approaches in capturing complex temporal dependencies.

METHODS: We introduce Trajectory Order Objective BERT (Bidirectional Encoder Representations from Transformers; TOO-BERT), a transformer-based model that advances the MLM pretraining approach by integrating a novel TOO to better learn the complex sequential dependencies between medical events. TOO-Bert enhanced the learned context by MLM by pretraining the model to distinguish ordered sequences of medical codes from permuted ones in a patient trajectory. The TOO is enhanced by a conditional selection process that focus on medical codes or visits that frequently occur together, to further improve contextual understanding and strengthen temporal awareness. We evaluate TOO-BERT on 2 extensive EHR datasets, MIMIC-IV hospitalization records and the Malmo Diet and Cancer Cohort (MDC)-comprising approximately 10 and 8 million medical codes, respectively. TOO-BERT is compared against conventional machine learning methods, a transformer trained from scratch, and a transformer pretrained on MLM in predicting heart failure (HF), Alzheimer disease (AD), and prolonged length of stay (PLS).

RESULTS: TOO-BERT outperformed conventional machine learning methods and transformer-based approaches in HF, AD, and PLS prediction across both datasets. In the MDC dataset, TOO-BERT improved HF and AD prediction, increasing area under the receiver operating characteristic curve (AUC) scores from 67.7 and 69.5 with the MLM-pretrained Transformer to 73.9 and 71.9, respectively. In the MIMIC-IV dataset, TOO-BERT enhanced HF and PLS prediction, raising AUC scores from 86.2 and 60.2 with the MLM-pretrained Transformer to 89.8 and 60.4, respectively. Notably, TOO-BERT demonstrated strong performance in HF prediction even with limited fine-tuning data, achieving AUC scores of 0.877 and 0.823, compared to 0.839 and 0.799 for the MLM-pretrained Transformer, when fine-tuned on only 50% (442/884) and 20% (176/884) of the training data, respectively.

CONCLUSIONS: These findings demonstrate the effectiveness of integrating temporal ordering objectives into MLM-pretrained models, enabling deeper insights into the complex temporal relationships inherent in EHR data. Attention analysis further highlights TOO-BERT's capability to capture and represent sophisticated structural patterns within patient trajectories, offering a more nuanced understanding of disease progression.

PMID:40465350 | DOI:10.2196/68138

Chaos. 2025 Jun 1;35(6):063115. doi: 10.1063/5.0270132.

ABSTRACT

Time-series analysis plays a crucial role in understanding the dynamics of real-world systems across various scientific and engineering disciplines. We in this paper propose a novel approach to identifying chaotic dynamics by a geometric method based on deep learning. Specifically, we construct a map from the observed time-series data and seek the existence of a topological horseshoe in the map, which indicates chaotic behavior. We demonstrate the effectiveness of our method by numerical experiments on the Hénon map, the Lorenz system, and the Duffing system. The results show that the topological horseshoe theory combined with deep neural works provides a valuable tool for detection of chaos in complex nonlinear systems from time series.

PMID:40465250 | DOI:10.1063/5.0270132

Neuroradiology. 2025 Jun 4. doi: 10.1007/s00234-025-03637-7. Online ahead of print.

ABSTRACT

PURPOSE: Human brain fiber tractography using diffusion magnetic resonance imaging is a crucial stage in mapping brain white matter structures, pre-surgical planning, and extracting connectivity patterns. Accurate and reliable tractography, by providing detailed geometric information about the position of neural pathways, minimizes the risk of damage during neurosurgical procedures.

METHODS: Both tractography itself and its post-processing steps such as bundle segmentation are usually used in these contexts. Many approaches have been put forward in the past decades and recently, multiple data-driven tractography algorithms and automatic segmentation pipelines have been proposed to address the limitations of traditional methods.

RESULTS: Several of these recent methods are based on learning algorithms that have demonstrated promising results. In this study, in addition to introducing diffusion MRI datasets, we review learning-based algorithms such as conventional machine learning, deep learning, reinforcement learning and dictionary learning methods that have been used for white matter tract, nerve and pathway recognition as well as whole brain streamlines or whole brain tractogram creation.

CONCLUSION: The contribution is to discuss both tractography and tract recognition methods, in addition to extending previous related reviews with most recent methods, covering architectures as well as network details, assess the efficiency of learning-based methods through a comprehensive comparison in this field, and finally demonstrate the important role of learning-based methods in tractography.

