Dose Prediction Research Articles

Deep learning-based Monte Carlo dose prediction for heavy-ion online adaptive radiotherapy and fast quality assurance: A feasibility study.

Online adaptive radiotherapy (OART) and rapid quality assurance (QA) are essential for effective heavy ion therapy (HIT). However, there is a shortage of deep learning (DL) models and workflows for predicting Monte Carlo (MC) doses in such treatments. This study seeks to address this gap by developing a DL model for independent MC dose (MCDose) prediction, aiming to facilitate OART and rapid QA implementation for HIT. A MC dose prediction DL model called CAM-CHD U-Net for HIT was introduced, based on the GATE/Geant4 MC simulation platform. The proposed model improved upon the original CHD U-Net by adding a Channel Attention Mechanism (CAM). Two experiments were conducted, one with CHD U-Net (Experiment 1) and another with CAM-CHD U-Net (Experiment 2), and involved data from 120 head and neck cancer patients. Using patient CT images, three-dimensional energy matrices, and ray-masks as inputs, the model completed the entire MC dose prediction process within a few seconds. In Experiment 2, within the Planned Target Volume (PTV) region, the average gamma passing rate (3%/3mm) between the predicted dose and true MC dose reached 99.31%, and 96.48% across all body voxels. Experiment 2 demonstrated a 46.15% reduction in the mean absolute difference in in organs at risk compared to Experiment 1. By extracting relevant parameters of radiotherapy plans, the CAM-CHD U-Net model can directly and accurately predict independent MC dose, and has a high gamma passing rate with the ground truth dose (the dose obtained after a complete MC simulation). Our workflow enables the implementation of heavy ion OART, and the predicted MCDose can be used for rapid QA of HIT.

Modeling dosimetric benefits from daily adaptive RT for gynecological cancer patients with and without knowledge-based dose prediction.

Daily online adaptive radiotherapy (ART) improves dose metrics for gynecological cancer patients, but the on-treatment process is resource-intensive requiring longer appointments and additional time from the entire adaptive team. To optimize resource allocation, we propose a model to identify high-priority patients. For 49 retrospective cervical and endometrial cancer patients, we calculated two initial plans: the treated standard-of-care (InitialSOC) and a reduced margin initial plan (InitialART) for adapting with the Ethos treatment planning system. Daily doses corresponding to standard and reduced margins (DailySOC and DailyART) were determined by re-segmenting the anatomy based on the treatment CBCT and calculating dose on a synthetic CT. These initial and daily doses were used to estimate the ART benefit ( =DailySOC-DailyART) versus initial plan differences ( =InitialSOC-InitialART) via multivariate linear regression. Dosimetric benefits were modeled with initial plan differences ( ) of (cc), (Gy), and (Gy). Anatomy (intact uterus or post-hysterectomy), DoseType (simultaneous integrated boost [SIB] vs. single dose), and/or prescription value. To establish a logistic model, we classified the top 10% in each metric as high-benefit patients. We then built a logistic model to predict these patients from the previous predictors. Leave-one-out validation and ROC analysis were used to evaluate the accuracy. To improve the clinical efficiency of this predictive process, we also created knowledge-based plans for the ΔInitial plans ( ) and repeated the analysis. In both and our multivariate analysis showed low R2 values 0.34-0.52 versus 0.14-0.38. The most significant predictor in each multivariate model was the corresponding ∆Initial metric (e.g., Bowel (V40Gy), p<1e-05). In the logistic model, the metrics with the strongest correlation to the high-benefit patients were (cc), (Gy), , and prescription. The models for original and knowledge-based plans had an AUC of 0.85 versus 0.78. The sensitivity and specificity were 0.92/0.72 and 0.69/0.80, respectively. This methodology will allow clinics to prioritize patients for resource-intensive daily online ART.

