J Med Imaging (Bellingham). 2025 Mar;12(2):024002. doi: 10.1117/1.JMI.12.2.024002. Epub 2025 Mar 26.

ABSTRACT

PURPOSE: Previous studies on echocardiogram segmentation are focused on the left ventricle in parasternal long-axis views. Deep-learning models were evaluated on the segmentation of the ventricles in parasternal short-axis echocardiograms (PSAX-echo). Segmentation of the ventricles in complementary echocardiogram views will allow the computation of important metrics with the potential to aid in diagnosing cardio-pulmonary diseases and other cardiomyopathies. Evaluating state-of-the-art models with small datasets can reveal if they improve performance on limited data.

APPROACH: PSAX-echo was performed on 33 volunteer women. An experienced cardiologist identified end-diastole and end-systole frames from 387 scans, and expert observers manually traced the contours of the cardiac structures. Traced frames were pre-processed and used to create labels to train two domain-specific (Unet-Resnet101 and Unet-ResNet50), and four general-domain [three segment anything (SAM) variants, and the Detectron2] deep-learning models. The performance of the models was evaluated using the Dice similarity coefficient (DSC), Hausdorff distance (HD), and difference in cross-sectional area (DCSA).

RESULTS: The Unet-Resnet101 model provided superior performance in the segmentation of the ventricles with 0.83, 4.93 pixels, and 106 pixel 2 on average for DSC, HD, and DCSA, respectively. A fine-tuned MedSAM model provided a performance of 0.82, 6.66 pixels, and 1252 pixel 2 , whereas the Detectron2 model provided 0.78, 2.12 pixels, and 116 pixel 2 for the same metrics, respectively.

CONCLUSIONS: Deep-learning models are suitable for the segmentation of the left and right ventricles in PSAX-echo. We demonstrated that domain-specific trained models such as Unet-ResNet provide higher accuracy for echo segmentation than general-domain segmentation models when working with small and locally acquired datasets.

PMID:40151505 | PMC:PMC11943840 | DOI:10.1117/1.JMI.12.2.024002

Front Big Data. 2025 Mar 13;8:1455442. doi: 10.3389/fdata.2025.1455442. eCollection 2025.

ABSTRACT

The exponential growth of image and video data motivates the need for practical real-time content-based searching algorithms. Features play a vital role in identifying objects within images. However, feature-based classification faces a challenge due to uneven class instance distribution. Ideally, each class should have an equal number of instances and features to ensure optimal classifier performance. However, real-world scenarios often exhibit class imbalances. Thus, this article explores the classification framework based on image features, analyzing balanced and imbalanced distributions. Through extensive experimentation, we examine the impact of class imbalance on image classification performance, primarily on large datasets. The comprehensive evaluation shows that all models perform better with balancing compared to using an imbalanced dataset, underscoring the importance of dataset balancing for model accuracy. Distributed Gaussian (D-GA) and Distributed Poisson (D-PO) are found to be the most effective techniques, especially in improving Random Forest (RF) and SVM models. The deep learning experiments also show an improvement as such.

PMID:40151465 | PMC:PMC11948280 | DOI:10.3389/fdata.2025.1455442

Ophthalmol Sci. 2024 Dec 17;5(3):100685. doi: 10.1016/j.xops.2024.100685. eCollection 2025 May-Jun.

ABSTRACT

PURPOSE: Early glaucoma detection is key to preventing vision loss, but screening often requires specialized eye examination or photography, limiting large-scale implementation. This study sought to develop artificial intelligence models that use self-reported health data from surveys to prescreen patients at high risk for glaucoma who are most in need of glaucoma screening with ophthalmic examination and imaging.

DESIGN: Cohort study.

PARTICIPANTS: Participants enrolled from May 1, 2018, to July 1, 2022, in the nationwide All of Us Research Program who were ≥18 years of age, had ≥2 eye-related diagnoses in their electronic health record (EHR), and submitted surveys with self-reported health history.

