Enhanced thermoelectric performance of copper iodide particles/nanowires composite in the low-temperature range.

Thermoelectric (TE) energy harvesting presents a viable method for reducing energy waste by transforming waste thermal energy into electricity. In this study, we fabricated copper iodide (CuI) composites using synthesized CuI nanowires (NWs) and particles to enhance TE performance in the low-temperature range. The Seebeck coefficient (S) was notably higher when a combination of CuI particles and NWs was used, reaching a maximum S of 1614.24 µV K-1 with a 60% NWs content at RT. Electrical conductivity (σ) exhibited an inverse correlation with S, with higher values detected when either particles or NWs were used only. The highest power factor (PF) of 128.44 µW m-1K-2 was recorded at RT with 60% NWs content, demonstrating improved TE performance. Thermal conductivity (κ) diminished when different material structures were employed, enhancing phonon scattering. The maximum figure of merit (ZT) achieved was ∼0.14 with 60% NWs content at 425 K, indicating the potential of this method for improving TE performance. This study offers valuable insights into optimizing TE performance using CuI composites, proposing a promising strategy for energy harvesting from low-temperature sources.

In Silico Analysis and Biochemical Characterization of Streptomyces PET Hydrolase with Bis(2-Hydroxyethyl) Terephthalate Biodegradation Activity.

Polyethylene terephthalate (PET), one of the most widely used plastics in the world, causes serious environmental problems. Recently, scientists have been focused on the enzymatic degradation of PET, an environmentally friendly method that offers an attractive approach to the degradation and recycling of PET. In this work, PET hydrolase from Streptomyces sp. W2061 was biochemically characterized, and the biodegradation of PET was performed using the PET model substrate bis (2- hydroxyethyl terephthalate) (BHET). PET hydrolase has an isoelectric point of 5.84, and a molecular mass of about 50.31 kDa. The optimum pH and temperature were 7.0 and 40°C, respectively. LC-MS analysis of the enzymatic products showed that the PET hydrolase successfully degraded a single ester bond of BHET, leading to the formation of MHET. Furthermore, in silico characterization of the PET hydrolase protein sequence and its predicted three-dimensional structure was designed and compared with the well-characterized IsPETase from Ideonella sakaiensis. The structural analysis showed that the (Gly-x1-Ser-x2-Gly) serine hydrolase motif and the catalytic triad (Ser, Asp, and His) were conserved in all sequences. In addition, we integrated molecular dynamics (MD) simulations to analyze the variation in the structural stability of the PET hydrolase in the absence and presence of BHET. These simulations showed the formation of a stable complex between the PET hydrolase and BHET. To the best of our knowledge, this is the first study on Streptomyces sp. W2061 to investigate the BHET degradation activity of PET hydrolase, which has potential application in the biodegradation of plastics in the environment.

Optimizing Disease Outbreak Forecast Ensembles.

On the basis of historical influenza and COVID-19 forecasts, we found that more than 3 forecast models are needed to ensure robust ensemble accuracy. Additional models can improve ensemble performance, but with diminishing accuracy returns. This understanding will assist with the design of current and future collaborative infectious disease forecasting efforts.

Horizontal Lithium Electrodeposition on Atomically Polarized Monolayer Hexagonal Boron Nitride.

Both uncontrolled Li dendrite growth and corrosion are major obstacles to the practical application of Li-metal batteries. Despite numerous attempts to address these challenges, effective solutions for dendrite-free reversible Li electrodeposition have remained elusive. Here, we demonstrate the horizontal Li electrodeposition on top of atomically polarized monolayer hexagonal boron nitride (hBN). Theoretical investigations revealed that the hexagonal lattice configuration and polarity of the monolayer hBN, devoid of dangling bonds, reduced the energy barrier for the surface diffusion of Li, thus facilitating reversible in-plane Li growth. Moreover, the single-atom-thick hBN deposited on a Cu current collector (monolayer hBN/Cu) facilitated the formation of an inorganic-rich, homogeneous solid electrolyte interphase layer, which enabled the uniform Li+ flux and suppressed Li corrosion. Consequently, Li-metal and anode-free full cells containing the monolayer hBN/Cu exhibited improved rate performance and cycle life. This study suggests that the monolayer hBN is a promising class of underlying seed layers to enable dendrite- and corrosion-free, horizontal Li electrodeposition for sustainable Li-metal anodes in next-generation batteries.

