Simple strategy of the use of pharmaceutically functionalized ionic liquids in a new generation of polymer nanocarriers for the combined delivery of ionic p-aminosalicylate and ampicillin.

Single and dual bioactive linear poly(ionic liquid)s (PIL) were synthesized for use as nanocarriers in drug delivery systems (DDS). These PILs were obtained through the (co)polymerization of the choline-based monomeric ionic liquids (MIL) with pharmaceutical anions possessing antibacterial properties, specifically [2-(methacryloyloxy)ethyl]trimethyl-ammonium with ampicillin and p-aminosalicylate (TMAMA/AMP and TMAMA/PAS). The copolymers exhibited varying chain lengths defined by a degree of polymerization (DPn = 122-370), and differing contents of ionic fraction and drugs (TMAMA 61-92 %, AMP 61-93 % and PAS 16-21 %). These parameters were adjustable by the monomer conversion (33-92 %) and the initial ratio of comonomers. In aqueous solution, the polymer particles reached nanosizes, i.e. 190-328 nm for AMP systems and 200-235 nm for AMP/PAS systems. In the release process, the pharmaceutical anions were released through exchange by phosphate anions in PBS at pH 7.4 at 37 °C. Depending on the copolymer composition the release of AMP was attained in 72-100 % (11.1-19.5 µg/mL) within 26 h by the single drug systems, while the dual drug systems released 61-100 % of AMP (14.8-24.7 µg/mL) and 82-100 % of PAS (3.1-4.8 µg/mL) within 72 h. The effectiveness in the drug delivery of the designed TMAMA polymers seems to be promising for future applications in antibiotic therapy and the combined therapy.

The comparisons of non-linear models to describe the growth performance of Lori-Bakhtiari sheep.

BACKGROUND: The study of growth traits is of interest to many animal scientists, regardless of specialization, due to the economic importance of growth rate, mature weight and other related traits. OBJECTIVE: This study aimed to compare six non-linear models for describing the growth of Lori-Bakhtiari sheep. METHODS: In order to collect weight data, 85 lambs (41 males and 44 females) were reared from birth to 140 days of age, and their growth patterns were recorded by measuring their body weight at 10-day intervals. Various mathematical functions, including the negative exponential, Brody, Gompertz, Logistic, Morgan-Mercer-Flodin (MMF) and Weibull, were used to model the relationship between body weight records and age. RESULTS: The results showed that the MMF and Gompertz models provided the best fit to the body weight data, whereas the negative exponential model exhibited the worst fit. In all models, the asymptotic weight of male lambs was higher than females. The research also revealed differences in growth patterns between male and female lambs. Overall, females had a lower absolute growth rate than males, but they reached their peak growth at an earlier period, and their growth rate declined faster. CONCLUSIONS: The differences in growth patterns between males and females indicate the importance of analysing male and female data separately when describing growth. As a result, Gompertz model can be recommended to Lori-Bakhtiari female and male lamb breeders to determine more accurate growth traits. In addition, it should be considered that feeding male and female lambs separately according to absolute growth rate values may increase growth performance.

A Bayesian non-stationary heteroskedastic time series model for multivariate critical care data.

We propose a multivariate GARCH model for non-stationary health time series by modifying the observation-level variance of the standard state space model. The proposed model provides an intuitive and novel way of dealing with heteroskedastic data using the conditional nature of state-space models. We follow the Bayesian paradigm to perform the inference procedure. In particular, we use Markov chain Monte Carlo methods to obtain samples from the resultant posterior distribution. We use the forward filtering backward sampling algorithm to efficiently obtain samples from the posterior distribution of the latent state. The proposed model also handles missing data in a fully Bayesian fashion. We validate our model on synthetic data and analyze a data set obtained from an intensive care unit in a Montreal hospital and the MIMIC dataset. We further show that our proposed models offer better performance, in terms of WAIC than standard state space models. The proposed model provides a new way to model multivariate heteroskedastic non-stationary time series data. Model comparison can then be easily performed using the WAIC.

Enhancing deep neural network training efficiency and performance through linear prediction.