METHODS: We developed models to predict the risk of glaucoma, as determined by EHR diagnosis codes, using 3 machine learning approaches: (1) penalized logistic regression, (2) XGBoost, and (3) a fully connected neural network. Glaucoma diagnosis was identified based on International Classification of Diseases codes extracted from EHR data. An 80/20 train-test split was implemented, with cross-validation employed for hyperparameter tuning. Input features included self-reported demographics, general health, lifestyle factors, and family and personal medical history.

MAIN OUTCOME MEASURES: Models were evaluated using standard classification metrics, including area under the receiver operating characteristic curve (AUROC).

RESULTS: Among the 8205 patients, 873 (10.64%) were diagnosed with glaucoma. Across models, AUROC scores for identifying which patients had glaucoma from survey health data ranged from 0.710 to 0.890. XGBoost achieved the highest AUROC of 0.890 (95% confidence interval [CI]: 0.860-0.910). Logistic regression followed with an AUROC of 0.772 (95% CI: 0.753-0.795). Explainability studies revealed that key features included traditionally recognized risk factors for glaucoma, such as age, type 2 diabetes, and a family history of glaucoma.

CONCLUSIONS: Machine and deep learning models successfully utilized health data from self-reported surveys to predict glaucoma diagnosis without additional data from ophthalmic imaging or eye examination. These models may eventually enable prescreening for glaucoma in a wide variety of low-resource settings, after which high-risk patients can be referred for targeted screening using more specialized ophthalmic examination or imaging.

FINANCIAL DISCLOSURES: The author(s) have no proprietary or commercial interest in any materials discussed in this article.

PMID:40151359 | PMC:PMC11946806 | DOI:10.1016/j.xops.2024.100685

Comput Biol Med. 2025 Mar 26;190:110070. doi: 10.1016/j.compbiomed.2025.110070. Online ahead of print.

ABSTRACT

This study explores the development of a deep learning model using a neck-wearable piezoelectric sensor to accurately distinguish severe sleep apnea syndrome (SAS) from habitual snoring, addressing the underdiagnosis of SAS in adults. From 2018 to 2020, 60 adult habitual snorers underwent polysomnography while wearing a neck piezoelectric sensor that recorded snoring vibrations (70-250 Hz) and carotid artery pulsations (0.01-1.5 Hz). The initial dataset comprised 1167 silence, 1304 snoring, and 399 noise samples from 20 participants. Using a hybrid deep learning model comprising a one-dimensional convolutional neural network and gated-recurrent unit, the model identified snoring and apnea/hypopnea events, with sleep phases detected via pulse wave variability criteria. The model's efficacy in predicting severe SAS was assessed in the remaining 40 participants, achieving snoring detection rates of 0.88, 0.86, and 0.92, with respective loss rates of 0.39, 0.90, and 0.23. Classification accuracy for severe SAS improved from 0.85 for total sleep time to 0.90 for partial sleep time, excluding the first sleep phase, demonstrating precision of 0.84, recall of 1.00, and an F1 score of 0.91. This innovative approach of combining a hybrid deep learning model with a neck-wearable piezoelectric sensor suggests a promising route for early and precise differentiation of severe SAS from habitual snoring, aiding guiding further standard diagnostic evaluations and timely patient management. Future studies should focus on expanding the sample size, diversifying the patient population, and external validations in real-world settings to enhance the robustness and applicability of the findings.

PMID:40147187 | DOI:10.1016/j.compbiomed.2025.110070

Radiography (Lond). 2025 Mar 26;31(3):102929. doi: 10.1016/j.radi.2025.102929. Online ahead of print.

ABSTRACT

INTRODUCTION: A hybrid angio-CT system with 320-row detectors and deep learning-based reconstruction (DLR), provides additional imaging via 4D-CT angiography (CTA), potentially shortening procedure time and reducing DSA acquisitions, contrast media, and radiation dose. This study evaluates the feasibility of low-tube voltage 4D-CTA with low-contrast volume and DLR for selective renal artery embolization using a vessel phantom.