Abstract A007: Artificial intelligence in radiation therapy: Bringing in a new era of cancer therapy, accurate and fast

Abstract Introduction: The application of AI in RT has been adopted as innovative method that seeks to enhance accuracy, effectiveness and patient outcome of cancer treatment. This aims to understand current and potential uses of AI-based tools in RT; from planning, monitoring during therapy, adaptive approach, and after treatment assessment. This study defines the role of AI in facilitating decision-making, made to define treatment plans, and to organize the clinical practice in order to improve the therapeutic outcomes and provide individualized approach to the cancer patients. Methods and Experimental Procedures: Radiation therapy work flows which integrated AI algorithms for treatment planning and delivery were examined in a retrospective manner. Imaging data was used to predict tumor behavior and develop the best treatment plan since the models employed by practitioners involved machine learning. Another application of AI was for the shape matching of its patient anatomy during treatment and came up with better matching for therapy adjustments. This involved data from multiple clinical centers employing AI-assisted techniques for prostate, lung as well head and neck cancers. These QA procedures implemented with the help of AI were assessed with respect to their capability in identifying errors and maintaining dose accuracy. Results: Initial data showed that AI optimizes accuracy and time in treatment planning by streamlining the processes of outlining tumors and relevant tissues; planning time is cut in half in comparison with the conventional approach. Real-time positioning indicated by AI optimized adaptive therapy gave accuracy in tracking the anatomical variations thereby justifying the alterations in radiation delivery that enhanced the tumor control without affecting normal tissue. Automated quality assurance systems with AI capabilities successfully detected possible mistakes in the application of treatment processes and thus, decisively enhanced patients’ safety and minimized risks of radiation mishaps. The data also showed that models using AI enhanced accuracy in dose prediction and thus the overall treatment plans and results especially when the anatomy of a patient is unique and cannot easily be navigated. Conclusions: AI is becoming integrated into RT since it delivers, precision, personalization and efficiency that the industry desires. Advanced AI-based technologies help in treatment planning, facilitate monitoring and improve the technique of adaptive RT while also strengthening QA measures thus making patients safer and providing better therapeutic results. In the future, emerging technologies in AI offers prospects to address problems in radiation oncology for improved diagnosis and treatment of cancer. There is need for more clinical and research practice and development to actualize AI applications, remove the constraints and apply these systems in regular clinical services. The future of RT is going to be with AI incorporated in all the departments to ensure that cancer treatment is optimized for the benefit of the patient. Citation Format: Newtone Kisia. Artificial intelligence in radiation therapy: Bringing in a new era of cancer therapy, accurate and fast. [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Targeted Therapies in Combination with Radiotherapy; 2025 Jan 26-29; San Diego, CA. Philadelphia (PA): AACR; Clin Cancer Res 2025;31(2_Suppl):Abstract nr A007.

Contribution of Nuclear Fragmentation to Dose and RBE in Carbon-Ion Radiotherapy.

Variable relative biological effectiveness (RBE) of carbon radiotherapy may be calculated using several models, including the microdosimetric kinetic model (MKM), stochastic MKM (SMKM), repair-misrepair-fixation (RMF) model, and local effect model I (LEM), which have not been thoroughly compared. In this work, we compared how these four models handle carbon beam fragmentation, providing insight into where model differences arise. Monoenergetic and spread-out Bragg peak carbon beams incident on a water phantom were simulated using Monte Carlo. Using these beams, input parameters for each model (microdosimetric spectra, DNA double-strand break yield, kinetic energy spectra, physical dose fragment contributions) were calculated for each contributing carbon beam fragment (hydrogen, helium, lithium, beryllium, boron, secondary carbon, primary carbon, electrons, and "other"). Scored input parameters for each fragment were used to calculate linear (α) and quadratic (β) parameters according to each model, which were combined with reference α and β values and absorbed physical dose to calculate RBE. Contributions from secondary fragments were found to exceed 30% of the total physical dose. Using identical beam parameters, the four models produced not only different RBE values but also different RBE trends. In all models, RBE was highest for secondary carbon ions. Beyond secondary carbons, the RBE magnitude typically increased with the atomic number of the fragment, but RBE trends differed dramatically by model and beamline region (entrance, spread-out Bragg peak, and tail). Variations in fragment RBE were large enough to be apparent in biological dose predictions. This study demonstrated that fragmentation is a nonnegligible consideration in carbon radiotherapy. Our findings identified differences in RBE among specific fragments and the four models, contributing to variability in the total biological dose across models. Because these findings emphasize differences in how various models handle carbon beam fragments, greater care should be taken in characterization of secondary fragments in particle therapy.