Deep neural networks have achieved remarkable success in various fields. However, training an effective deep neural network still poses challenges. This paper aims to propose a method to optimize the training effectiveness of deep neural networks, with the goal of improving their performance. Firstly, based on the observation that parameters (weights and bias) of deep neural network change in certain rules during training process, the potential of parameters prediction for improving training efficiency is discovered. Secondly, the potential of parameters prediction to improve the performance of deep neural network by noise injection introduced by prediction errors is revealed. And then, considering the limitations comprehensively, a deep neural network Parameters Linear Prediction method is exploit. Finally, performance and hyperparameter sensitivity validations are carried out on some representative backbones. Experimental results show that by employing proposed Parameters Linear Prediction method, as opposed to SGD, has led to an approximate 1% increase in accuracy for optimal model, along with a reduction of about 0.01 in top-1/top-5 error. Moreover, it also exhibits stable performance under various hyperparameter settings, shown the effectiveness of the proposed method and validated its capacity in enhancing network's training efficiency and performance.

Autopsy case of linear nevus sebaceous syndrome with KRAS (G12D) mutation.

Linear nevus sebaceous syndrome (LNSS) is a neurocutaneous syndrome associated with systemic complications that involve multiple organs, including the skin, central nervous system, eyes, and skeleton. LNSS is considered to be caused by mosaic RAS gene mutation. In this report, we present an autopsy case of LNSS in a Japanese boy. The affected neonate had hydrops fetalis and was born at 28 weeks and 4 days of gestation, weighing 2104 g. He had bilateral inverted eyelids, verrucous linear nevus separated along Blaschko's line, myocardial hypertrophy, and pharyngeal constriction, and underwent intensive treatment in NICU for arrhythmia, hydrocephalus, and respiratory distress. The hydrocephalus progressed gradually and he died at the age of 181 days, 12 days after a sudden cardiac arrest and recovery. KRAS G12D mutation was found in a skin biopsy specimen but not in blood cells, suggesting a postzygotic mosaicism. Autopsy revealed novel pathological findings related to LNSS, including intracranial lipomatous hamartoma and mesenteric lymphangioma, in addition to previously reported findings such as multicystic dysplastic kidney. There was the limited expression of mutated KRAS protein in kidneys.

Design of a Cavity for the High-Power Radio-Frequency Quadrupole Coupler Test for the ANTHEM Project.

The ANTHEM (Advanced Technologies for Human-centered Medicine) Radio-Frequency Quadrupole (RFQ) will employ eight coaxial power couplers, which will be magnetically coupled to the device through a loop antenna. The coupler design can support up to 140 kW in continuous wave operation. This paper presents the design of the cavity used for high-power testing, with the primary objectives of both optimizing the coupling between the couplers and ensuring operations at the designated operating frequency. Furthermore, the paper encompasses thermal and structural assessments conducted through numerical simulations.

Single-Track Magnetic Tape Absolute Position Sensor with Self-Adaptivity.

In this study, we demonstrate a single-track magnetic code tape-based absolute position sensor system. Unlike traditional dual-track systems, our method simplifies manufacturing and avoids crosstalk between tracks, offering higher tolerance to alignment errors. The sensing system employs an array of magnetic field sensing elements that recognize the bit sequence encoded on the tape. This approach allows for accurate position determination even when the number of sensing elements is fewer than the number of bits covered, and without the need for specific spacing between sensing elements and bit length. We demonstrate the system's ability to learn and adapt to various magnetic code patterns, including those that are irregular or have been altered. Our method can identify and localize the sensed magnetic field pattern directly within a self-learned magnetic field map, providing robust performance in diverse conditions. This self-adaptive capability enhances operational safety and reliability, as the system can continue functioning even when the magnetic tape is misaligned or has undergone changes.

Wearable Alcohol Monitoring Device for the Data-Driven Transcutaneous Alcohol Diffusion Model.