METHODS: A custom-made phantom simulating contrast-enhanced vessels filled with contrast medium was scanned. The study assessed image quality under varying image noise and vessel contrast. Quantitative analysis included peak contrast-to-noise ratio (pCNR) and image noise. Qualitative assessment was performed by seven radiologists using a 4-point scale; each radiologist independently recorded their evaluations on an assessment sheet.

RESULTS: A pCNR of approximately 15.0 was identified as the threshold for acceptable image quality. The pCNR decreased as the noise index increased (by 25-75 % when comparing a noise index of 30-70 HU).Vessels with a CT value of 500 Hounsfield units (HU) achieved sufficient image quality with a noise index of 50 HU. Dose reduction was substantial compared to traditional DSA, with effective radiation dose remaining within acceptable clinical levels.

CONCLUSION: 4D-CTA, combined with DLR, demonstrated the potential to reduce radiation and contrast agent usage while preserving diagnostic quality for renal artery angiography. Further clinical validation is required to confirm these findings in clinical settings.

IMPLICATIONS FOR PRACTICE: 4D-CTA with low-tube voltage and deep learning-based reconstruction (DLR) can reduce radiation and contrast use while maintaining image quality. This approach might improve safety, particularly in patients with renal impairment, and serve as a viable alternative to conventional DSA for selective renal artery embolization.

PMID:40147091 | DOI:10.1016/j.radi.2025.102929

Acad Radiol. 2025 Mar 26:S1076-6332(25)00263-6. doi: 10.1016/j.acra.2025.03.030. Online ahead of print.

NO ABSTRACT

PMID:40148166 | DOI:10.1016/j.acra.2025.03.030

Lancet Digit Health. 2025 Apr;7(4):e264-e274. doi: 10.1016/j.landig.2025.01.001.

ABSTRACT

BACKGROUND: Left ventricular systolic dysfunction (LVSD) is independently associated with cardiovascular events in patients with congenital heart disease. Although artificial intelligence-enhanced electrocardiogram (AI-ECG) analysis is predictive of LVSD in the general adult population, it has yet to be applied comprehensively across congenital heart disease lesions.

METHODS: We trained a convolutional neural network on paired ECG-echocardiograms (≤2 days apart) across the lifespan of a wide range of congenital heart disease lesions to detect left ventricular ejection fraction (LVEF) of 40% or less. Model performance was evaluated on single ECG-echocardiogram pairs per patient at Boston Children's Hospital (Boston, MA, USA) and externally at the Children's Hospital of Philadelphia (Philadelphia, PA, USA) using area under the receiver operating (AUROC) and precision-recall (AUPRC) curves.

FINDINGS: The training cohort comprised 124 265 ECG-echocardiogram pairs (49 158 patients; median age 10·5 years [IQR 3·5-16·8]; 3381 [2·7%] of 124 265 ECG-echocardiogram pairs with LVEF ≤40%). Test groups included internal testing (21 068 patients; median age 10·9 years [IQR 3·7-17·0]; 3381 [2·7%] of 124 265 ECG-echocardiogram pairs with LVEF ≤40%) and external validation (42 984 patients; median age 10·8 years [IQR 4·9-15·0]; 1313 [1·7%] of 76 400 ECG-echocardiogram pairs with LVEF ≤40%) cohorts. High model performance was achieved during internal testing (AUROC 0·95, AUPRC 0·33) and external validation (AUROC 0·96, AUPRC 0·25) for a wide range of congenital heart disease lesions. Patients with LVEF greater than 40% by echocardiogram who were deemed high risk by AI-ECG were more likely to have future dysfunction compared with low-risk patients (hazard ratio 12·1 [95% CI 8·4-17·3]; p<0·0001). High-risk patients by AI-ECG were at increased risk of mortality in the overall cohort and lesion-specific subgroups. Common salient features highlighted across congenital heart disaese lesions include precordial QRS complexes and T waves, with common high-risk ECG features including deep V2 S waves and lateral precordial T wave inversion. A case study on patients with ventricular pacing showed similar findings.