Evaluation of fluid responsiveness using liver compression in dogs with experimentally induced hypovolemia.

OBJECTIVE: To investigate whether liver compression (LC) could increase stroke volume (SV) by more than 15% in healthy, anesthetized dogs with hypovolemia and suggest LC as a novel method to evaluate fluid responsiveness. ANIMALS: 6 healthy Beagles. METHODS: This prospective, nonrandomized experimental study was conducted from November 2023 to February 2024. The dogs were anesthetized with isoflurane and mechanically ventilated under neuromuscular blockade. After instrumentation, the dogs underwent the following 4 experimental stages in a sequential, nonrandomized manner: stage 1, baseline; stage 2, 30% withdrawal of circulating blood volume; stage 3, 50% infusion of the collected blood; and stage 4, the remaining 50% infusion of the collected blood. At each stage, SV via pulmonary artery thermodilution and hemodynamic variables were measured before, during, and after the LC. RESULTS: In stage 2, LC significantly increased mean SV by 30%, from 6.9 to 9 mL/beat. Simultaneously, LC significantly increased mean arterial pressure by 11 mm Hg and mean central venous pressure by 2 mm Hg, while pulse pressure variation significantly decreased from 28% to 22%. In stages 1, 3, and 4, LC did not significantly change mean SV, mean arterial pressure, and pulse pressure variation; however, mean central venous pressure significantly increased during stage 3. CLINICAL RELEVANCE: This study demonstrates that LC at 22 mm Hg for 1 minute could increase SV more than 15% in anesthetized, hypovolemic dogs and LC could be used as a novel method to evaluate fluid responsiveness.

Long-Term Enhancement of Botulinum Toxin Injections for Post-Stroke Spasticity by Use of Stretching Exercises-A Randomized Controlled Trial.

Botulinum toxin A (BONT/A) injections play a central role in the treatment of upper limb spasticity in stroke patients. We proposed structured stretching exercises to enhance the effect of post-stroke spasticity relief of the upper limbs following BONT/A injections. A total of 43 patients who had a stroke with grade 2 spasticity or higher on the Modified Ashworth Scale (MAS) in their upper-limb muscles were randomly assigned to the intervention (n = 21) or control group (n = 22). The former received structured stretching exercises after their BONT/A injections for 20 min, 5 days per week, for 6 months at a hospital, while the others conducted self-stretching exercises at home. The outcome measures were assessed before the intervention (T0) and after three (T1) and six months (T2). Significantly greater improvements in the MAS scores of the elbows, wrists, and fingers were found in the intervention group's patients at T1 and T2. The behavioral outcome measures, including shoulder pain, activities of daily living, and quality of life, and our electrophysiological studies also showed a significantly higher enhancement in this patient group. In conclusion, the structured stretching exercises plus BONT/A injections for six months showed a superior effect in relieving post-stroke upper-limb spasticity compared to self-stretching exercises.

Scalable electron tomography for connectomics.

We demonstrate limited-tilt, serial section electron tomography (ET), which can non-destructively map brain circuits over large 3D volumes and reveal high-resolution, supramolecular details within subvolumes of interest. We show accelerated ET imaging of thick sections (>500 nm) with the capacity to resolve key features of neuronal circuits including chemical synapses, endocytic structures, and gap junctions. Furthermore, we systematically assessed how imaging parameters affect image quality and speed to enable connectomic-scale projects.

The Development of an Artificial Intelligence Video Analysis-Based Web Application to Diagnose Oropharyngeal Dysphagia: A Pilot Study.

The gold standard test for diagnosing dysphagia is the videofluoroscopic swallowing study (VFSS). However, the accuracy of this test varies depending on the specialist's skill level. We proposed a VFSS-based artificial intelligence (AI) web application to diagnose dysphagia. Video from the VFSS consists of multiframe data that contain approximately 300 images. To label the data, the server separated them into frames during the upload and stored them as a video for analysis. Then, the separated data were loaded into a labeling tool to perform the labeling. The labeled file was downloaded, and an AI model was developed by training with You Only Look Once (YOLOv7). Using a utility called SplitFolders, the entire dataset was divided according to a ratio of training (70%), test (10%), and validation (20%). When a VFSS video file was uploaded to an application equipped with the developed AI model, it was automatically classified and labeled as oral, pharyngeal, or esophageal. The dysphagia of a person was categorized as either penetration or aspiration, and the final analyzed result was displayed to the viewer. The following labeling datasets were created for the AI learning: oral (n = 2355), pharyngeal (n = 2338), esophageal (n = 1480), penetration (n = 1856), and aspiration (n = 1320); the learning results of the YOLO model, which analyzed dysphagia using the dataset, were predicted with accuracies of 0.90, 0.82, 0.79, 0.92, and 0.96, respectively. This is expected to help clinicians more efficiently suggest the proper dietary options for patients with oropharyngeal dysphagia.