Wearable alcohol monitoring devices demand noninvasive, real-time measurement of blood alcohol content (BAC) reliably and continuously. A few commercial devices are available to determine BAC noninvasively by detecting transcutaneous diffused alcohol. However, they suffer from a lack of accuracy and reliability in the determination of BAC in real time due to the complex scenario of the human skin for transcutaneous alcohol diffusion and numerous factors (e.g., skin thickness, kinetics of alcohol, body weight, age, sex, metabolism rate, etc.). In this work, a transcutaneous alcohol diffusion model has been developed from real-time captured data from human wrists to better understand the kinetics of diffused alcohol from blood to different skin epidermis layers. Such a model will be a footprint to determine a base computational model in larger studies. Eight anonymous volunteers participated in this pilot study. A laboratory-built wearable blood alcohol content (BAC) monitoring device collected all the data to develop this diffusion model. The proton exchange membrane fuel cell (PEMFC) sensor was fabricated and integrated with an nRF51822 microcontroller, LMP91000 miniaturized potentiostat, 2.4 GHz transceiver supporting Bluetooth low energy (BLE), and all the necessary electronic components to build this wearable BAC monitoring device. The %BAC data in real time were collected using this device from these volunteers' wrists and stored in the end device (e.g., smartphone). From the captured data, we demonstrate how the volatile alcohol concentration on the skin varies over time by comparing the alcohol concentration in the initial stage (= 10 min) and later time (= 100 min). We also compare the experimental results with the outputs of three different input profiles: piecewise linear, exponential linear, and Hoerl, to optimize the developed diffusion model. Our results demonstrate that the exponential linear function best fits the experimental data compared to the piecewise linear and Hoerl functions. Moreover, we have studied the impact of skin epidermis thickness within ±20% and demonstrate that a 20% decrease in this thickness results in faster dynamics compared to thicker skin. The model clearly shows how the diffusion front changes within a skin epidermis layer with time. We further verified that 60 min was roughly the time to reach the maximum concentration, Cmax, in the stratum corneum from the transient analysis. Lastly, we found that a more significant time difference between BACmax and Cmax was due to greater alcohol consumption for a fixed absorption time.

Using Support Vector Machines to Classify Road Surface Conditions to Promote Safe Driving.

Accurate detection of road surface conditions in adverse winter weather is essential for traffic safety. To promote safe driving and efficient road management, this study presents an accurate and generalizable data-driven learning model for the estimation of road surface conditions. The machine model was a support vector machine (SVM), which has been successfully applied in diverse fields, and kernel functions (linear, Gaussian, second-order polynomial) with a soft margin classification technique were also adopted. Two learner designs (one-vs-one, one-vs-all) extended their application to multi-class classification. In addition to this non-probabilistic classifier, this study calculated the posterior probability of belonging to each group by applying the sigmoid function to the classification scores obtained by the trained SVM. The results indicate that the classification errors of all the classifiers, excluding the one-vs-all linear learners, were below 3%, thereby accurately classifying road surface conditions, and that the generalization performance of all the one-vs-one learners was within an error rate of 4%. The results also showed that the posterior probabilities can analyze certain atmospheric and road surface conditions that correspond to a high probability of hazardous road surface conditions. Therefore, this study demonstrates the potential of data-driven learning models in classifying road surface conditions accurately.

LMI-Based MPC Design Applied to the Single-Phase PWM Inverter with LC Filter under Uncertain Parameters.

This work proposes a design methodology for predictive control applied to the single-phase PWM inverter with an LC filter. In the design, we considered that the PWM inverter has parametric uncertainties in the filter inductance and output load resistance. The control system purpose is to track a sinusoidal signal at the inverter output. The designed control system with an embedded integrator uses the principle of receding horizon control, which underpinned predictive control. The methodology was described by linear matrix inequalities, which can be solved efficiently using convex programming techniques, and the optimal solution is obtained. MATLAB-Simulink and real-time FPGA-in-the-loop simulations illustrate the viability of the proposed control system. The LMI-based MPC reveals an effective performance for tracking of a sinusoidal reference signal and disturbance rejection of input voltage and load perturbations for the inverter subject to uncertainties.

Development and Experiment of Semi-Physical Simulation Platform for Space Manipulator.

To address the extended development cycle, high costs, and maintenance difficulties associated with existing microgravity simulation methods, this study has developed a semi-physical simulation platform for robotic arms tailored to different gravity environments and loading conditions. The platform represents difficult-to-model joints as physical objects, while the easily modeled components are simulated based on principles of similarity. In response to the strong coupling, nonlinearity, and excess force disturbance issues in the electric variable load loading system, a fractional-order linear active disturbance rejection control algorithm was employed. The controller parameters were tuned using an improved particle swarm algorithm with modified weight coefficients, and experimental results demonstrate that a fractional-order linear active disturbance rejection control improves response speed and disturbance rejection performance compared to linear sliding mode control. The study investigated the differences in the drive force of joint motors in space robotic arms under varying gravity environments and loading conditions. Experimental results indicate that load torque is the primary influencing factor on joint motor drive force, while radial force serves as a secondary influencing factor. Additionally, when the axis of the joint motor is perpendicular to the ground, it can, to some extent, simulate microgravity conditions on the ground.

Distributed State Observer for Systems with Multiple Sensors under Time-Delay Information Exchange.

