共找到 20 条结果
Retinoic acid (RA), a bioactive metabolite of vitamin A, plays roles in early embryogenesis and hematopoietic development. However, its precise function in directing the hematopoietic lineage outcomes of human pluripotent stem cells (hPSCs) remains unclear. Here, we uncovered a distinct, stage-specific role for RA as a lineage-specifying modulator during late-stage hematopoietic differentiation, rather than as a promoter of hematopoietic progenitor generation. Using a stepwise hPSC differentiation system, we demonstrated that RA exerted minimal or inhibitory effects when applied during early mesoderm or hemogenic endothelial stages. In contrast, RA treatment during days 13-15 significantly enhanced progenitor maturation, proliferation, and functional output. Notably, RA acted cooperatively with external cytokines to modulate lineage fate. In the presence of erythropoietin (EPO), RA strongly promoted erythroid differentiation by activating EPOR signaling and upregulating erythroid transcriptional programs, including GATA1, KLF1, and globin gene expression. Conversely, under GM-CSF/M-CSF stimulation, RA biased progenitor differentiation toward macrophages, consistent with its role as an amplifier of the prevailing cytokine-directed lineage fate rather than an independent suppressor of erythropoiesis. These effects were highly dose- and context-dependent, with low-dose RA optimally enhancing lineage bias without cytotoxicity. Importantly, RA modulated the transcriptional and proliferative dynamics of committed progenitors. Taken together, our findings reveal a previously unrecognized role of RA as a versatile and tunable modulator of hematopoietic lineage fate that offers a novel strategy for in vitro blood cell engineering. This study advances approaches for lineage-specific blood production relevant to disease modeling, drug screening, and regenerative medicine.
With the growing use of AI-powered conversational technology, people with dementia may benefit from these tools. Finetuning and adapting such systems require an understanding of the language use of people with dementia and communication patterns. Daily conversations with 120 Korean older adults at various stages of dementia were collected over a two-year period and transcribed. Text mining techniques were applied to identify language patterns and latent meanings, including Term Frequency (TF), Term Frequency-Inverse Document Frequency (TF-IDF), co-occurrence network and concordance analyses, and topic modeling using both Latent Dirichlet Allocation (LDA) and BERTopic. TF/TF-IDF analyses of 6,989 speech segments by people with dementia revealed frequent use of pronouns and vague demonstratives. Word pairs and in-context usage of frequently used words were closely tied to their lived experiences. Topic modeling identified five themes via LDA and nine key topics from 29 BERTopic subcategories, covering hardship, family and relationships, emotions, reminiscence, and daily pleasures. BERTopic also captured health-related needs, identity, and agency, offering insight into the inner world of people with dementia. Despite declining cognitive abilities, people with dementia experience a wide range of life events and emotions. This study highlights the need for language-based support to help them retain agency, rather than be viewed solely as dependent or in need of care. Furthermore, the words and themes from this study could serve as valuable input for developing AI-powered communication algorithms, enabling more dementia-friendly and engaging conversations with people with dementia.
Electron spin resonance in scanning tunneling microscopy enabled the study of electronic transitions of magnetic impurities on surfaces at the atomic scale. This ESR-STM technique allows to spectroscopically probe and coherently manipulate spins using an all-electrical method without oscillating external magnetic driving fields. Here, we aim to review recent advancements in ESR-STM. We will discuss possible fundamental mechanisms by which the electric field drives spin resonance based on Heisenberg exchange, Kondo scattering, and Anderson impurity models. We validate theoretical predictions against experimental observations, to understand how electronic correlations, spin exchange, and many-body effects manifest in ESR-STM signals. After reviewing coherent spin control in the STM junction, we discuss potential applications of the ESR-STM method for coherent multi-spin control which enables multiple-qubit operations. Finally, we address recent developments in coupled electron-nuclear spin systems, including hyperfine-resolved ESR spectroscopy, and the driving and polarization of nuclear spins in ESR-STM.