Therapeutic Efficacy of Ultrasound-Guided Selective Nerve Block on Chronic Cervical Radiculopathy.

Background and Objectives: Cervical radiculopathy (CR) manifests as pain and sensorimotor disturbances in the upper extremities, often resulting from nerve root compression due to intervertebral disc herniation, degenerative changes, or trauma. While conservative treatments are initially preferred, persistent or severe cases may require surgical intervention. Ultrasound-guided selective nerve root block (SNRB) has emerged as a promising intervention for alleviating symptoms and potentially obviating the need for surgery. This study evaluates the therapeutic efficacy of ultrasound-guided SNRB in managing chronic CR, aiming to determine its potential in symptom relief and delaying or avoiding surgical procedures. Materials and Methods: A retrospective analysis was conducted on 720 outpatients treated for CR between October 2019 and March 2022. After excluding patients with traumatic CR, previous surgeries, malignancies, progressive neurological symptoms requiring immediate surgery, or inadequate conservative treatment, 92 patients who had experienced cervical radicular pain for more than three months and had failed to improve after more than six weeks of conservative treatment with VAS scores ≥ 5 were included. The patients underwent single or multiple ultrasound-guided SNRB procedures, involving the injection of dexamethasone and lidocaine under real-time ultrasound guidance. Symptom severity was assessed at the baseline, and at 4, 8, and 12 weeks post-procedure using the Visual Analog Scale (VAS). The data collected included age, sex, presence of neck and/or radicular pain, physical examination findings, recurrence of symptoms, improvement in symptoms, and whether surgical intervention was ultimately required. Statistical analyses were performed to identify the factors associated with symptom improvement or recurrence. Results: Significant symptom improvement was observed in 69 (75.0%) participants post-SNRB, with 55 (79.7%) showing improvement at 4 weeks, 11 (15.9%) at 8 weeks, and 3 (4.4%) at 12 weeks. Symptom recurrence, defined by an increase in VAS score accompanied by a pain flare lasting at least 24 h after a pain-free interval of at least one month, was noted in 48 (52.2%) patients. The presence of combined neck and radicular pain was a significant predictor of recurrence (p = 0.008). No significant associations were found between symptom relief and factors such as age, gender, initial pain severity, or MRI findings. Conclusions: Ultrasound-guided SNRB effectively manages chronic CR, providing substantial symptom relief and potentially reducing the need for surgical intervention. This technique offers a promising conservative treatment option, especially given its real-time visualization advantages and minimal radiation exposure.

HiRENet: Novel convolutional neural network architecture using Hilbert-transformed and raw electroencephalogram (EEG) for subject-independent emotion classification.

BACKGROUND AND OBJECTIVES: Convolutional neural networks (CNNs) are the most widely used deep-learning framework for decoding electroencephalograms (EEGs) due to their exceptional ability to extract hierarchical features from high-dimensional EEG data. Traditionally, CNNs have primarily utilized multi-channel raw EEG data as the input tensor; however, the performance of CNN-based EEG decoding may be enhanced by incorporating phase information alongside amplitude information. METHODS: This study introduces a novel CNN architecture called the Hilbert-transformed (HT) and raw EEG network (HiRENet), which incorporates both raw and HT EEG as inputs. This concurrent use of HT and raw EEG aims to integrate phase information with existing amplitude information, potentially offering a more comprehensive reflection of functional connectivity across various brain regions. The HiRENet model was developed using two CNN frameworks: ShallowFBCSPNet and a CNN with a residual block (ResCNN). The performance of the HiRENet model was assessed using a lab-made EEG database to classify human emotions, comparing three input modalities: raw EEG, HT EEG, and a combination of both signals. Additionally, the computational complexity was evaluated to validate the computational efficiency of the ResCNN design. RESULTS: The HiRENet model based on ResCNN achieved the highest classification accuracy, with 86.03% for valence and 84.01% for arousal classifications, surpassing traditional CNN methodologies. Considering computational efficiency, ResCNN demonstrated superiority over ShallowFBCSPNet in terms of speed and inference time, despite having a higher parameter count. CONCLUSION: Our experimental results showed that the proposed HiRENet can be potentially used as a new option to improve the overall performance for deep learning-based EEG decoding problems.