Evaluating the Efficacy of Repurposed Antiretrovirals in Hepatitis B Virus Treatment: A Narrative Review of the Pros and Cons

Human immunodeficiency virus (HIV) and hepatitis B virus (HBV) continue to be global public health issues. Globally, about 39.9 million persons live with HIV in 2023, according to the Joint United Nations Programme on HIV/AIDS (UNAIDS) 2024 Fact Sheet. Consequently, the World Health Organisation (WHO) reported that about 1.5 million new cases of HBV occur, with approximately 820 thousand mortalities yearly. Conversely, the lower percentage of HBV (30%) cases that receive a diagnosis is a setback in achieving the WHO 2030 target for zero HBV globally. This has necessitated a public health concern to repurpose antiretroviral (ARV) drugs for the treatment of HBV diseases. This review provides an introductory background, including the pros and cons of repurposing antiretrovirals (ARVs) for HBV treatment. We examine the similarities in replication mechanisms between HIV and HBV. We further investigate some clinical studies and trials of co-infected and mono-infected patients with HIV–HBV. The topical keywords including repurposing ARV drugs, repurposing antiretroviral therapy, Hepatitis B drugs, HBV therapy, title, and abstracts are searched in PubMed, Web of Science, and Google Scholar. The advanced search includes the search period 2014–2024, full text, clinical trials, randomized control trials, and review. The search results filtered from 361 to 51 relevant articles. The investigations revealed that HIV and HBV replicate via a common route known as ‘reverse transcription’. Clinical trial results indicate that an early initiation of ARVs, particularly with tenofovir disoproxil fumarate (TDF) as part of a regimen, significantly reduced the HBV viral load in co-infected patients. In mono-infected HBV, timely and correct precise medication is essential for HBV viral load reduction. Therefore, genetic profiling is pivotal for successful ARV drug repurposing in HBV treatment. Pharmacogenetics enables the prediction of the right dosages, specific individual responses, and reactions. This study uniquely explores the intersection of pharmacogenetics and drug repurposing for optimized HBV therapy. Additional in vivo, clinical trials, and in silico research are important for validation of the potency, optimum dosage, and safety of repurposed antiretrovirals in HBV therapy. Furthermore, a prioritization of research collaborations comprising of regulators and funders to foster clinically adopting and incorporating repurposed ARVs for HBV therapy is recommended.

The Estimation Value of 99mTc-MAA in Comparison with 90Y-PET/MR-based Dosimetry in Selective Internal Radiation Therapy (SIRT) for Liver Malignancies.

This study intended to compare the radiation dose estimates to target and nontarget liver compartments from 99mTc-MAA SPECT/CT and 90Y-PET/MR scans in liver tumors treated by 90Y-glass microspheres. Dose estimation was performed for twenty-three eligible patients (13M, 10F) after 99mTc-MAA simulation using SPECT/CT imaging, and over 90Y-PET/MR images after 90Y-microsphere therapy. Simplicit90Y™ software was used for voxel-based dosimetry over the liver parenchyma. Dose estimates were obtained for whole healthy liver (HL), healthy injected liver (HIL), and tumor volumes. Pearson correlation, Bland-Altman plot, and Wilcoxon signed-ranks test were used for statistical analysis. The mean tumor dose was 270±111 Gy, the whole liver parenchyma dose was 26 ±12 Gy, and the healthy injected liver dose was 55±18 Gy from 99mTc-MAA simulation. 90YPET/ MR dosimetry yielded a mean tumor dose of 271±125 Gy, a HIL mean dose of 54±18 Gy, and a liver parenchyma dose of 25±12 Gy. An excellent agreement was demonstrated between tumor doses (R2=0.90) and liver doses (R2=0.87), while the agreement was less for HIL doses (R2=0.80). Wilcoxon signed-ranks test yielded no significant difference between the dose estimates for all liver compartments. It was deduced that 99mTc-MAA SPECT/CT simulation provides valuable dose prediction in 90Y-glass microsphere therapy. Despite the difference in volume measurements and dose estimates with 90Y-PET/MR, the predictive value of the 99mTc-MAA simulation was not affected.