PMID:40464927 | DOI:10.1007/s00234-025-03637-7

Brain Inform. 2025 Jun 4;12(1):13. doi: 10.1186/s40708-025-00260-3.

ABSTRACT

Alzheimer's disease (AD) is a progressive neurodegenerative disorder that primarily affects memory, thinking, and behavior, leading to dementia, a severe cognitive decline. While no cure currently exists, recent advancements in preventive drug trials and therapeutic management have increased interest in developing clinical algorithms for early detection and biomarker identification. Electroencephalography (EEG) is noninvasive, cost-effective, and has high temporal resolution, making it a promising tool for automated AD detection. However, conventional machine learning approaches often fall short in accurately detecting AD due to their limited architectures. We also need to investigate the impact of EEG signal segment length on classification accuracy. To address these issues, a deep learning-based framework is proposed to detect AD using EEG data, focusing on determining the optimal segment length for classification. This framework contains EEG data collection, pre-processing for noise removal, temporal segmentation, convolutional neural network (CNN) model training and classification, and finally, evaluation. We have tested different segment lengths to test the impact on AD detection. We have used both 10-fold and leave-one-out cross-validation techniques and obtained accuracy of 97.08% and 96.90%, respectively, on a publicly available dataset from AHEPA General University Hospital of Thessaloniki. We have also tested the generalizability of the proposed model by testing it to detect frontotemporal dementia and obtained better results than existing studies. Furthermore, we have validated our proposed CNN model using several ablation studies and layer-wise extracted feature visualization. This study will establish a pioneering direction for future researchers and technology experts in the field of neurodiseases.

PMID:40464817 | DOI:10.1186/s40708-025-00260-3

ACS Nano. 2025 Jun 4. doi: 10.1021/acsnano.5c05698. Online ahead of print.

ABSTRACT

Accurate preoperative diagnosis of papillary thyroid carcinoma (PTC) histological subtypes and lymph node metastasis is essential for formulating personalized treatment strategies. However, their preoperative diagnosis is challenged by the limited reliability of cytological identification of histological subtypes and the low accuracy of lymph node detection using ultrasound imaging. Herein, a deep learning-assisted surface-enhanced Raman scattering (SERS) chip is developed for the preoperative diagnosis of PTC histological subtypes and evaluation of lymph node metastasis, using fine-needle aspiration (FNA) samples. The convolutional neural network algorithm is used to analyze Raman spectral fingerprints, successfully distinguishing PTC subtypes and lymph node metastasis with an accuracy of 95.83%. Moreover, the deep learning-assisted SERS platform has been successfully employed to identify central cervical lymph node metastasis with an accuracy of 100%. This approach highlights the potential of personalized medicine, facilitating the development of individualized treatment strategies, reducing overtreatment, and mitigating recurrence risk.

PMID:40464771 | DOI:10.1021/acsnano.5c05698

J Med Chem. 2025 Jun 4. doi: 10.1021/acs.jmedchem.5c00712. Online ahead of print.

ABSTRACT

Cyclic peptides have emerged as promising modulators of protein-protein interactions due to their unique pharmacological properties and ability to target extensive flat binding interfaces. However, traditional strategies for developing cyclic peptides are often hindered by significant resource constraints. Recent advancements in computational techniques and artificial intelligence-driven methodologies have significantly enhanced the cyclic peptide drug discovery pipeline, while breakthroughs in automated synthesis platforms have accelerated experimental validation, presenting transformative potential for pharmaceutical innovation. In this review, we examine state-of-the-art computational and artificial intelligence-driven strategies that address challenges such as peptide flexibility, limited data availability, and complex conformational landscapes. We discuss how the integration of physics-based simulations with deep learning techniques is redefining the design and optimization of cyclic peptide therapeutics and propose future perspectives to advance the precision and efficiency of cyclic peptide drug development, ultimately offering innovative solutions to unmet medical needs.

PMID:40464341 | DOI:10.1021/acs.jmedchem.5c00712

J Oral Rehabil. 2025 Jun 4. doi: 10.1111/joor.14045. Online ahead of print.

ABSTRACT

BACKGROUND: Artificial intelligence (AI) and various subunits of AI such as artificial neural networks (ANN), Convolutional neural networks (CNNs), machine learning (ML), deep learning (DL) and deep neural networks (DNN) are being tried to diagnose and plan treatment for periodontal diseases.

AIM: This narrative review aims to discuss the current evidence on the applications of AI for the diagnosis and risk prediction of periodontal/peri-implant diseases.

