Search Results (1,448)

Oculomotor dysfunction is an eye movement disorder frequently experienced in patients with Parkinson’s disease. Many patients tend to experience visual symptoms, and this can exacerbate cognitive symptoms when visual tasks become more demanding. The purpose of this review is to characterize oculomotor dysfunction in patients with Parkinson’s disease based on the distinct types of ocular motor deficits and their corresponding impact on the patient’s symptoms, visual perception, activities of daily living, and quality-of-life. A systematic literature search was conducted to identify relevant articles. The results from the sensorimotor framework analysis are categorized in five domains: visual-sensory, visual-motor, visual-perceptual, cognitive processing, and psychosocial challenges. The findings suggest that clinical evaluation and neurorehabilitation should move beyond speed-dependent metrics, but focus on specific non-speed-dependent ocular motor deficits. By fostering interdisciplinary collaboration, healthcare professionals can take proactive steps to address the vision-related challenges faced by patients with Parkinson’s disease. Full article

Background/Objectives: Multi-strain Limosilactobacillus fermentum supplementation has been reported to promote weight loss outcomes in free-living conditions, but limited evidence exists on these probiotic strains added to an energy-restricted diet and walking program in overweight adults. Methods: In a double-blind, placebo-controlled, parallel-arm randomized trial, overweight adults (35.2 ± 13.2 years old, 167.6 ± 8.6 cm, 79.9 ± 11.8 kg, 28.4 ± 2.7 kg/m² body mass index, 36.1 ± 6.6% body fat) completed a 12-week weight loss program that included a 500 kcal/day energy deficit and walking 10 k steps/d. Participants ingested one daily capsule containing a three-strain probiotic blend (L. fermentum K7-Lb1, L. fermentum K8-Lb1, L. fermentum K11-Lb3; 6 billion CFU/day) (PRO) or maltodextrin placebo (PLA). Assessments were performed at baseline, week 6, and week 12 and included body composition, resting energy expenditure, substrate utilization, peak oxygen uptake, dietary intake, step counts, blood biomarkers, quality of life, and side effects. Data were analyzed using multivariate and univariate repeated-measures general linear models (GLM), with mean changes from baseline presented alongside 95% confidence intervals. Results: All participants significantly reduced body weight, fat mass, body fat percentage, and waist circumference. At 12 weeks, PRO reduced fat mass more than PL (−2680.7 ± 1276.7 g; p = 0.039). In PRO, android and gynoid fat percentage decreased at 6 weeks (p < 0.001; p = 0.008) and 12 weeks (p = 0.004; p < 0.001), respectively. Visceral adipose tissue mass, volume, and area were lower at 6 weeks and trended lower at 12 weeks. In PRO, bone mineral content and bone mineral area decreased at 12 weeks, while bone mineral density paradoxically increased (0.007 ± 0.003 g/cm²; p = 0.024). Conclusions: During a 12-week weight loss program, supplementation of a multi-strain L. fermentum probiotic significantly reduced body fat and central adiposity. Full article

In China, policy-driven rural renewal projects have transformed many village public spaces, but some renewed sites are still weakly integrated into villagers’ everyday routines. This study asks why some renewed public spaces sustain routine use and low-intensity social interaction, while others remain materially complete but socially weak. The study was conducted in a rural township in Puyang County, Henan Province. Twelve renewed public spaces across several villages were examined through structured spatial audits and 579 resident questionnaires. Five spatial conditions were assessed: visibility, stay support, activity accommodation, interaction-supportive arrangement, and experienced locational convenience. Two behavioral outcomes were used to describe post-renewal vitality: use frequency and social participation. The analysis combines necessary condition analysis (NCA) and fuzzy-set qualitative comparative analysis (fsQCA). NCA is used as a diagnostic tool for identifying upper-limit constraints, while fsQCA is used to identify sufficient combinations of conditions. The results suggest that experienced locational convenience is the clearest bottleneck condition for both outcomes. When a site is difficult to incorporate into residents’ daily walking routines, internal design quality has limited capacity to translate into sustained behavioral use. Among better-located spaces, high vitality is associated with several design configurations. The most stable recurrent pattern combines visibility, stay support, and locational convenience as core conditions, together with either interaction-supportive arrangement or activity accommodation. Low-vitality spaces follow a different logic, being characterized by the simultaneous absence of several supporting conditions rather than by the absence of one isolated feature. The paper therefore proposes a two-step diagnostic logic for rural public-space renewal: first checking whether a site is embedded in everyday mobility and then matching internal spatial conditions with local patterns of use. Full article