In recent years, crystalline porous zeolite materials have been extensively investigated owing to their distinct characteristics, such as high surface area, pore volume, size, and thermal stability. In this work, 500 nm to 1 μm sized thermally stable cobalt-supported zeolite 4 A was developed via impregnation method, employed as an efficient and reusable catalyst in synthesizing novel thiosemicarbazide and 1,2,4 triazole derivatives using Co-Zeolite 4 A in ethanol: water (1:1) as a green solvent, and found most active among Ni and Fe-supported Zeolite 4 A, with 89% yield. FT-IR, XRD, and BET data revealed characteristic metal-support interactions, crystalline domains, and reduction in surface area (10.1 to 9.4 m2/g), confirming successful cobalt incorporation and governing its catalytic activity. The catalyst has maintained its efficacy over five reaction cycles. The anticancer activity was meticulously evaluated with most potent 3,5-bis trifluoromethyl 1,2,4 triazole 4c, displayed IC50 8.74 µM and 29.51 µM against HeLa and PC3 cancer cell lines. In silico docking indicated that compound 4c strongly interacted with the tumor suppressor protein p21, and DFT studies confirmed the chemical reactivity of compounds 3a-3c. This synthetic approach offers high yields, faster reaction times, robust and reusable catalyst, and environmentally sustainable platform for organic transformations with therapeutic potential.
Social content and thinking about information in relation to the self may attenuate age-related declines in associative memory. However, it's unclear whether these strategies are similarly effective across cultures due to potential differences in the prioritization of social and self-relevant information. The present study compared Taiwanese and American younger and older adults on associative memory for social information across two encoding conditions. Specifically, they related object-scene image pairs containing varying levels of social information (i.e. none, low and high) to themselves (i.e. self-referencing) or to a distant-other (i.e. other-referencing). We replicated a prior finding in a new sample of Taiwanese participants in which other-referencing (vs self-referencing) enhanced older adults' memory for high social trials, relative to younger adults. In contrast, this pattern did not emerge for Americans; older American participants' memory performance was relatively consistent across self- vs. other-referencing and the level of social information, and there was no interaction with age. These findings suggest that cultural differences in memory for high social (relative to low and nonsocial) information emerge for older, but not younger, adults, particularly when participants are asked to think about another person. Therefore, culture may influence the effectiveness of strategies to reduce age-related associative memory impairments.
Soft tissue grafts are commonly used in periodontal surgery around teeth and implants. However, few studies have examined donor-site pain through patient-reported outcomes following soft tissue grafting. This study aimed to minimize donor-site pain during autologous gingival grafting using a three-dimensionally (3D)-printed stent and evaluate its effectiveness from a patient-centered perspective. In this randomized controlled trial, 32 patients requiring autogenous gingival grafts were equally allocated to the control group (Omnivac stent) or the test group (3D-printed stent). Patient-reported outcomes were assessed using a visual analog scale (VAS) and the Oral Health Impact Profile-14 (OHIP-14) questionnaire on the day of surgery and at 1, 7, and 14 days postoperatively. Data from 16 control and 14 test patients were analyzed (two patients were lost to follow-up). The test group reported significantly lower VAS scores than the control group. Despite a lack of statistically significant temporal differences, marked reduction was observed on the day of surgery and on postoperative day 7. OHIP-14 scores were significantly lower in the test group than in the control group across all time points. Analysis of the individual OHIP-14 items at different intervals revealed significant reductions in discomfort-specifically in pronunciation, taste, mastication, and daily activities-when using 3D-printed stents. This study applied 3D-printed stents to reduce donor-site discomfort after autogenous soft tissue grafting and evaluated their effectiveness using patient-reported outcomes. Although 3D-printed stents may not directly reduce postoperative pain, they help alleviate discomfort during routine oral functions and daily activities. This study focuses on the free gingival graft, one of the most frequently employed soft tissue grafts in periodontal practice. While autogenous gingival grafting on the palatal side is currently considered the gold standard, it causes additional pain and discomfort in patients. This study demonstrates that three-dimensional technology can be used to improve stent design to reduce patient pain and discomfort. This randomized controlled trial demonstrates that compared with a conventional stent, a customized 3D-printed palatal stent significantly improves patient-reported comfort and oral function after autologous gingival grafting, thereby enhancing postoperative quality of life.