INTERPRETATION: Our externally validated algorithm shows promise in prediction of current and future LVSD in patients with congenital heart disease, providing a clinically impactful, inexpensive, and convenient cardiovascular health tool in this population.

FUNDING: Kostin Innovation Fund, Thrasher Research Fund Early Career Award, Boston Children's Hospital Electrophysiology Research Education Fund, National Institutes of Health, National Institute of Childhood Diseases and Human Development, and National Library of Medicine.

PMID:40148010 | DOI:10.1016/j.landig.2025.01.001

J Microbiol Biotechnol. 2025 Mar 26;35:e2411010. doi: 10.4014/jmb.2411.11010.

ABSTRACT

Group A Streptococcus (GAS) is a major pathogen with diverse clinical outcomes linked to its genetic variability, making accurate phenotype prediction essential. While previous studies have identified many GAS-associated genetic factors, translating these findings into predictive models remains challenging due to data complexity. The current study aimed to integrate deep learning models with genome-wide association study-derived genetic variants to predict pathogenic phenotypes in GAS. We evaluated the performance of several deep neural network models, including CNN, ResNet18, LSTM, and their ensemble approach in predicting GAS phenotypes. It was found that the ensemble model consistently achieved the highest prediction accuracy across phenotypes. Models trained on the full 4722-genotype set outperformed those trained on a reduced 175-genotype set, underscoring the importance of comprehensive variant data in capturing complex genotype-phenotype interactions. Performance changes in the reduced 175-genotype set compared to the full-set genotype scenarios revealed the impact of data dimensionality on model effectiveness, with CNN remaining robust, while ResNet18 and LSTM underperformed. Our findings emphasized the potential of deep learning in phenotype prediction and the critical role of data-model compatibility.

PMID:40147921 | DOI:10.4014/jmb.2411.11010

Zhonghua Kou Qiang Yi Xue Za Zhi. 2025 Mar 28;60(4):326-331. doi: 10.3760/cma.j.cn112144-20250116-00021. Online ahead of print.

ABSTRACT

Artificial intelligence (AI) has significantly enhanced the precision and efficiency of dental restoration design, smile analysis, and personalized treatment in the field of esthetic dentistry through technologies such as deep learning. This advancement provides patients with more accurate and efficient esthetic treatment solutions. However, its application still faces challenges such as technical limitations, ethical concerns, and insufficient data diversity. In the future, with the integration of high-quality data, optimization of real-time learning technologies, and the advancement of multidisciplinary collaboration, AI is expected to further promote the intelligent and human-centered development of esthetic dentistry, bringing profound and positive impacts to patients and clinical practice.

PMID:40147889 | DOI:10.3760/cma.j.cn112144-20250116-00021

AJNR Am J Neuroradiol. 2025 Mar 27. doi: 10.3174/ajnr.A8567. Online ahead of print.

ABSTRACT

BACKGROUND AND PURPOSE: Deep learning (DL)-based reconstruction enables improving the quality of MR images acquired with a short scan time. We aimed to prospectively compare the image quality and diagnostic performance in evaluating cervical degenerative spine diseases and myelopathy between conventional cervical MRI and accelerated cervical MRI with a commercially available vendor-neutral DL-based reconstruction.

MATERIALS AND METHODS: Fifty patients with degenerative cervical spine disease or myelopathy underwent both conventional cervical MRI and accelerated cervical MRI by using a DL-based reconstruction operating within the DICOM domain. The images were evaluated both quantitatively, based on SNR and contrast-to-noise ratio (CNR), and qualitatively, by using a 5-point scoring system for the overall image quality and clarity of anatomic structures on sagittal T1WI, sagittal contrast-enhanced (CE) T1WI, and axial/sagittal T2WI. Four radiologists assessed the sensitivity and specificity of the 2 protocols for detecting degenerative diseases and myelopathy.