Background: Spatial cognition, a skill paramount to survival, is impaired in Parkinson’s disease (PD) but has been little researched. Spatial cognition is utilized during motor–cognitive integration, which impacts daily functioning and quality of life in PD. As PD is a unilateral-onset condition, spatial–cognitive and motor–cognitive ability may differ by side of onset. Spatial cognition is suggested to be modulated by the right hemisphere; thus, we hypothesize to observe worse spatial and motor–cognitive performance by people with left-onset PD (LOPD) than right-onset PD (ROPD). Methods: 216 participants with PD were recruited (LOPD = 107; M = 62; mean age = 69.80 ± 8.5). Spatial outcomes were collected via the body position spatial task (BPST), Reverse Corsi Blocks, and visuospatial items of the Montreal Cognitive Assessment (MoCA); motor–cognitive outcomes were collected by a Trails test, a Four Square Step Test (FSST), and a Timed Up and Go test. An independent t-test and the Mann–Whitney U test compared outcome variables between onset groups. Results: No significant differences were found between onset groups. Exploratory subgroup analyses revealed differences. Significantly worse performance by LOPD in single- and dual-task TUG was found within people with bilateral symptoms and postural instability (Hoehn & Yahr stage, >2; LOPD, N = 33; single, p = 0.001; dual, p = 0.021) and worse performance in single-task TUG in people with MoCA < 18 (LOPD, N = 5; single, p = 0.036) and people with freezing of gait (FOGQ, >0; LOPD, N = 14, p = 0.048). Significantly larger DTC by LOPD was found within frequent freezers (FOGQ, >3; LOPD, N = 9; p = 0.003). Conclusions: LOPD may tend to perform worse in motor–cognitive tasks among subgroups of those with more severe symptoms, i.e., those at later stages of disease. These findings may have implications for prognoses of those with LOPD versus ROPD and suggest that those with LOPD may have worse long-term outcomes in spatial cognition and motor–cognitive integration. Full article

Objective: This study aimed to evaluate the effects of 12 weeks of Astragalus membranaceus saponins (AMS) supplementation on functional performance, knee joint mobility, self-reported outcomes, and biomarkers of inflammation and cartilage in healthy subjects with knee discomfort. Methods: A randomized, double-blind, placebo-controlled trial was conducted in healthy subjects aged 20–70 years with knee discomfort but without clinically diagnosed knee osteoarthritis. Participants were assigned to receive one capsule of AMS or a placebo once daily for 12 weeks. The pre-specified primary endpoints were the SLSD step count and knee ROM; KOOS total score was a key secondary endpoint; serum biomarkers were exploratory. The results included functional performance assessed by the Single Leg Step Down (SLSD) test, knee range of motion (ROM), and self-reported outcomes using the Knee injury and Osteoarthritis Outcome Score (KOOS). Knee ROM was measured with a goniometer and recorded as both active ROM and passive ROM for knee flexion and extension. Serum biomarkers of inflammation (IL-8, IL-1β, MIP-1α), cartilage degradation (CTX-II, COMP, MMP-13, COL2A1), and cartilage synthesis (PIINP) were evaluated at baseline and Week 12. Results: Within the AMS group, SLSD step count increased significantly by 16.83% (Δ = +12.78 steps; p < 0.05) and recovery time decreased significantly by 19.12% (Δ = −108.91 s; p < 0.05) compared with baseline, whereas the placebo group showed smaller, non-significant changes (+4.48 steps and −56.48 s, respectively); however, neither between-group difference in change scores reached statistical significance. Significant improvements in active and passive knee ROM were observed in both flexion and extension (all p < 0.05) within the AMS group, whereas the placebo group showed no significant changes. KOOSs improved significantly in all domains within the AMS group, with the largest gains observed in sport/recreation (+22.23%) and quality of life (+18.38%). In the exploratory biomarker analysis, several inflammation and cartilage-related biomarkers changed after AMS supplementation showed within-group reductions (IL-8, COMP, MMP-13) and PIINP increased. Conclusions: 12 weeks of AMS supplementation was associated with improvements in selected functional, mobility, and outcomes in generally healthy adults with self-reported knee discomfort. AMS was also associated with changes in selected circulating biomarkers related to inflammation and cartilage metabolism. These findings should be interpreted as a preliminary, safe, complementary strategy to support joint health in healthy subjects with knee discomfort. Full article