Multiple system atrophy (MSA) is a clinically heterogeneous disorder. Conventional motor phenotype-based classification provides limited prognostic information. Previously, we identified data-driven subtypes of early-stage MSA using a latent class analysis (LCA) that incorporated both motor and non-motor features. However, the prognostic significance of these subtypes remains unclear. We analyzed the survival outcomes in a previously reported cohort of 61 patients with probable or possible MSA enrolled within three years of motor symptom onset. Patients were classified according to LCA-derived subtypes and dichotomized into an extensive or restricted dysautonomia group based on shared autonomic profiles. Overall survival was assessed using Kaplan-Meier analysis and compared using the log-rank test. Cox proportional hazards models were used to estimate hazard ratios (HRs) with progressive adjustments for the age at onset, sex, baseline disease severity (Unified Multiple System Atrophy Rating Scale, UMSARS Part I), and disease duration at enrollment. Survival analysis was performed in 60 patients, 47 of whom died by the end of follow-up. The extensive dysautonomia group showed a significantly shorter median survival than the restricted group (6.0 vs. 7.0 years; log-rank p = 0.008). After adjusting for the age at onset and sex, the restricted group had a lower risk of mortality (adjusted HR: 0.538, 95% CI 0.290-0.996; p = 0.049). This association was attenuated after additional adjustments for the baseline disease severity (UMSARS Part I) and disease duration. Data-driven subtypes defined by early symptom patterns correspond to clinically meaningful survival differences in patients with MSA. Extensive dysautonomia reflects a more malignant, globally severe phenotype than isolated autonomic involvement, highlighting the prognostic relevance of incorporating non-motor features into early MSA classification.
In this work, two coumarin-linked Schiff bases, Cou-NEt and Cou-Nap, together with their Zn(II)-derived products, were synthesized and investigated by spectroscopic, computational, and biological methods. Structural characterization revealed distinct Zn(II)-dependent outcomes: Cou-Nap afforded the expected Zn(II) complex, whereas reaction of Cou-NEt with ZnCl2 induced imine cleavage to yield an amine-derived Zn(II) species. All compounds showed similar visible absorption bands, while Zn(II) incorporation modified the emission response and enhanced fluorescence, most prominently for Zn-Nap. TD-DFT calculations indicated predominantly ligand-centered excited states and suggested that the weak fluorescence of Cou-NEt is associated with access to a red-shifted keto-like excited state, whereas Zn(II) coordination suppresses nonradiative deactivation through chelation-induced rigidification. DNA-binding experiments revealed moderate affinity toward ctDNA, consistent with non-covalent interaction. In vitro antiproliferative assays against MCF-7, HepG2, and A549 cells showed that Cou-NEt and Zn-Nap were the most active compounds in the present series. Molecular docking and molecular dynamics simulations using the human mTORΔN-mLST8 complex supported favorable binding profiles for these two compounds, while confocal imaging in HeLa cells demonstrated efficient cellular uptake of the Zn(II) species. Overall, the results show that Zn(II) coordination can tune the photophysical and anticancer-related behavior of coumarin-based Schiff-base systems in a ligand-dependent manner, while further structural optimization is required to improve antiproliferative potency.
Neurodegenerative disorders such as Parkinson's disease are closely associated with dysregulated activity of monoamine oxidase-B (MAO-B), which leads to dopamine depletion and oxidative stress. Despite the availability of numerous monoamine oxidase (MAO) inhibitors on the market, their irreversibility and associated side effects necessitate the development of more effective and reversible MAO-B inhibitors. In this study, a series of twenty-one indole-based derivatives (PSH1-PSH21), collectively designated as PSH, was synthesised and evaluated for inhibitory activity against MAO isoforms. Most synthesised compounds showed higher inhibitory activity toward MAO-B than MAO-A, indicating selectivity for MAO-B. Among the PSH derivatives, PSH18 exhibited the most potent MAO-B inhibitory activity (IC50 = 0.95 ± 0.02 µM), followed by PSH6 (IC50 = 1.79 ± 0.40 µM) and PSH2 (IC50 = 1.96 ± 0.08 µM). Compound PSH18 showed the highest selectivity index value of 42.11, followed by PSH6 (22.35) and PSH2 (20.41). Additionally, PSH18 was confirmed to be a competitive and reversible inhibitor of MAO-B with an inhibition constant value of 0.89 ± 0.035 µM. Notably, PSH18 exhibited good permeability across the blood-brain barrier in parallel artificial membrane permeability assay experiments, along with acceptable absorption, distribution, metabolism, excretion, and toxicity parameters predicted through in silico modelling, suggesting its potential as a central nervous system-targeted molecule. Molecular docking and a 200 ns molecular dynamics simulation demonstrated stable binding of the ligand to the MAO-B active-site pocket, driven by hydrophobic and π-π stacking interactions with key amino acid residues lining the aromatic cage. Moreover, the calculated binding energy indicated strong ligand-protein interactions. Overall, these results indicate that the PSH scaffold could serve as a promising lead structure for the development of potent and selective MAO-B inhibitors. These compounds may have therapeutic potential for the treatment of neurodegenerative diseases.