Validating the accuracy of mathematical model-based pharmacogenomics dose prediction with real-world data.

The study aims to verify the usage of mathematical modeling in predicting patients' medication doses in association with their genotypes versus real-world data. The work relied on collecting, extracting, and using real-world data on dosing and patients' genotypes. Drug metabolizing enzymes, i.e., cytochrome CYP 450, were the focus. A total number of 1914 subjects from 26 studies were considered, and CYP2D6 and CYP2C19 gene polymorphisms were used for the verification. Results show that the mathematical model was able to predict the reported optimal dosing of the values provided in the considered studies. Predicting patients' optimal doses circumvents trial and error in patients' treatments. The authors discussed the advantages of using a mathematical model in patients' dosing and identified multiple issues that would hinder the usability of raw data in the future, especially in the era of artificial intelligence (AI). The authors recommend that researchers and healthcare professionals use simple descriptive metabolic activity terms for patients and use allele activity scores for drug dosing rather than phenotype/genotype classifications. The authors verified that a mathematical model could assist in providing data for better-informed decision-making in clinical settings and drug research and development.

Nuclear interaction correction based on dual-energy computed tomography in carbon-ion radiotherapy.

&#xD;Accurate dose predictions are crucial to maximizing the benefits of carbon-ion therapy. Carbon beams incident on the human body cause nuclear interactions with tissues, resulting in changes in the constituent nuclides and leading to dose errors that are conventionally corrected using conventional single-energy computed tomography (SECT). Dual-energy computed tomography (DECT) has frequently been used for stopping power estimation in particle therapy and is well suited for correcting nuclear reactions because of its detailed body-tissue elemental information. This study proposes a correction method for the absolute dose in carbon-ion therapy that considers changes in nuclide composition resulting from nuclear reactions with body tissues, as a novel application of DECT.&#xD;Approach:&#xD;The change in dose associated with nuclear reactions is determined by correcting each integrated depth dose component of the carbon beam using a nuclear interaction correction factor. This factor is determined based on the stopping power, mass density, and nuclear interaction cross-section in body tissue. The stopping power and mass density were calculated using established methods, whereas the nuclear interaction cross-section was newly defined through a conversion equation derived from the effective atomic number.&#xD;Main results:&#xD;Nuclear interaction correction factors and corrected doses were determined for 85 body tissues with known compositions, comparing them with existing SECT-based methods. The root-mean-square errors of the SECT- and DECT-based nuclear interaction correction factors relative to theoretical values were 0.66% and 0.39%, respectively.&#xD;Significance:&#xD;This indicates a notable enhancement in the estimation accuracy with DECT. The dose calculations in uniform body tissues derived from SECT showed slight over-correction in adipose and bone tissues, whereas those based on DECT were almost consistent with theoretical values. Our proposed method demonstrates the potential of DECT for enhancing dose calculation accuracy in carbon-ion therapy, complementing its established role in stopping power estimation.&#xD.

An Efficient 3D Convolutional Neural Network for Dose Prediction in Cancer Radiotherapy from CT Images.