Background/Objectives: The safety of specific parenteral glucose intake values within the range currently recommended by international guidelines for the early acute phase in preterm neonates has not been established. This study aimed to evaluate whether exceeding a data-driven parenteral dextrose intake threshold during the first week of life is independently associated with hyperglycemia, hypertriglyceridemia, metabolic acidosis, and extrauterine growth restriction (EUGR). Methods: This was a single-center retrospective study involving preterm neonates (gestational age ≤ 34 weeks and/or birth weight ≤ 1500 g) admitted to the Neonatal Intensive Care Unit of Policlinico Umberto I, Rome, between 2015 and 2022. The analysis followed two pre-specified steps: (1) data-driven identification of an exposure threshold by restricted cubic spline logistic regression; (2) multivariable analyses with the dichotomized exposure, adjusting for gestational age, birth weight, enteral nutrition timing, neonatal morbidity, and perinatal compromise. Results: 389 preterm neonates met eligibility. The data-driven inflection point of the spline-derived log-odds curve identified a threshold of 7 g/kg/day. Exceeding this threshold during the first week of life was independently associated with both hyperglycemia (adjusted odds ratio 5.55, 95% confidence interval 2.56 to 12.03; p < 0.001) and hypertriglyceridemia (adjusted odds ratio 4.36, 95% confidence interval 1.41 to 13.45; p = 0.010), but not with metabolic acidosis or with EUGR at 36 weeks postmenstrual age. The divergence in daily parenteral glucose intake between cases and controls was apparent from the second day of life. Conclusions: Exceeding 7 g/kg/day of parenteral dextrose was independently associated with early metabolic complications, but not with growth outcomes. A safety threshold for parenteral glucose may exist within the currently recommended intake range; prospective multicenter studies are needed before clinical recommendations can be drawn. Full article

Accurate volatility forecasting and timely risk early warning are foundational requirements of financial risk management: Value-at-Risk estimates, portfolio risk limits, derivative hedging ratios, and stress-test scenario calibrations all depend on forward-looking volatility signals that remain reliable when market conditions depart from average. This paper develops a public-data causal multiscale wavelet spillover learning (CMWSL) framework that jointly addresses stock-index volatility forecasting and high-volatility early warning under strict walk-forward evaluation. CMWSL integrates three components: a heterogeneous autoregressive (HAR) persistence block as the dominant linear baseline, causal stationary wavelet transform (SWT) summaries that encode within-index multiscale market dynamics, and a cross-index spillover layer that tests whether medium- and long-scale wavelet energy from peer indices carries incremental risk-relevant information. The empirical analysis covers the S&P 500, Nasdaq-100, and Dow Jones Industrial Average over a 2513-step out-of-sample evaluation period from 2016 to 2025, with forecast horizons $h\in \{1,5,10\}$ and OHLC-based volatility targets. All preprocessing, wavelet decomposition, calibration rules, and warning thresholds are re-estimated inside each rolling training window to eliminate look-ahead bias. HAR remains the strongest average model in the main Rogers–Satchell specification, confirming that daily index volatility risk is highly persistence-driven. The multiscale extension delivers statistically significant improvements at longer horizons, in richer public macro-financial information environments, and under the Parkinson target. Clark–West tests detect significant spillover gains in five of nine index–horizon cells (CW $=4.83$, $p<0.001$ for S&P 500 at $h=10$). Critically, tail-conditioned and rolling-window diagnostics show that multiscale and cross-index gains concentrate in upper-volatility regimes and synchronized stress episodes—precisely the conditions in which risk management decisions are most consequential. For market-risk early warning, a logistic classifier built on the same causal feature pipeline delivers the most stable precision–recall performance across all settings, providing an interpretable and operationally auditable alert mechanism suitable for practical risk monitoring. Full article