This study investigated the coordinated molecular reorganization of faba bean protein isolate (FBPI) induced by ratio controlled dual polyphenol complexation. Gallic acid (GA) and rutin were selected as model polyphenols with different molecular sizes and interaction characteristics. Their distinct effects on protein structure were further examined in relation to interfacial assembly and Pickering emulsion stability. GA promoted structural reorganization, reducing particle size (207.5 to 194.9 nm) and increasing solubility (89.4 to 96.2%), whereas increasing rutin proportion (≥ 0.8:1.0) enhanced intermolecular associations and structural relaxation. Spectroscopic and thermodynamic analyses revealed that hydrogen bonding and van der Waals interactions governed the formation of protein polyphenol complexes, leading to localized conformational rearrangement and increased accessibility of reactive residues. The optimal rutin:GA ratio of 0.4:1.0 produced complexes with superior interfacial performance, exhibiting the highest adsorbed protein fraction (63.76%), high interfacial concentration (11.50 mg/m2), and the lowest Turbiscan Stability Index (7.6). Interfacial adsorption behavior and microscopic observations confirmed the formation of cohesive protein layers and reduced droplet coalescence. These findings demonstrate that dual polyphenol complexation provides an effective strategy for regulating protein structure and stabilizing particulate interfacial films, offering potential for the development of plant protein-based Pickering emulsions.
Autoimmune features are increasingly recognized in myelodysplastic syndromes (MDS), but their clinical and molecular significance remains unclear. We aimed to characterize the clinicopathologic, cytogenetic, molecular, and prognostic features of MDS according to autoimmune status. We retrospectively analyzed 163 patients newly diagnosed with MDS classified into 3 groups: MDS with autoimmune disease (MDS-AID; n = 27), MDS with laboratory evidence of autoimmunity only (MDS-ALE; n = 15), and MDS without autoimmune features (MDS-control; n = 121). Clinical characteristics, genomic alterations, and survival outcomes were compared among groups. Patients with MDS-AID were younger, more often female, and were more frequently classified as lower-risk MDS subtypes, with fewer tier 1 or 2 mutations. Within lower-risk MDS subtypes, trisomy 8 and 9 were more frequent in MDS-AID than in MDS-control (P < .001 and P = .047, respectively), and MDS-ALE was characterized by an increased frequency of ASXL1 mutations (P = .01). Survival analyses showed no significant differences in overall survival or progression-free survival between MDS-AID or MDS-ALE and MDS-control in Cox proportional hazards models. In multivariable analyses, the IPSS-M score was the only variable independently associated with survival outcomes. These findings suggest that MDS with autoimmune features exhibit clinical and molecular heterogeneity, whereas IPSS-M, but not autoimmune status, was independently associated with survival in this cohort.