Introduction: Cancer is a highly lethal disease with a significantly high mortality rate. One of the most commonly used methods for treatment is radiation therapy. However, cancer treatment using radiotherapy is a time-consuming process that requires significant manual work from planners and doctors. In radiation therapy treatment planning, determining the dose distribution for each of the regions of the patient's body is one of the most difficult and important tasks. Nowadays, artificial intelligence has shown promising results in improving the quality of disease treatment, particularly in cancer radiation therapy. Objectives: The main objective of this study is to build a high-performance deep learning model for predicting radiation therapy doses for cancer and to develop software to easily manipulate and use this model. Materials and Methods: In this paper, we propose a custom 3D convolutional neural network model with a U-Net-based architecture to automatically predict radiation doses during cancer radiation therapy from CT images. To ensure that the predicted doses do not have negative values, which are not valid for radiation doses, a rectified linear unit (ReLU) function is applied to the output to convert negative values to zero. Additionally, a proposed loss function based on a dose-volume histogram is used to train the model, ensuring that the predicted dose concentrations are highly meaningful in terms of radiation therapy. The model is developed using the OpenKBP challenge dataset, which consists of 200, 100, and 40 head and neck cancer patients for training, testing, and validation, respectively. Before the training phase, preprocessing and augmentation techniques, such as standardization, translation, and flipping, are applied to the training set. During the training phase, a cosine annealing scheduler is applied to update the learning rate. Results and Conclusions: Our model achieved strong performance, with a good DVH score (1.444 Gy) on the test dataset, compared to previous studies and state-of-the-art models. In addition, we developed software to display the dose maps predicted by the proposed model for each 2D slice in order to facilitate usage and observation. These results may help doctors in treating cancer with radiation therapy in terms of both time and effectiveness.

Therapeutic dose prediction using score-based diffusion model for pretreatment patient-specific quality assurance.

Implementing pre-treatment patient-specific quality assurance (prePSQA) for cancer patients is a necessary but time-consuming task, imposing a significant workload on medical physicists. Currently, the prediction methods used for prePSQA fall under the category of supervised learning, limiting their generalization ability and resulting in poor performance on new data. In the context of this work, the limitation of traditional supervised models was broken by proposing a conditional generation method utilizing unsupervised diffusion model. A conditional generation method base on the score-based diffusion model was proposed, which employed diffusion model for the first time to predict the predict patients' therapeutic doses (TherapDose). The proposed diffusion model TherapDose prediction method (DMTP) learns the data distribution of dose images. The data distribution contains the quantitative relationship between the radiotherapy dose (RTDose) derived from the VMAT plan files of the Treatment Planning System (TPS) and the measured Dose (MDose, i.e., TherapDose) obtained from the Dolphin Compass physical system. By sampling from the learnt distribution, efficient prediction of TherapDose was achieved. The training dataset comprises RTDose, and the MDose. The three-dimensional information of dose slice was utilized to predict TherapDose, aiming to enhance the accuracy and efficiency of TherapDose prediction. Root mean square error (RMSE), mean absolute error (MAE), and structural similarity (SSIM) metrics were leveraged to validate the effectiveness of the proposed method. Meanwhile, CT images were further added to test the impacts of CT images on the prediction effect of MDose. The DMTP method has demonstrated superior performance in predicting TherapDose within key anatomical regions including the head and neck, chest, and abdomen, outperforming existing state-of-the-art methods by achieving high-quality predictions as measured across different evaluation metrics. It indicates that the proposed method is highly effective and accurate in its dose prediction capabilities. The proposed method has proven to be highly effective, consistently outperforming state-of-the-art techniques in MDose prediction across multiple anatomical regions and evaluation metrics. This method can serve as a clinical aid to assist medical physicists in diminishing the measurement workload associated with prePSQA.

Integrating Deep Learning-Based Dose Distribution Prediction with Bayesian Networks for Decision Support in Radiotherapy for Upper Gastrointestinal Cancer.

Selecting the better techniques to harbor optimal motion management, either a stereotactic linear accelerator delivery using TrueBeam (TBX) or magnetic resonance-guided gated delivery using MRIdian (MRG), is time-consuming and costly. To address this challenge, we aimed to develop a decision-supporting algorithm based on a combination of deep learning-generated dose distributions and clinical data. We retrospectively analyzed 65 patients with liver or pancreatic cancer who underwent both TBX and MRG simulations and planning process. We trained three-dimensional U-Net deep learning models to predict dose distributions and generated dose volume histograms (DVHs) for each system. We integrated predicted DVH metrics into a Bayesian network (BN) model incorporating clinical data. The MRG prediction model outperformed the TBX model, demonstrating statistically significant superiorities in predicting normalized dose to the planning target volume (PTV) and liver. We developed a final BN prediction model integrating the predictive DVH metrics with patient factors like age, PTV size, and tumor location. This BN model an area under the receiver operating characteristic curve index of 83.56%. The decision tree derived from the BN model showed that the tumor location (abutting vs. apart of PTV to hollow viscus organs) was the most important factor to determine TBX or MRG. It provided a potential framework for selecting the optimal radiation therapy (RT) system based on individual patient characteristics. We demonstrated a decision-supporting algorithm for selecting optimal RT plans in upper gastrointestinal cancers, incorporating both deep learning-based dose prediction and BN-based treatment selection. This approach might streamline the decision-making process, saving resources and improving treatment outcomes for patients undergoing RT.