Direct oral anticoagulants for oral anticoagulants-naïve Asian patients with atrial fibrillation and end-stage renal disease undergoing dialysis.

In Asian patients with atrial fibrillation (AF) and end-stage renal disease (ESRD) undergoing dialysis, the use of direct oral anticoagulants (DOACs) remains debatable. From the national health insurance claims data in South Korea, we included 425 new users of OAC among patients with non-valvular AF and ESRD undergoing dialysis between 2013 and 2020. Patients were categorized into DOAC (n = 106) and warfarin group (n = 319). Clinical outcomes, including ischemic stroke, myocardial infarction (MI), intracranial hemorrhage (ICH), and gastrointestinal (GI) bleeding, were compared between the two groups using inverse probability of treatment weighting (IPTW) analysis. During the median follow-up of 3.2 years, the incidence of ischemic stroke was significantly reduced in the DOAC compared to the warfarin group [Hazard ratio (HR) 0.07; P = 0.001]. However, the incidence of MI (HR 1.32; P = 0.41) and GI bleeding (HR 1.78; P = 0.06) were not significantly different between the two groups. No ICH events occurred in the DOAC group, although the incidence rate did not differ significantly between the two groups (P = 0.17). In Asian patients with AF and ESRD undergoing dialysis, DOACs may be associated with a reduced risk of ischemic stroke compared with warfarin. The MI, ICH, and GI bleeding rates may be comparable between DOACs and warfarin.

Unilateral retrobulbar haemorrhage in a cat secondary to suspected anticoagulant rodenticide intoxication.

A 6-month-old intact female mixed-breed kitten presented with severe exophthalmos of the left eye. Periocular lesions, including subconjunctival haemorrhage, third eyelid protrusion, and left eyelid oedema, were detected in the absence of globe retropulsion. The left intraocular pressure was increased, and ocular ultrasonography revealed ipsilateral retrobulbar fluid. Coagulation panels were markedly prolonged and severe anaemia was detected. Ultrasound-guided retrobulbar centesis performed to decrease intraocular pressure yielded blood. Based on the history and clinical findings, anticoagulant rodenticide intoxication was suspected. Treatment included partial tarsorrhaphy and the administration of topical antibiotics, artificial tears, and vitamin K1. Fresh whole blood and fresh frozen plasma were transfused for supportive therapy. Coagulation parameters improved after 7 days of hospitalisation. The periocular lesions resolved within 14 days, despite persistent optic nerve damage and blindness. This case report raises the possibility that anticoagulant rodenticide toxicity may result in retrobulbar haemorrhage in the absence of other typical cavitary bleeding. Although uncommon, anticoagulant rodenticide toxicity should be considered in cats with retrobulbar haemorrhage.

Restoration Temperature Control through Glass Transition Temperature Modulation of Shape Memory Polymer for Thermally Switchable Adhesive.

Shape memory polymers (SMPs) undergo changes between arbitrary shapes and programmed shapes upon exposure to specific stimulus, allowing them to restore their original shape. All kinds of external stimuli have a threshold to change the shape of the SMP. Especially, for the thermal type SMP, the critical temperature for shape restoration is typically near the glass transition temperature (Tg). In this study, the controllability of the restoration temperature is analyzed by adjusting the Tg of the polymer using Norland Optical Adhesive 63, which can be cured with UV irradiation. By varying the ambient temperature from 20 to 120 °C during UV exposure, Tg changes ranging from 35.84 to 50.50 °C are obtained, with corresponding changes in restoration temperature. As a practical application, a thermal-activated SMP dry adhesive is developed with programmable Tg and switchable adhesion. The fabricated SMP dry adhesive exhibited strong adhesion to substrates with various surface roughness. Additionally, the shape memory effect allowed for easy detachment through shape recovery, and different adhesive performances at different temperatures are achieved by programming various Tg values. Moreover, the simple manufacturing process of the SMP dry adhesive is confirmed to be suitable for continuous fabrication processes based on roll-to-roll methods.

Laminin as a Key Extracellular Matrix for Proliferation, Differentiation, and Maturation of Porcine Muscle Stem Cell Cultivation.