The hydrological prediction in transboundary river basins is difficult because of their heterogeneous data distribution, restrictions of data sovereignty, and the irregular nature of the discharge pattern. In this paper, Federated Learning (FL) with an LSTM neural network is proposed for next-day predictions of discharge in the Chu–Talas transboundary basin. The basin area belongs to both Kazakhstan and Kyrgyzstan. In an FL scenario, two hydrological stations from the basin were selected as client nodes, representing two different discharge regimes. Station 15125—the Chu main channel is characterized by the highest discharge regime among stations located in the basin, while Station 15233—Merke tributary represents a small catchment with an irregular regime. The federated LSTM model is compared against a centralized LSTM and a local-only LSTM baseline model. The training process is based on nearly three decades of daily discharge measurements. The preprocessing step includes synchronization, lag calculation, and windowing operation. The models are trained using three metrics: root mean square error, mean absolute error, and Nash–Sutcliffe Efficiency, as well as using Monte Carlo Dropout for estimation of the probabilistic uncertainty. The results demonstrate that the federated model demonstrates comparable performance with the centralized one for the Chu main channel. It also improves prediction accuracy for the smaller Merke tributary compared with both centralized and local-only models. These findings show that FL can work effectively with non-IID and heterogeneous hydrological data. The study makes three main contributions: (i) it implements the FedAvg algorithm on transboundary, heterogeneous hydrological data, proving that decentralized optimization can effectively capture autoregressive temporal hydrology without data centralization; (ii) it systematically compares federated, centralized, and local-only models, demonstrating that the federated approach eliminates the scale bias that traditionally neglects smaller, high-variance catchments; and (iii) it utilizes Monte Carlo Dropout to translate deterministic AI outputs into risk-aware probabilistic bounds. Ultimately, the results of this study demonstrate the practical and scientific usefulness of FL in operational water management, as the method presents a privacy-saving means of increasing predictive capacity and enabling risk-based decision-making in transboundary river basins. Full article

Chemical dosing control in wastewater treatment plants (WWTPs) necessitates a dynamic equilibrium between effluent compliance and operational costs, exemplifying a typical multi-objective sequential decision problem. Given the significant operational and compliance risks associated with online trial-and-error methods in full-scale plants, this study introduces a model-based reinforcement learning (MBRL) framework aimed at optimizing dosing. Utilizing 227 historical operation records, we construct a multilayer perceptron (MLP) virtual WWTP that maps 15 process states and 2 dosing actions to 6 effluent indicators, thereby providing a secure training environment for policy development. A composite reward function is formulated to incorporate penalties for effluent quality, constraints for abnormal conditions, and terms related to chemical costs. Three actor–critic algorithms—Soft Actor–Critic (SAC), Twin Delayed Deep Deterministic Policy Gradient (TD3), and Proximal Policy Optimization (PPO)—are trained for 50,000 steps and evaluated over 50 test episodes against both a random baseline and a practically deployed proportional feedforward control baseline. All three reinforcement learning methods yield significantly higher rewards: SAC achieves a mean score of 70.23 (95% CI: [67.94, 72.52]), TD3 scores 70.33 (95% CI: [68.02, 72.64]), and PPO scores 68.78 (95% CI: [66.47, 71.09]), compared to 61.01 (95% CI: [58.23, 63.80]) for proportional control and 45.87 (95% CI: [41.29, 50.45]) for random dosing. Notably, both off-policy agents (SAC and TD3) demonstrate statistically equivalent, state-of-the-art control performance (paired Wilcoxon, p = 0.0631) and converge towards highly economical, low-dosage strategies. This study validates the feasibility of data-driven virtual commissioning for WWTP dosing optimization and supports the advancement of hybrid intelligent control systems that incorporate mechanistic constraints. In addition to algorithm comparisons, this framework can be further refined to serve as a practical, data-driven decision support tool for wastewater treatment plant (WWTP) operators. It enables them to formulate daily chemical dosing plans while adhering to compliance and cost constraints. Full article

The presence of multiple uncertainties in integrated electricity–heat energy systems (E-HIES) poses significant challenges to system dispatch. To achieve an effective balance between economy and robustness, this paper proposes an optimal scheduling method based on fuzzy chance-constrained Information Gap Decision Theory (Fuzzy-IGDT), accounting for uncertainties in wind power output, photovoltaic output, electrical load, and thermal load. The method employs trapezoidal fuzzy numbers to model the four types of uncertain variables and constructs a fuzzy robust model (F-RM) for conservative decision-makers and a fuzzy opportunity model (F-OM) for aggressive decision-makers. An Adaptive Step Ratio (ASR) optimization method is then developed to solve the proposed models. Case studies demonstrate the effectiveness of the proposed methodology. Results show that: compared with conventional IGDT, pure fuzzy and stochastic programming, Fuzzy-IGDT simultaneously optimizes economy, stability and reliability: daily operating cost is reduced by 12.7%, the standard deviation of cost volatility shrinks by 34.5%, and the loss-of-load probability is only 0.3%. Relative to the traditional Weighted Offset Coefficient (WOC) method, ASR directly coordinates the deviation ratios of multiple variables through its step-ratio mechanism, cutting system risk cost by 21.3%, raising solution efficiency by 42%, and improving convergence stability by a factor of 3.8. This research provides new theoretical support and practical tools for optimal scheduling of E-HIES under multiple uncertainties. Full article