FDDM: Frequency-Decomposed Diffusion Model for Dose Prediction in Radiotherapy

FDDM: Frequency-Decomposed Diffusion Model for Dose Prediction in Radiotherapy

Prediction of lacosamide concentrations to support dose optimization during pregnancy.

We aimed to quantify and predict lacosamide exposure during pregnancy by developing a pregnancy physiologically-based pharmacokinetic model, allowing the prediction of potential dose increases to support maintaining a patient's preconception lacosamide concentrations. Models for nonpregnant adults and pregnant female patients were constructed using physiochemical and pharmacological parameters identified from literature review. Evaluation of plasma concentration data from human males was digitized from the literature. Concentration data in nonpregnant and pregnant human females were available from the Maternal Outcomes and Neurodevelopmental Effects of Antiepileptic Drugs (MONEAD) study, a longitudinal observational study which followed 11 nonpregnant and 16 pregnant women receiving lacosamide. Evaluation was conducted qualitatively with visual overlay (>80% of observed concentrations within 90% confidence interval) and quantitatively with average fold error and absolute average fold error (0.8-1.25 ratio acceptance criteria). Simulations of intensively-sampled dosing regimens at steady-state dosing across multiple gestational ages were conducted in Simcyp to evaluate the potential changes in lacosamide pharmacokinetics during pregnancy. Additional simulations were performed to explore the effects of cytochrome polymorphisms and glomerular filtration rate variability. The model adequately described the evaluation data in nonpregnant adults and pregnant adults between 10 and 40 weeks of gestation. Estimates in patients at 40- weeks of gestation indicated that lacosamide clearance increased by 48.2% compared to nonpregnant patients. Maximum lacosamide concentration (Cmax) during a simulated dosing interval also fell by 30% from preconception to 40 weeks. A simulated dose increase of 50 mg once daily at 10 weeks of gestation supported maintenance of preconception concentration for a typical patient taking the most common dosing regimen of 200 mg, twice daily (BID), preconception. Simulated changes in lacosamide concentration align with the limited data available in observational studies. Our simulations support the use of therapeutic drug monitoring and dose adjustments to maintain the efficacy of lacosamide pharmacotherapy.

MoA Studies of the TEAD P-Site Binding Ligand MSC-4106 and Its Optimization to TEAD1-Selective Amide M3686.

Taking the structural information into account, we were able to tune the TEAD selectivity for a specific chemotype. However, different TEAD selectivity profiles did not affect the compound potency or efficacy in the NCI-H226 viability assay. Amides based on MSC-4106 or analogues showed improved viability efficacy compared with the corresponding acids. The amide M3686 exhibited AUC-driven efficacy in NCI-H226 xenograft models and had an improved 25-fold lower human dose prediction than MSC-4106. MSC-4106 was also used in HDX-MS studies to aid in the understanding of the MoA of P-site binding TEAD inhibitors. Artificial P-site binders rigidify certain areas in the periphery of the transcription factor that seem to be crucial for cofactor interaction, whereas a native fatty acid increased the protein dynamics of cofactor-binding interfaces.