RESULTS: The DL-based protocol reduced MRI acquisition time by 47%-48% compared with the conventional protocol. DL-reconstructed images demonstrated a higher SNR on sagittal T1WI (P = .046) and a higher CNR on sagittal T2WI (P = .03) than conventional images. The SNR on sagittal T2WI and the CNR on sagittal T1WI did not significantly differ (P > .05). DL-reconstructed images had better overall image quality on sagittal T1WI (P < .001), sagittal T2WI (Dixon in-phase or TSE) (P < .001), and sagittal T2WI (Dixon water-only) (P = .013) and similar image quality on axial T2WI and sagittal CE T1WI (P > .05). DL-reconstructed images had better clarity of anatomic structures (P values were < .001 for all structures, except for the neural foramen [P = .024]). DL-reconstructed images had a higher sensitivity for detecting neural foraminal stenosis (P = .005) and similar sensitivities for diagnosing other degenerative spinal diseases and myelopathy (P > .05). The specificities for diagnosing degenerative spinal diseases and myelopathy did not differ between the 2 images (P > .05).

CONCLUSIONS: The accelerated cervical MRI reconstructed with a vendor-neutral DL-based reconstruction algorithm did not compromise image quality and had higher or similar diagnostic performance for diagnosing cervical degenerative spine diseases and myelopathy compared with the conventional protocol.

PMID:40147833 | DOI:10.3174/ajnr.A8567

Semin Cancer Biol. 2025 Mar 25:S1044-579X(25)00052-5. doi: 10.1016/j.semcancer.2025.03.004. Online ahead of print.

ABSTRACT

Pancreatic ductal adenocarcinoma (PDAC) is recognized as one of the most lethal malignancies, characterized by late-stage diagnosis and limited therapeutic options. Risk stratification has traditionally been performed using epidemiological studies and genetic analyses, through which key risk factors, including smoking, diabetes, chronic pancreatitis, and inherited predispositions, have been identified. However, the multifactorial nature of PDAC has often been insufficiently addressed by these methods, leading to limited precision in individualized risk assessments. Advances in artificial intelligence (AI) have been proposed as a transformative approach, allowing the integration of diverse datasets-spanning genetic, clinical, lifestyle, and imaging data into dynamic models capable of uncovering novel interactions and risk profiles. In this review, the evolution of PDAC risk stratification is explored, with classical epidemiological frameworks compared to AI-driven methodologies. Genetic insights, including genome-wide association studies and polygenic risk scores, are discussed, alongside AI models such as machine learning, radiomics, and deep learning. Strengths and limitations of these approaches are evaluated, with challenges in clinical translation, such as data scarcity, model interpretability, and external validation, addressed. Finally, future directions are proposed for combining classical and AI-driven methodologies to develop scalable, personalized predictive tools for PDAC, with the goal of improving early detection and patient outcomes.

PMID:40147701 | DOI:10.1016/j.semcancer.2025.03.004

Neuroimage. 2025 Mar 25:121151. doi: 10.1016/j.neuroimage.2025.121151. Online ahead of print.

ABSTRACT

Alzheimer's disease (AD) is a primary degenerative brain disorder commonly found in the elderly, Mild cognitive impairment (MCI) can be considered a transitional stage from normal aging to Alzheimer's disease. Therefore, distinguishing between normal aging and disease-induced neurofunctional impairments is crucial in clinical treatment. Although deep learning methods have been widely applied in Alzheimer's diagnosis, the varying data formats used by different methods limited their clinical applicability. In this study, based on the ADNI dataset and previous clinical diagnostic experience, we propose a method using oblique coronal MRI to assist in diagnosis. We developed an algorithm to extract oblique coronal slices from 3D MRI data and used these slices to train classification networks. To achieve subject-wise classification based on 2D slices, rather than image-wise classification, we employed ensemble learning methods. This approach fused classification results from different modality images or different positions of the same modality images, constructing a more reliable ensemble classification model. The experiments introduced various decision fusion and feature fusion schemes, demonstrating the potential of oblique coronal MRI slices in assisting diagnosis. Notably, the weighted voting from decision fusion strategy trained on oblique coronal slices achieved accuracy rates of 97.5% for CN vs. AD, 100% for CN vs. MCI, and 94.83% for MCI vs. AD across the three classification tasks.