This paper deals with the optimal integration of power plants, including a storage device. For such systems, numerous structures are possible, involving different numbers of heat exchangers, and for each of them, optimal operating temperatures need to be found. Moreover, the heat-storage system can be located at different temperature levels, offering another degree of freedom when optimizing the whole system. If process simulators are presently very powerful tools for optimizing complex processes, they need to propose a primary design before any optimization steps. Finite-Dimension Thermodynamics (FDT) could help engineers to propose this primary design, close to the optimal one. To this aim, the FDT method is generalized for power-generation systems including a storage device and any number of heat exchangers. The optimization step consists of maximizing the power generation submitted to the thermodynamics constraints (first and second laws) related to each heat exchanger, power block, and thermal storage system. Two types of heat transfer law are studied and compared: Newton’s law $\left(K\times ∆T\right)$ and phenomenological law issued from thermodynamics of irreversible processes $\left(L\times ∆\left(1/T\right)\right)$). Remarkable results have been found: (i) all the studied structures lead to the Curzon–Ahlborn efficiency when optimized with Newton’s law, (ii) for the same driving source (same high temperature and same power), and without any storage system, the output power production varies as N⁻², N being the number of the heat exchangers, (iii) Charge and discharge times scenarios have a big impact on the optimal operating temperatures and on the resulting daily energy production. Efficiencies of operational plants, including nuclear or solar plants and ORC, are finally compared with the theoretical efficiency found at the maximum power point. This shows that FDT provides a good assessment of the actual efficiency of existing power plants. Full article

Decomposition of nonlinear, nonstationary multicomponent signals remains challenging for existing decomposition strategies, including frequency-based, data-driven, and subspace methods, which can suffer from mode mixing, leakage across components, and unreliable isolation of transients. Motivated by this gap, this study proposes Orthogonal Self-Similarity Decomposition (OSSD), which exploits a self-similarity structure in delay-embedded orbit geometry so that temporal organization, rather than spectrum alone, guides component construction. OSSD-Basic introduces three algorithmic novelties within a single pipeline: (1) an adaptive proxy-correlation band merging on the delay axis, (2) a dominant-component cascade that prevents energy-dominant carriers from masking weaker components, and (3) a double MGS + LS reprojection that collapses the inter-mode orthogonality index to numerical zero, regardless of merging and pruning operations. Synthetic experiments with known ground truth show that OSSD-Basic provides a parsimonious four-mode representation with exact inter-mode orthogonality (OI = 9.4 × 10⁻¹⁸), the highest reconstruction SNR among the evaluated baselines (27.14 dB), and the highest ground-truth diagonal correlation sum (3.038) among the tested methods, while using two fewer modes than EMD, VMD, and SSA. Daily streamflow forecasting on a U.S. Geological Survey discharge record further shows that augmenting OSSD-derived inputs with fractal descriptors and fractional-order differencing features yields progressive accuracy gains over the AR-ANN baseline, with R² improving from 0.855 to 0.915 at one-step-ahead and from 0.388 to 0.699 at four-step-ahead forecasting in the single-input setting, within a single-station case study on USGS 01554000. Overall, OSSD-Basic offers an interpretable multiscale decomposition with guaranteed inter-mode orthogonality and a structured feature pathway for oscillatory–transient mixtures. Full article