The issues of state estimations based on distributed observers for linear time-invariant (LTI) systems with multiple sensors are discussed in this paper. We deal with the scenario when the information exchange has known time delays, and aim at designing a distributed observer for each subsystem such that each distributed observer can estimate the system state asymptotically by rejecting the time delay. To begin with, by rewriting the target system in a connecting form, a subsystem which is affected by the time-delay states of other nodes is established. And then, for this subsystem, a distributed observer with time delay is constructed. Moreover, an equivalent state transformation is made for the observer error dynamic system based on the observable canonic decomposition theorem. Further, in order to ensure that the distributed observer error dynamic system is asymptotically stable even if there exists a time delay, a linear matrix inequality (LMI) which is relative to the Laplace matrix is elaborately set up, and a special Lyapunov function candidate based on the LMI is considered. Next, based on the Lyapunov function and Lyapunov stability theory, we prove that the error dynamic system of the distributed observer is asymptotically stable, and the observer gain is determined by a feasible solution of the LMI. Finally, a simulation example is given to illustrate the effectiveness of the proposed method.

Prevalence of mandibular asymmetries in the pediatric population of Jazan: A radiographic analytical study.

Background: Facial asymmetry results from variation in mandibular linear and angular dimensions on the right and left sides of the face. Mandibular asymmetry is of great significance to oral surgeons and orthodontists as it directly impacts the facial profile of an individual. Aim: The present study aimed to measure the prevalence of mandibular asymmetry and its fluctuations during the mixed dentition growth phase in healthy children aged 6-8 years in the Jazan region of Saudi Arabia. Method: This retrospective observational study was conducted by measuring linear asymmetrical measurements of mandible on orthopantomograms of 390 healthy children (182 boys and 208 girls, aged 6-8 years) with mixed dentition. Linear measurements from orthopantomograms were obtained using a standardized digitizer. Two sets of mandibular measurements were recorded, alongside subjective assessments of mandibular first molar development. An independent t-test was employed to assess the significance between measurements on both sides, while one-way ANOVA was used to demonstrate facial asymmetry significance among different age groups. Result: The result of this study revealed a significant statistical difference (p-value≤ 0.05) for both sides of the mandible across two dimensions: condylar and ramus height (p value = 0.03) and mandibular length (p value = 0.04). The asymmetry index resulted in no asymmetry among most of the included subjects. However, compared to the other three linear measurements, many seven-year-old participants possess mandibular asymmetry on condylar height (54.5 %). Conclusion: Within the limitation it could be concluded that children in growing age have a significant mandibular asymmetry (mainly 7 years), which, however, is only seldom clinically significant. Hence, treatment plan should be cautiously planned.

Paddy insect identification using deep features with lion optimization algorithm.

Pests are a significant challenge in paddy cultivation, resulting in a global loss of approximately 20 % of rice yield. Early detection of paddy insects can help to save these potential losses. Several ways have been suggested for identifying and categorizing insects in paddy fields, employing a range of advanced, noninvasive, and portable technologies. However, none of these systems have successfully incorporated feature optimization techniques with Deep Learning and Machine Learning. Hence, the current research provided a framework utilizing these techniques to detect and categorize images of paddy insects promptly. Initially, the suggested research will gather the image dataset and categorize it into two groups: one without paddy insects and the other with paddy insects. Furthermore, various pre-processing techniques, such as augmentation and image filtering, will be applied to enhance the quality of the dataset and eliminate any unwanted noise. To determine and analyze the deep characteristics of an image, the suggested architecture will incorporate 5 pre-trained Convolutional Neural Network models. Following that, feature selection techniques, including Principal Component Analysis (PCA), Recursive Feature Elimination (RFE), Linear Discriminant Analysis (LDA), and an optimization algorithm called Lion Optimization, were utilized in order to further reduce the redundant number of features that were collected for the study. Subsequently, the process of identifying the paddy insects will be carried out by employing 7 ML algorithms. Finally, a set of experimental data analysis has been conducted to achieve the objectives, and the proposed approach demonstrates that the extracted feature vectors of ResNet50 with Logistic Regression and PCA have achieved the highest accuracy, precisely 99.28 %. However, the present idea will significantly impact how paddy insects are diagnosed in the field.

Motor Control in Parkinson's Disease During the Performance of Multi-Joint Reversal Movements.