PMID:40147601 | DOI:10.1016/j.neuroimage.2025.121151

Cell. 2025 Mar 24:S0092-8674(25)00213-2. doi: 10.1016/j.cell.2025.02.025. Online ahead of print.

ABSTRACT

Efforts to use genome-wide assays or brain scans to diagnose autism have seen diminishing returns. Yet the clinical intuition of healthcare professionals, based on longstanding first-hand experience, remains the gold standard for diagnosis of autism. We leveraged deep learning to deconstruct and interrogate the logic of expert clinician intuition from clinical reports to inform our understanding of autism. After pre-training on hundreds of millions of general sentences, we finessed large language models (LLMs) on >4,000 free-form health records from healthcare professionals to distinguish confirmed versus suspected autism cases. By introducing an explainability strategy, our extended language model architecture could pin down the most salient single sentences in what drives clinical thinking toward correct diagnoses. Our framework flagged the most autism-critical DSM-5 criteria to be stereotyped repetitive behaviors, special interests, and perception-based behaviors, which challenges today's focus on deficits in social interplay, suggesting necessary revision of long-trusted diagnostic criteria in gold-standard instruments.

PMID:40147442 | DOI:10.1016/j.cell.2025.02.025

Comput Med Imaging Graph. 2025 Mar 23;123:102527. doi: 10.1016/j.compmedimag.2025.102527. Online ahead of print.

ABSTRACT

Medical image registration is a critical task in aligning medical images from different time points, modalities, or individuals, essential for accurate diagnosis and treatment planning. Despite significant progress in deep learning-based registration methods, current approaches still face considerable challenges, such as insufficient capture of local details, difficulty in effectively modeling global contextual information, and limited robustness in handling complex deformations. These limitations hinder the precision of high-resolution registration, particularly when dealing with medical images with intricate structures. To address these issues, this paper presents a novel registration network (TCDE-Net), an unsupervised medical image registration method based on a dual-encoder architecture. The dual encoders complement each other in feature extraction, enabling the model to effectively handle large-scale nonlinear deformations and capture intricate local details, thereby enhancing registration accuracy. Additionally, the detail-enhancement attention module aids in restoring fine-grained features, improving the network's capability to address complex deformations such as those at gray-white matter boundaries. Experimental results on the OASIS, IXI, and Hammers-n30r95 3D brain MR dataset demonstrate that this method outperforms commonly used registration techniques across multiple evaluation metrics, achieving superior performance and robustness. Our code is available at https://github.com/muzidongxue/TCDE-Net.

PMID:40147215 | DOI:10.1016/j.compmedimag.2025.102527

Comput Biol Med. 2025 Mar 26;190:110074. doi: 10.1016/j.compbiomed.2025.110074. Online ahead of print.