Activation of endothelial Tie2 signaling has emerged as a potential therapeutic strategy for ameliorating vascular abnormalities and hyperpermeability in ocular diseases. Early therapeutic strategies have focused on inhibiting angiopoietin-2 (Ang2), an antagonistic ligand of Tie2, thereby indirectly promoting Tie2 activation. However, accumulating evidence indicates that indirect Tie2 activation via Ang2 blockade is insufficient for enhanced therapeutic efficacy, underscoring the need for Tie2 agonists. In addition, since it is still necessary to inhibit neovascularization induced by vascular endothelial growth factor (VEGF), appropriate therapeutic efficacy could be achieved by direct Tie2 agonism and VEGF neutralization. An affinity-matured Tie2-activating antibody, MT-101, was identified by phage display panning using a complementarity-determining region-targeted mutagenesis library. Tie2xVEGF bispecific antibodies were generated with MT-101 fused to five different anti-VEGF modules, and their functionality in Tie2 and VEGF signaling was compared using endothelial cells. The therapeutic efficacy of the bispecific antibody fusion MT-103, comprising MT-101 and VEGFR domains, was evaluated in mouse oxygen-induced retinopathy (OIR) and laser-induced choroidal neovascularization (LI-CNV) models compared with anti-VEGF agent Aflibercept or Ang2xVEGF bispecific antibody. MT-101 activated the Tie2 signaling pathway, including AKT-eNOS and ERK cascades, and exhibited efficacy comparable to that of Aflibercept in OIR and LI-CNV models. By directly comparing five different Tie2xVEGF bispecifics, we selected the most potent construct, MT-103, generated by fusing VEGFR1/2 domains to MT-101. MT-103 demonstrated approximately four- and five-fold greater potency than the Ang2xVEGF bispecific antibody in inhibiting VEGF signaling and reducing permeability, respectively, in retinal endothelial cells. MT-103 further demonstrated improved efficacy, reducing neovascularization by 14% compared with Aflibercept in the LI-CNV model and suppressing vascular leakage by 20% compared with Ang2xVEGF bispecific antibody in the OIR model. Moreover, MT-103 elicited robust Tie2 activation and vessel stabilization by enhancing pericyte coverage relative to Ang2xVEGF bispecific antibody. These findings demonstrate that a therapeutic strategy combining direct Tie2 activation and VEGF blockade may provide improved therapeutic potential compared to neutralizing VEGF alone or Ang2 and VEGF, representing a promising therapeutic strategy that warrants further validation for the treatment of various ocular diseases.
Inflammation is a metabolically intensive and tightly regulated process, driven primarily by innate immune cells. Cellular metabolism actively instructs immune signaling and cell fate decisions. Bioenergetic pathways, including glycolysis, mitochondrial respiration, and the tricarboxylic acid cycle, reshape cytokine production and regulate inflammatory cell death pathways. In this review, we synthesize emerging evidence on how metabolic intermediates and pathways regulate inflammasome signaling and the execution of diverse inflammatory cell death modalities, including pyroptosis, necroptosis, PANoptosis, and ferroptosis. We propose that metabolic inputs-including redox balance, mitochondrial dynamics, and lipid modifications-constitute an interconnected metabolic regulatory network that determines the threshold and outcome of inflammatory signaling. This framework offers new insights into immunometabolic dysregulation and therapeutic strategies in inflammatory, infectious, and neoplastic diseases.
Smoking cessation reduces hip fracture risk; however, subsequent weight loss may offset this benefit. In a nationwide cohort, weight loss after smoking cessation was associated with increased hip fracture risk. Maintaining stable body weight after quitting smoking may be important for fracture prevention. Smoking cessation reduces fracture risk but is often accompanied by changes in body weight that may influence bone health. However, it remains unclear whether weight change after smoking cessation is associated with hip fracture risk. We therefore examined the association between post-cessation weight change and the incidence of hip fractures. This retrospective cohort study used data from the Korean National Health Insurance Service. Adults aged ≥ 40 years who underwent consecutive biennial health examinations in 2007 and 2009 were categorized as nonsmokers, quitters, or current smokers based on their smoking status. Quitters were further stratified by weight change between two examinations: loss (> 5% decrease), weight maintenance (± 5%), or weight gain (> 5% increase). Participants were followed from 2010 through 2018 for incident hip fractures. Hazard ratios (HRs) and 95% confidence intervals (CIs) were estimated using multivariable Cox proportional hazards models. Among 913,805 participants (672,858 nonsmokers; 34,143 quitters; 206,804 current smokers), 6,001 incident hip fractures occurred during a mean follow-up of 8.1 years. Compared with nonsmokers, current smokers had a higher risk of hip fracture (aHR = 1.700, 95% CI 1.557-1.856). Quitters also showed increased risks across all weight-change categories: weight loss (aHR = 1.878, 95% CI 1.164-3.031), weight maintenance (aHR = 1.461, 95% CI 1.188-1.797), and weight gain (aHR = 1.611, 95% CI 1.151-2.255). Quitters who maintained or gained weight had numerically lower risk estimates than current smokers, whereas those who lost weight had a higher point estimate; however, confidence intervals overlapped across groups. In analyses of the entire cohort, both weight loss (aHR = 1.425, 95% CI 1.336-1.519) and weight gain (aHR = 1.150, 95% CI 1.064-1.242) were associated with higher hip fracture risk compared with weight maintenance. Smoking cessation was associated with a lower risk of hip fracture than continued smoking, although the risk remained higher than in nonsmokers. Weight loss after smoking cessation was associated with higher fracture risk, while maintaining stable body weight appeared to be associated with more favorable outcomes. These findings suggest that maintaining a stable body weight after smoking cessation may be important for hip fracture prevention.