Extracellular matrix (ECM) proteins play a crucial role in culturing muscle stem cells (MuSCs). However, there is a lack of extensive research on how each of these proteins influences proliferation and differentiation of MuSCs from livestock animals. Therefore, we investigated the effects of various ECM coatings-collagen, fibronectin, gelatin, and laminin-on the proliferation, differentiation, and maturation of porcine MuSCs. Porcine MuSCs, isolated from 14-day-old Berkshire piglets, were cultured on ECM-coated plates, undergoing three days of proliferation followed by three days of differentiation. MuSCs on laminin showed higher proliferation rate than others (p<0.05). There was no significant difference in the mRNA expression levels of PAX7, MYF5, and MYOD among MuSCs on laminin, collagen, and fibronectin (p>0.05). During the differentiation period, MuSCs cultured on laminin exhibited a significantly higher differentiation rate, resulting in thicker myotubes compared to those on other ECMs (p<0.05). Also, MuSCs on laminin showed higher expression of mRNA related with maturated muscle fiber such as MYH1 and MYH4 corresponding to muscle fiber type IIx and muscle fiber type IIb, respectively, compared with MuSCs on other ECM coatings (p<0.05). In summary, our comparison of ECMs revealed that laminin significantly enhances MuSC proliferation and differentiation, outperforming other ECMs. Specifically, muscle fibers cultured on laminin exhibited a more mature phenotype. These findings underscore laminin's potential to advance in vitro muscle research and cultured meat production, highlighting its role in supporting rapid cell proliferation, higher differentiation rates, and the development of mature muscle fibers.

A framework for inferring and analyzing pharmacotherapy treatment patterns.

BACKGROUND: To discover pharmacotherapy prescription patterns and their statistical associations with outcomes through a clinical pathway inference framework applied to real-world data. METHODS: We apply machine learning steps in our framework using a 2006 to 2020 cohort of veterans with major depressive disorder (MDD). Outpatient antidepressant pharmacy fills, dispensed inpatient antidepressant medications, emergency department visits, self-harm, and all-cause mortality data were extracted from the Department of Veterans Affairs Corporate Data Warehouse. RESULTS: Our MDD cohort consisted of 252,179 individuals. During the study period there were 98,417 emergency department visits, 1,016 cases of self-harm, and 1,507 deaths from all causes. The top ten prescription patterns accounted for 69.3% of the data for individuals starting antidepressants at the fluoxetine equivalent of 20-39 mg. Additionally, we found associations between outcomes and dosage change. CONCLUSIONS: For 252,179 Veterans who served in Iraq and Afghanistan with subsequent MDD noted in their electronic medical records, we documented and described the major pharmacotherapy prescription patterns implemented by Veterans Health Administration providers. Ten patterns accounted for almost 70% of the data. Associations between antidepressant usage and outcomes in observational data may be confounded. The low numbers of adverse events, especially those associated with all-cause mortality, make our calculations imprecise. Furthermore, our outcomes are also indications for both disease and treatment. Despite these limitations, we demonstrate the usefulness of our framework in providing operational insight into clinical practice, and our results underscore the need for increased monitoring during critical points of treatment.

2D MOFs and Zeolites for Composite Membrane and Gas Separation Applications: A Brief Review.

Commercial membranes have predominantly been fabricated from polymers due to their economic viability and processability. This choice offers significant advantages in energy efficiency, cost-effectiveness, and operational simplicity compared to conventional separation techniques like distillation. However, polymeric membranes inherently exhibit a trade-off between their permeability and selectivity, which is summarized in the Robeson upper bound. To potentially surpass these limitations, mixed-matrix membranes (MMMs) can be an alternative solution, which can be constructed by combining polymers with inorganic additives such as metal-organic frameworks (MOFs) and zeolites. Incorporating high-aspect-ratio fillers like MOF nanosheets and zeolite nanosheets is of significant importance. This incorporation not only enhances the efficiency of separation processes but also reinforces the mechanical robustness of the membranes. We outline synthesis techniques for producing two-dimensional (2D) crystals (including nanocrystals with high aspect ratio) and provide examples of their integration into membranes to customize separation performances. Moreover, we propose a potential trajectory for research in the area of high-aspect-ratio materials-based MMMs, supported by a mathematical-model-based performance prediction.

Metabolomic changes in culture media with varying passage numbers of pig muscle stem cell culture for cultured meat production.