ABSTRACT

Numerical simulations of fluid dynamics in tube-like structures are important to biomedical research to model flow in blood vessels and airways. It is further useful to some clinical applications, such as predicting arterial fractional flow reserves, and assessing vascular flow wall shear stresses to predict atherosclerosis disease progression. Traditionally, they are conducted via computational fluid dynamics (CFD) simulations, which, despite optimization, still take substantial time, limiting clinical adoption. To improve efficiency, we investigate the use of the multi-case Neural Network (NN) to enable real-time predictions of fluid dynamics (both steady and pulsatile flows) in a 3D curved tube (with a narrowing in the middle mimicking a stenosis) of any shape within a geometric range, using only geometric parameters and boundary conditions. We compare the unsupervised approach guided by physics governing equations (physics informed neural network or PINN) to the supervised approach of using mass CFD simulations to train the network (supervised network or SN). We find that multi-case PINN can generate accurate velocity, pressure and wall shear stress (WSS) results under steady flow (spatially maximum error < 2-5 %), but this requires a specific enhancement strategies: (1) estimating the curvilinear coordinate parameters via a secondary NN to use as inputs into PINN, (2) imposing no-slip wall boundary condition as a hard constraint, and (3) advanced strategy to better spatially propagate effects of boundary conditions. However, we cannot achieve reasonable accuracy for pulsatile flow with PINN. Conversely, SN provides very accurate velocity, pressure, and WSS predictions under both steady and pulsatile flow scenarios (spatially and/or temporally maximum error averaged over all geometries <1 %), and is much less computationally expensive to train. To achieve this, strategies (1) and (2) above and a spectral encoding strategy for pulsatile flow are necessary. Thus, interestingly, the use of physics constraints is not effective in our application.

PMID:40147188 | DOI:10.1016/j.compbiomed.2025.110074

Med Ultrason. 2025 Mar 8. doi: 10.11152/mu-4495. Online ahead of print.

ABSTRACT

AIM: This study aimed to develop a deep learning (DL) model for automatic detection and diagnosis of gouty arthritis (GA) in the first metatarsophalangeal joint (MTPJ) using ultrasound (US) images.

MATERIALS AND METHODS: A retrospective study included individuals who underwent first MTPJ ultrasonography between February and July 2023. A five-fold cross-validation method (training set = 4:1) was employed. A deep residual convolutional neural network (CNN) was trained, and Gradient-weighted Class Activation Mapping (Grad-CAM) was used for visualization. Different ResNet18 models with varying residual blocks (2, 3, 4, 6) were compared to select the optimal model for image classification. Diagnostic decisions were based on a threshold proportion of abnormal images, determined from the training set.

RESULTS: A total of 2401 US images from 260 patients (149 gout, 111 control) were analyzed. The model with 3 residual blocks performed best, achieving an AUC of 0.904 (95% CI: 0.887~0.927). Visualization results aligned with radiologist opinions in 2000 images. The diagnostic model attained an accuracy of 91.1% (95% CI: 90.4%~91.8%) on the testing set, with a diagnostic threshold of 0.328.

CONCLUSION: The DL model demonstrated excellent performance in automatically detecting and diagnosing GA in the first MTPJ.

PMID:40146981 | DOI:10.11152/mu-4495

JMIR Diabetes. 2025 Mar 27;10:e66831. doi: 10.2196/66831.

ABSTRACT

Type 2 diabetes mellitus has seen a continuous rise in prevalence in recent years, and a similar trend has been observed in the increased availability of glucose-lowering drugs. There is a need to understand the variation in treatment response to these drugs to be able to predict people who will respond well or poorly to a drug. Electronic health records, clinical trials, and observational studies provide a huge amount of data to explore predictors of drug response. The use of artificial intelligence (AI), which includes machine learning and deep learning techniques, has the capacity to improve the prediction of treatment response in patients. AI can assist in the analysis of vast datasets to identify patterns and may provide valuable information on selecting an effective drug. Predicting an individual's response to a drug can aid in treatment selection, optimizing therapy, exploring new therapeutic options, and personalized medicine. This viewpoint highlights the growing evidence supporting the potential of AI-based methods to predict drug response with accuracy. Furthermore, the methods highlight a trend toward using ensemble methods as preferred models in drug response prediction studies.

PMID:40146874 | DOI:10.2196/66831

IEEE Trans Neural Netw Learn Syst. 2025 Mar 27;PP. doi: 10.1109/TNNLS.2025.3549816. Online ahead of print.