暂无摘要（点击查看详情）
The amount of biologic data produced by biomedical research has increased significantly in both volume and complexity in recent years. Advances in computational power have increasingly enabled the use of machine learning (ML) to analyze and predict patterns from large-scale, complex biologic datasets. In nonhuman primate (NHP) infectious disease models, such high-dimensional datasets containing a large number of features are often obtained by next-generation sequencing-based multiomics and immunologic analyses. As a result, ML is particularly valuable for effective analysis and predictive modeling in this context. This review demonstrates that the application of ML in NHP infectious disease models has increased over time. Ensemble methods, particularly random forest, have emerged as the most frequently used algorithms, followed by regression and clustering approaches. Logistic regression and hierarchical clustering were the most commonly applied regression and clustering methods, respectively. These techniques are primarily used for vaccine response prediction, biomarker discovery, disease progression analysis, gene and pathway identification, and immune response characterization. Despite this increasing trend, the overall adoption of ML in NHP infectious disease models remains limited, which may reflect gaps in familiarity and computational expertise among researchers. Recent advances in generative artificial intelligence and user-friendly analytical platforms are expected to improve accessibility and promote broader adoption. This review aims to support understanding and facilitate wider application of ML in NHP infectious disease models.
Echo-planar imaging (EPI), commonly employed in functional MRI (fMRI), is highly susceptible to magnetic field inhomogeneities, leading to pronounced geometric distortions in reconstructed images. These distortions can result in substantial structural discrepancies between EPI and anatomical images acquired using the magnetization-prepared 2 rapid acquisition gradient echoes (MP2RAGE) method, thereby making it challenging to achieve accurate co-registration and subsequent localization of functional mapping. This issue can be effectively addressed by employing an anatomical imaging sequence that exhibits distortion profiles identical to those in EPI, referred to here as MP2EPI (magnetization-prepared 2 EPI). While this approach enables effortless co-registration with functional scans, it also introduces geometric distortions into the anatomical reference imaging, which limits its utility for analyses that rely on morphometric measurements or atlas-based segmentation. Distortion in MP2EPI can be corrected using additional data acquired with the reversed phase-encoding (PE) direction, which, however, significantly increases total acquisition time. To overcome this limitation, this work presents a novel MP2EPI sequence that simultaneously acquires reversed PE data within a single MP2EPI acquisition, without increasing the overall scan time. The primary focus of the current work is the technical implementation and validation of this sequence in the context of submillimeter fMRI at 7T.
Accurate pre-harvest yield estimation of underground bulb crops such as onion and garlic is important for precision agriculture, harvest planning, and food-security-oriented decision-making. However, their harvestable organs develop below ground and cannot be directly observed using conventional remote sensing methods. This study aimed to develop a non-destructive yield estimation framework by integrating UAV-based hyperspectral imaging with hybrid machine learning models. Field experiments were conducted in Muan-gun, Korea, using onion and garlic as representative underground bulb crops. UAV-based hyperspectral images, crop growth traits, and destructive live bulb weight measurements were collected during the growing period. Hyperspectral images were processed through geometric correction, radiometric correction, and Savitzky-Golay spectral smoothing. Three dimensionality reduction methods, including genetic algorithm (GA), principal component analysis (PCA), and clustering, were used to reduce spectral redundancy. Five prediction models, including random forest (RF), XGBoost, partial least squares regression (PLSR), multilayer perceptron (MLP), and residual network (ResNet), were then evaluated for live bulb weight prediction. Significant spectral differences were observed in the 550-680 nm and 730-800 nm bands, which were closely associated with crop yield and below-ground bulb development. GA was the most effective feature selection method for extracting yield-related spectral bands. For onion yield prediction, the GA + RF model achieved the highest predictive accuracy, with an R2 of 0.9656 and an NRMSE of 18.55%. For garlic yield prediction, PLSR showed the best performance, with an R2 of 0.9260 and an NRMSE of 27.20%. The proposed UAV-based hyperspectral framework enables accurate, real-time, and non-destructive yield estimation for underground bulb crops. This approach reduces reliance on labor-intensive destructive sampling and provides a practical tool for precision crop monitoring and data-driven agricultural management.