METHOD: A search strategy with the following keywords: (Artificial intelligence [MeSH Terms]) AND (Periodontal disease [MeSH Terms]) was used to search for articles from 2000 to 2024.

RESULTS: AI models using patient-related data, signs and symptoms of the disease, immunological biomarkers and microbial profiles aid in effective diagnosis and planning treatment. AI is also used in periodontal diagnosis of pathological and anatomical landmarks such as cementoenamel junction, bone levels, furcation defects, nature and system of dental implants placed, degree of implant or tooth fractures and periapical pathology, assessing the severity and grading of periodontal or peri-implant disease/conditions, assessing the signs and symptoms of periodontal/peri-implant disease and determining the prognosis of implant and periodontal treatment. Studies have compared the diagnosis made by dentists and AI-based models and found AI models to be more effective and quicker in diagnosis than dentists.

DISCUSSION: AI-based tools such as DL, ML, CNN, and ANN are more effective and quicker for timely diagnosis, risk assessment, and treatment plans for periodontal and peri-implant disease diagnosis. DL and CNN are the most commonly used tools for the diagnosis of bone levels around teeth or implants, periodontal disease staging and severity, and location of anatomical structures and teeth.

CONCLUSION: AI and its subsets are promising tools for the diagnosis/risk prediction and treatment planning for periodontal and peri-implant diseases.

PMID:40464289 | DOI:10.1111/joor.14045

J Cell Mol Med. 2025 Jun;29(11):e70639. doi: 10.1111/jcmm.70639.

ABSTRACT

A cellular characteristic of IPF is the transformation of fibrosis into myofibroblasts. This study identifies several transcription factors-STAT3, FOXP1, JUNB, ATF3, FosL2, BATF, Fra2 and AP-1-that play crucial roles in promoting pulmonary fibrogenesis. They achieve this by facilitating the differentiation of fibroblasts into myofibroblasts, as analysed through ATAC-seq and RNA-seq. Additionally, STAT3 ChIP-seq showed that STAT3 is significantly concentrated in accessible chromatin regions, including introns and intergenic areas. H3K27ac ChIP-seq and Co-IP demonstrated that STAT3 plays a role in the formation of super enhancer (SE), which promotes gene expression. CUT&RUN-qPCR and the pGL3-SE dual-luciferase reporter system assays proved that STAT3 enhanced pGL3-SE activities by facilitating H3K27ac modification, leading to promoting the transcription of target genes including RUNX1, JUNB, JUN, SMAD6, COL3A1 and PTPN1. In summary, this study shows that STAT3 contributes to the formation of SEs that accelerate the differentiation of fibroblasts into myofibroblasts, leading to IPF. This insight enhances our understanding of STAT3-related SEs and offers potential therapeutic strategies for fibrotic diseases.

PMID:40464702 | DOI:10.1111/jcmm.70639

Angew Chem Int Ed Engl. 2025 Jun 4:e202507254. doi: 10.1002/anie.202507254. Online ahead of print.

ABSTRACT

Crosslinking strategies have emerged as an attractive technology for deciphering protein complexes and protein-protein interactions (PPIs). However, commonly used crosslinking strategies present significant challenges for the precise analysis of protein complexes and dynamic PPIs in native biological environments. Here, we report the development of the first visible-light-inducible lysine-specific homobifunctional photo-crosslinkers and introduce Visible-light-controlled Lysine-selective crosslinking (VL-XL) strategy for in-depth analysis of protein complexes and profiling dynamic interactomes in live cells. By synergistically integrating the advantages of temporal control, high biocompatibility, and lysine selectivity, the VL-XL strategy not only provides an effective solution for protein complexes studies-achieving residue-specific crosslinked peptides, delivering high-confidence data and streamlined MS data analysis-but also reveals dynamic interactomes in various scenarios. The VL-XL strategy successfully profiles the time-resolved, EGF-stimulated EGFR interactome, providing valuable insights into regulatory mechanisms of EGFR signaling. More importantly, the VL-XL strategy effectively unveils molecular glue degrader-induced E3 ligase interactome, leading to discovery of neo-substrates such as SESN2 and opening an innovative avenue for identifying novel targets for degradation. Overall, the VL-XL strategy offers a robust chemical tool for decoding protein complexes and dynamic interactomes, inspiring innovative solutions to address unresolved questions in proteomics, systems biology and drug discovery.

PMID:40464585 | DOI:10.1002/anie.202507254