We tested if the movement slowness of individuals with Parkinson's disease is related to their decreased ability to generate adequate net torques and linearly coordinate them between joints. This cross-sectional study included ten individuals with Parkinson's disease and ten healthy individuals. They performed planar movements with a reversal over three target distances. We calculated joint kinematics of the elbow and shoulder using spatial orientation. The muscle, interaction, and net torques were integrated into the acceleration/deceleration phases of the fingertip speed. We calculated the linear correlations of those torques between joints. Both groups modulated the elbow and shoulder net torques with target distances. They linearly coupled the production of torques. Both groups did not modulate the interaction torques. The movement slowness in Parkinson's disease was related to the difficulty in generating the appropriate muscle and net torques in the task. The interaction torques do not seem to play any role in movement control.

Generalized nonlinearity in animal ecology: Research, review, and recommendations.

Generalized linear models (GLMs) are an integral tool in ecology. Like general linear models, GLMs assume linearity, which entails a linear relationship between independent and dependent variables. However, because this assumption acts on the link rather than the natural scale in GLMs, it is more easily overlooked. We reviewed recent ecological literature to quantify the use of linearity. We then used two case studies to confront the linearity assumption via two GLMs fit to empirical data. In the first case study we compared GLMs to generalized additive models (GAMs) fit to mammal relative abundance data. In the second case study we tested for linearity in occupancy models using passerine point-count data. We reviewed 162 studies published in the last 5 years in five leading ecology journals and found less than 15% reported testing for linearity. These studies used transformations and GAMs more often than they reported a linearity test. In the first case study, GAMs strongly out-performed GLMs as measured by AIC in modeling relative abundance, and GAMs helped uncover nonlinear responses of carnivore species to landscape development. In the second case study, 14% of species-specific models failed a formal statistical test for linearity. We also found that differences between linear and nonlinear (i.e., those with a transformed independent variable) model predictions were similar for some species but not for others, with implications for inference and conservation decision-making. Our review suggests that reporting tests for linearity are rare in recent studies employing GLMs. Our case studies show how formally comparing models that allow for nonlinear relationships between the dependent and independent variables has the potential to impact inference, generate new hypotheses, and alter conservation implications. We conclude by suggesting that ecological studies report tests for linearity and use formal methods to address linearity assumption violations in GLMs.

Analysis of Virus-Specific B Cell Epitopes Reveals Extensive Antigen Degradation Prior to Recognition.

B cell epitopes must be visible for recognition by cognate B cells and/or antibodies. Here, we studied that premise for known linear B cell epitopes that were collected from the Immune Epitope Database as being recognized by humans during microbial infections. We found that the majority of such known B cell epitopes are virus-specific linear B cell epitopes (87.96%), and most are located in antigens that remain enclosed in host cells and/or virus particles, preventing antibody recognition (18,832 out of 29,225 epitopes). Moreover, we estimated that only a minority (32.72%) of the virus-specific linear B cell epitopes that are found in exposed viral regions (e.g., the ectodomains of envelope proteins) are solvent accessible on intact antigens. Hence, we conclude that ample degradation/processing of viral particles and/or infected cells must occur prior to B cell recognition, thus shaping the B cell epitope repertoire.

Predicting Longitudinal Cognitive Decline and Alzheimer's Conversion in Mild Cognitive Impairment Patients Based on Plasma Biomarkers.

The increasing burden of Alzheimer's disease (AD) emphasizes the need for effective diagnostic and therapeutic strategies. Despite available treatments targeting amyloid beta (Aß) plaques, disease-modifying therapies remain elusive. Early detection of mild cognitive impairment (MCI) patients at risk for AD conversion is crucial, especially with anti-Aß therapy. While plasma biomarkers hold promise in differentiating AD from MCI, evidence on predicting cognitive decline is lacking. This study's objectives were to evaluate whether plasma protein biomarkers could predict both cognitive decline in non-demented individuals and the conversion to AD in patients with MCI. This study was conducted as part of the Korean Longitudinal Study on Cognitive Aging and Dementia (KLOSCAD), a prospective, community-based cohort. Participants were based on plasma biomarker availability and clinical diagnosis at baseline. The study included MCI (n = 50), MCI-to-AD (n = 21), and cognitively unimpaired (CU, n = 40) participants. Baseline plasma concentrations of six proteins-total tau (tTau), phosphorylated tau at residue 181 (pTau181), amyloid beta 42 (Aß42), amyloid beta 40 (Aß40), neurofilament light chain (NFL), and glial fibrillary acidic protein (GFAP)-along with three derivative ratios (pTau181/tTau, Aß42/Aß40, pTau181/Aß42) were analyzed to predict cognitive decline over a six-year follow-up period. Baseline protein biomarkers were stratified into tertiles (low, intermediate, and high) and analyzed using a linear mixed model (LMM) to predict longitudinal cognitive changes. In addition, Kaplan-Meier analysis was performed to discern whether protein biomarkers could predict AD conversion in the MCI subgroup. This prospective cohort study revealed that plasma NFL may predict longitudinal declines in Mini-Mental State Examination (MMSE) scores. In participants categorized as amyloid positive, the NFL biomarker demonstrated predictive performance for both MMSE and total scores of the Korean version of the Consortium to Establish a Registry for Alzheimer's Disease Assessment Packet (CERAD-TS) longitudinally. Additionally, as a baseline predictor, GFAP exhibited a significant association with cross-sectional cognitive impairment in the CERAD-TS measure, particularly in amyloid positive participants. Kaplan-Meier curve analysis indicated predictive performance of NFL, GFAP, tTau, and Aß42/Aß40 on MCI-to-AD conversion. This study suggests that plasma GFAP in non-demented participants may reflect baseline cross-sectional CERAD-TS scores, a measure of global cognitive function. Conversely, plasma NFL may predict longitudinal decline in MMSE and CERAD-TS scores in participants categorized as amyloid positive. Kaplan-Meier curve analysis suggests that NFL, GFAP, tTau, and Aß42/Aß40 are potentially robust predictors of future AD conversion.