ABSTRACT

This article introduces the deep learning-based nonlinear model predictive controller with scene dynamics (DL-NMPC-SD) method for autonomous navigation. DL-NMPC-SD uses an a priori nominal vehicle model in combination with a scene dynamics model learned from temporal range sensing information. The scene dynamics model is responsible for estimating the desired vehicle trajectory, as well as to adjust the true system model used by the underlying model predictive controller. We propose to encode the scene dynamics model within the layers of a deep neural network, which acts as a nonlinear approximator for the high-order state space of the operating conditions. The model is learned based on temporal sequences of range-sensing observations and system states, both integrated by an Augmented Memory component. We use inverse reinforcement learning (IRL) and the Bellman optimality principle to train our learning controller with a modified version of the deep Q-learning (DQL) algorithm, enabling us to estimate the desired state trajectory as an optimal action-value function. We have evaluated DL-NMPC-SD against the baseline dynamic window approach (DWA), as well as against two state-of-the-art End2End and RL methods, respectively. The performance has been measured in three experiments: 1) in our GridSim virtual environment; 2) on indoor and outdoor navigation tasks using our RovisLab autonomous mobile test unit (AMTU) platform; and 3) on a full-scale autonomous test vehicle driving on public roads.

PMID:40146653 | DOI:10.1109/TNNLS.2025.3549816

Mol Inform. 2025 Mar;44(3):e202400325. doi: 10.1002/minf.202400325.

ABSTRACT

The blood brain barrier (BBB) is an endothelial-derived structure which restricts the movement of certain molecules between the general somatic circulatory system to the central nervous system (CNS). While the BBB maintains homeostasis by regulating the molecular environment induced by cerebrovascular perfusion, it also presents significant challenges in developing therapeutics intended to act on CNS targets. Many drug development practices rely partly on extensive cell and animal models to predict, to an extent, whether prospective therapeutic molecules can cross the BBB. In interest to reduce costs and improve prediction accuracy, many propose using advanced computational modeling of BBB permeability profiles leveraging empirical data. Given the scale of growth in machine learning and deep learning, we review the most recent machine learning approaches in predicting BBB permeability.

PMID:40146590 | DOI:10.1002/minf.202400325

Ophthalmol Ther. 2025 Mar 27. doi: 10.1007/s40123-025-01127-w. Online ahead of print.

ABSTRACT

INTRODUCTION: Existing deep learning (DL) algorithms lack the capability to accurately identify patients in immediate need of treatment for proliferative diabetic retinopathy (PDR). We aimed to develop a DL segmentation model to detect active PDR in six-field retinal images by the annotation of new retinal vessels and preretinal hemorrhages.

METHODS: We identified six-field retinal images classified at level 4 of the International Clinical Diabetic Retinopathy Disease Severity Scale collected at the Island of Funen from 2009 to 2019 as part of the Danish screening program for diabetic retinopathy (DR). A certified grader (grader 1) manually dichotomized the images into active or inactive PDR, and the images were then reassessed by two independent certified graders. In cases of disagreement, the final classification decision was made in collaboration between grader 1 and one of the secondary graders. Overall, 637 images were classified as active PDR. We then applied our pre-established DL segmentation model to annotate nine lesion types before training the algorithm. The segmentations of new vessels and preretinal hemorrhages were corrected for any inaccuracies before training the DL algorithm. After the classification and pre-segmentation phases the images were divided into training (70%), validation (10%), and testing (20%) datasets. We added 301 images with inactive PDR to the testing dataset.

RESULTS: We included 637 images of active PDR and 301 images of inactive PDR from 199 individuals. The training dataset had 1381 new vessel and preretinal hemorrhage lesions, while the validation dataset had 123 lesions and the testing dataset 374 lesions. The DL system demonstrated a sensitivity of 90% and a specificity of 70% for annotation-assisted classification of active PDR. The negative predictive value was 94%, while the positive predictive value was 57%.

CONCLUSIONS: Our DL segmentation model achieved excellent sensitivity and acceptable specificity in distinguishing active from inactive PDR.

PMID:40146482 | DOI:10.1007/s40123-025-01127-w