Selecting the primary cells in an optimal state for cultured meat production is a crucial challenge in commercializing cultured meat. We investigated the metabolomic changes in culture media according to passage numbers for indirectly assessing the state of primary cells. Pig skeletal muscle stem cells (PSCs) harvested from the biceps femoris muscles of 7-d-old crossbred pigs (Landrace × Yorkshire × Duroc, LYD) were used for cell characterization. Fresh media (FM) and spent media (SM) of PSCs during passages 1 to 3 in vitro culture were prepared for metabolomics analysis. SM was collected on the third day of proliferation for each passage of PSCs. Cell characterization analysis revealed that the proliferation rate was highest at passage 2; however, a significant loss of expression of myogenic marker genes was observed at passage 3. Based on metabolomic profiles of culture media, FM and SM groups (SM1, SM2, and SM3) were clearly separated by partial least squares-discriminant analysis. A total of seven differentially abundant metabolites (DAMs) were identified from FM and SM for each passage, based on the following criteria: P < 0.05, fold change > 1.5 or < 0.66, and a variable importance in projection score > 1.5. All seven DAMs and their interconnected metabolites might be primarily used as substrates for energy production and most of them were relatively abundant in SM3. Among the seven DAMs, the three potential biomarkers (γ-glutamyl-L-leucine, cytosine, and ketoleucine), which showed significant changes exclusively in SM3, each had an area under the curve value of 1. Therefore, monitoring the levels of these key metabolites in culture media could serve as a quality control measure for cultured meat production by enabling the indirect detection of suboptimal PSCs based on their proliferation ability.

Unravelling the main mechanism responsible for nocturnal CO2 uptake by dryland soils.

Soil respiration, or CO2 efflux from soil, is a crucial component of the terrestrial carbon cycle in climate models. Contrastingly, many dryland soils absorb atmospheric CO2 at night, but the exact mechanisms driving this uptake are actively debated. Here we used a mechanistic model with heuristic approaches to unravel the underlying processes of the observed patterns of soil-atmosphere CO2 fluxes. We show that the temperature drop during nighttime is the main driver of CO2 uptake by increasing CO2 solubility and local water pH of a thin water film on soil particle surfaces, providing favourable conditions for carbonate precipitation. Our data demonstrate that the nocturnal inorganic carbon absorption is a common soil process, but often offset by biological CO2 production. The uptake rates can be impacted by different successional stages of biocrusts that consume or produce CO2 and modify the pH of the soil water film, which can be maintained by non-rainfall water inputs, such as pore space condensation. Annual estimates of nocturnal carbon uptake, based on in situ continuous measurements at the soil level in drylands are still very scarce, but fluxes of up to several tens of g C m-2 y-1 have been reported, potentially accounting for a considerable fraction of the global residual terrestrial carbon sink.

Artificial intelligence predicts undiagnosed atrial fibrillation in patients with embolic stroke of undetermined source using sinus rhythm electrocardiograms.

BACKGROUND: Artificial intelligence (AI)-enabled sinus rhythm (SR) electrocardiogram (ECG) interpretation can aid in identifying undiagnosed paroxysmal atrial fibrillation (AF) in patients with embolic stroke of undetermined source (ESUS). OBJECTIVE: The purpose of this study was to assess the efficacy of an AI model in identifying AF based on SR ECGs in patients with ESUS. METHODS: A transformer-based vision AI model was developed using 737,815 SR ECGs from patients with and without AF to detect current paroxysmal AF or predict the future development of AF within a 2-year period. Probability of AF was calculated from baseline SR ECGs using this algorithm. Its diagnostic performance was further tested in a cohort of 352 ESUS patients from 4 tertiary hospitals, all of whom were monitored using an insertable cardiac monitor (ICM) for AF surveillance. RESULTS: Over 25.1-month follow-up, AF episodes lasting ≥1 hour were identified in 58 patients (14.4%) using ICMs. In the receiver operating curve (ROC) analysis, the area under the curve for the AI algorithm to identify AF ≥1 hour was 0.806, which improved to 0.880 after integrating the clinical parameters into the model. The AI algorithm exhibited greater accuracy in identifying longer AF episodes (ROC for AF ≥12 hours: 0.837, for AF ≥24 hours: 0.879) and a temporal trend indicating that the AI-based AF risk score increased as the ECG recording approached the AF onset (P for trend <.0001). CONCLUSIONS: Our AI model demonstrated excellent diagnostic performance in predicting AF in patients with ESUS, potentially enhancing patient prognosis through timely intervention and secondary prevention of ischemic stroke in ESUS cohorts.