Comprehensive quality control of silkworm chrysalis using chemical fingerprints combined with antioxidant activity and acetylcholine content.

In recent years, some insects have become foods due to their high nutritional value. In order to solve the problem of the lack of quality control methods for insect foods, this study proposes a comprehensive control model using silkworm chrysalis (SC) as an example. Firstly, five-wavelength mean fusion fingerprints (FWMFF) and UV quantum fingerprints of 21 batches of SC were established. And the 21 batches of SC were classified into different grades from different perspectives by using the comprehensive linear quantified fingerprint method (CLQFM) as a quality evaluation method for qualitative and quantitative analysis. Secondly, this paper fully considered the issue of the reliability of fingerprint evaluation, which guaranteed the accuracy of the evaluation results. On this basis, the antioxidant capacity of the samples was used in vitro 1,1-Diphenyl-2-picrylhydrazylradical (DPPH) scavenging assay using IC50. The relationship between fingerprints and antioxidant activity was also discussed. Finally, the content of endogenous neurotransmitter (ACh) in SC determined by ultra-high performance liquid chromatography-tandem mass spectrometry (UHPLC-MS/MS) in the range of 0.25-2.11µg/g. Overall, the present study proposes a comprehensive quality control strategy for functional foods based on the quality assessment of SC.

A deep-learning-based surrogate model for Monte-Carlo simulations of the linear energy transfer in primary brain tumor patients treated with proton-beam radiotherapy.

OBJECTIVE: This study explores the use of neural networks (NNs) as surrogate models for Monte-Carlo (MC) simulations in predicting the dose-averaged linear energy transfer (LETd) of protons in proton-beam therapy based on the planned dose distribution and patient anatomy in the form of computed tomography (CT) images. As LETdis associated with variability in the relative biological effectiveness (RBE) of protons, we also evaluate the implications of using NN predictions for normal tissue complication probability (NTCP) models within a variable-RBE context. Approach: The predictive performance of three-dimensional NN architectures was evaluated using five-fold cross-validation on a cohort of brain tumor patients (n=151). The best-performing model was identified and externally validated on patients from a different center (n=107). LETdpredictions were compared to MC-simulated results in clinically relevant regions of interest. We assessed the impact on NTCP models by leveraging LETdpredictions to derive RBE-weighted doses, using the Wedenberg RBE model. Main results: We found NNs based solely on the planned dose profile, i.e. without additional usage of CT images, can approximate MC-based LETddistributions. Root mean squared errors (RMSE) for the median LETdwithin the brain, brainstem, CTV, chiasm, lacrimal glands (ipsilateral/contralateral) and optic nerves (ipsilateral/contralateral) were 0.36, 0.87, 0.31, 0.73, 0.68, 1.04, 0.69 and 1.24~keV/µm, respectively. Although model predictions showed statistically significant differences from MC outputs, these did not result in substantial changes in NTCP predictions, with RMSEs of at most 3.2 percentage points. Significance: The ability of NNs to predict LETdbased solely on planned dose profiles suggests a viable alternative to the compute-intensive MC simulations in a variable-RBE setting. This is particularly useful in scenarios where MC simulation data are unavailable, facilitating resource-constrained proton therapy treatment planning, retrospective patient data analysis and further investigations on the variability of proton RBE.