Offshore high-water-rate gas wells can often sustain stable production for a considerable period after liquid first appears at the wellhead. Unlike conventional onshore gas wells with relatively low liquid production, these wells can remain in stable production during the middle and late production stages even when the gas velocity in the wellbore has fallen far below the critical value predicted by conventional liquid-carrying criteria. Under such conditions, the wellbore flow pattern commonly shifts from annular mist flow to churn flow and slug flow, and liquid transport becomes governed by a dynamic balance jointly controlled by pressure differential and gas entrainment. As a result, conventional critical liquid-carrying theory alone is no longer sufficient for accurate production state identification. To address this issue, this study proposes a production state identification method for offshore high-water-rate gas wells based on dynamic pressure profile calibration and nodal analysis. In this method, the wellbore pressure profile serves as the key link between outflow capacity and production state evaluation. Measured data from flowing pressure tests are used to calibrate the pressure profile within the selected multiphase flow correlation by introducing two calibration coefficients, namely the liquid holdup calibration coefficient and the two-phase friction calibration coefficient. Gaussian process regression is then applied to model the temporal evolution of the calibration coefficients, generate their fitted trajectories, and predict their values at the next time step. By using the predicted calibration coefficients to recalibrate the pressure profile, dynamic calibration of the wellbore pressure profile is achieved. Field applications to four offshore high-water-rate gas wells show that the calibrated pressure profiles are in closer agreement with the measured pressure points. The accuracy of production-state identification is also significantly improved, and the gas production rates calculated from calibrated nodal analysis are closer to the values reported in daily production records than those obtained before calibration. These results demonstrate that the proposed method effectively improves both wellbore pressure profile prediction and production-state identification for offshore high-water-rate gas wells. The study provides a practical method for production state evaluation and production management of offshore high-water-rate gas wells during the middle and late stages of field development. Full article

Considering the daily importance of vegetables in the human diet, breeders are expected to find faster and more accurate methods of creating new varieties of vegetables. To more precisely meet the demands of vegetable producers and consumers, breeders are increasingly combining hybrid and molecular techniques. The integration of hybrid and molecular breeding represents a logical step towards the development of efficient vegetable breeding strategies. For this purpose, the aim of this review is to point out several representative vegetable species (tomato, pepper, cabbage, lettuce, and cucumber) the possibilities, advantages, and disadvantages of the integration of hybrid and molecular methods of breeding. While conventional breeding techniques are based on selective breeding, mass selection, pure line selection, backcrossing, and hybrid breeding that exploit the effects of heterosis, advanced techniques such as phenomics, molecular markers, genome-wide association studies, and next-generation sequencing facilitate the identification and selection of desirable traits and improve nutritional quality. Breeding a new and promising vegetable cultivar can take 10 to 15 years before it becomes available for commercial production. Molecular techniques definitely represent a faster and more precise part of this mentioned integration. However, classical-hybrid breeding still develops stable, uniform, and marketable varieties without the high costs and significant access of advanced laboratory infrastructure, and without the regulatory barriers that accompany genetic engineering. Full article

Accurate hourly crop evapotranspiration (ETc) estimation is important for data-driven irrigation management support in tropical horticulture, yet existing approaches are constrained by data requirements and an inability to capture multi-scale temporal dynamics. This study proposes a Transformer encoder model for one-step-ahead hourly FAO-56 PM-derived ETc estimation in a durian orchard in Chanthaburi Province, Eastern Thailand, using 36,528 hourly meteorological observations obtained from the Visual Crossing Weather API for the orchard location over four years, with ETc computed from these inputs using the FAO-56 Penman–Monteith equation. The model employs a 168-h (7-day) look-back window, three stacked encoder blocks with multi-head self-attention ($h=8$, $d_{\mathrm{model}}=128$), and five meteorological input features (air temperature, relative humidity, solar radiation, wind speed, and ETc). A SARIMA$(2,1,2){(1,0,0)}_{24}$ model trained on the same dataset served as the statistical baseline. The Transformer achieved an RMSE of 0.0308 mm/h, MAE of 0.0188 mm/h, and $R^2$ of 0.9018 on the 168-h test set, outperforming SARIMA (RMSE = 0.0717, MAE = 0.0593, $R^2$ = 0.4688), representing a 57.0% reduction in RMSE, a 68.3% reduction in MAE, and a 92.4% improvement in $R^2$. The Transformer also achieved a daytime-only RMSE of 0.0414 mm/h vs. 0.0791 mm/h for SARIMA, and a daily cumulative ETc MAE of 0.1599 mm/day vs. 0.5901 mm/day, demonstrating superior accuracy during agronomically critical periods. The Transformer accurately reproduced both the 24-h diurnal cycle and the 7-day weekly pattern of ETc, whereas SARIMA exhibited a damped amplitude response. A recursive 168-h heuristic simulation demonstrated that the model generates physically plausible ETc patterns under an approximated meteorological scenario, suggesting the approach warrants further investigation as a component of future irrigation decision-support research. These results highlight the potential of Transformer-based deep learning for site-specific, proof-of-concept ETc estimation from meteorological inputs in tropical fruit production, pending validation across diverse sites and seasons. Full article