Prediction of NPK doses based on targeted fruit sugar content in Cucumis melo L. ‘Cantaloupe’ using a simple regression method

The fruit sweetness is the main target in melon plant production. The highest criterion of sweetness is excellent, with 16% of total sugar content. Modification of essential plant nutrients is the alternative to reach that category. So, this study aims to obtain optimum NPK doses using a simple regression method. The experiment was conducted in a greenhouse with a soilless culture hydroponic system from August until November 2023 using a completely randomized design (CRD) with five treatments and four replications. The parameters included leaf area, plant dry matter, leaf nutrient uptake, fruit weight, and fruit sugar content. Pearson correlation analysis showed that the total sugar content in fruit has a significantly positive correlation with potassium in NPK fertilizer treatments such as K2O dose and K2O uptake at 7 WAP, i.e., 0.932 and 0.973, respectively. According to the regression model y = -50.7 + 1.079 N + 0.251 P2O5 + 0.528 K2O, the NPK formula fertilizer containing 31.56 g N, 23.99 g P2O5, and 50.42 g K2O can be used by grower to produce excellent fruit sugar content.

Toward Dose Prediction at Point of Design.

Human dose prediction (HDP) is a useful tool for compound optimization in preclinical drug discovery. We describe here our exclusively in silico HDP strategy to triage compound designs for synthesis and experimental profiling. Our goal is a model that provides a preliminary estimate of the dose for a given exposure target based on chemical structure. First, we construct machine learning models to estimate rat pharmacokinetics, which are subsequently allometrically scaled to estimate human pharmacokinetics. Second, we establish a 10 nM free concentration target for early HDP where potency data are not yet available. Finally, we assess the uncertainty associated with each model and propagate these into the final estimate, providing us with actionable guidance on the level of accuracy of these estimates. We find that this strategy can reduce preparation of compounds with poor properties relative to an unstructured approach, but extensive experimental testing remains required.

Artificial Intelligence and Machine Learning Applications to Pharmacokinetic Modeling and Dose Prediction of Antibiotics: A Scoping Review.

Background and Objectives: The use of artificial intelligence (AI) and, in particular, machine learning (ML) techniques is growing rapidly in the healthcare field. Their application in pharmacokinetics is of potential interest due to the need to relate enormous amounts of data and to the more efficient development of new predictive dose models. The development of pharmacokinetic models based on these techniques simplifies the process, reduces time, and allows more factors to be considered than with classical methods, and is therefore of special interest in the pharmacokinetic monitoring of antibiotics. This review aims to describe the studies that use AI, mainly oriented to ML techniques, for dose prediction and analyze their results in comparison with the results obtained by classical methods. Furthermore, in the review, the techniques employed and the metrics to evaluate the precision are described to improve the compression of the results. Methods: A systematic search was carried out in the EMBASE, OVID, and PubMed databases and the results obtained were analyzed in detail. Results: Of the 13 articles selected, 10 were published in the last three years. Vancomycin was monitored in seven and none of the studies were performed on new antibiotics. The most used techniques were XGBoost and neural networks. Comparisons were conducted in most cases against population pharmacokinetic models. Conclusions: AI techniques offer promising results. However, the diversity in terms of the statistical metrics used and the low power of some of the articles make the overall assessment difficult. For now, AI-based ML techniques should be used in addition to classical population pharmacokinetic models in clinical practice.

Investigating LETd optimization strategies in carbon ion radiotherapy for pancreatic cancer: a dosimetric study using an anthropomorphic phantom.

Clinical carbon ion beams offer the potential to overcome hypoxia-induced radioresistance in pancreatic tumors, due to their high dose-averaged Linear Energy Transfer (LETd), as previous studies have linked a minimum LETd within the tumor to improved local control. Current clinical practices at the Heidelberg Ion-Beam Therapy Center (HIT), which use two posterior beams, do not fully exploit the LETd advantage of carbon ions, as the high LETd is primarily focused on the beams' distal edges. Different LETd-boosting strategies, such as Spot-scanning Hadron Arc (SHArc), could enhance LETd distribution by concentrating high-LETd values in potential hypoxic tumor cores while sparing organs at risk. This study aims to investigate and verify different LETd-boosting strategies using an anthropomorphic pancreas phantom. Various LETd-boosting strategies were investigated for a cylindrical and a pancreas-shaped target in an anthropomorphic pancreas phantom. Treatment plans were optimized using single field optimization (SFO) or multi field optimization (MFO), with objective functions based on either physical dose (Phys), relative biological effectiveness(RBE)-weighted dose, or a combination of RBE and LETd-based objectives (LETopt). The LETd-boosting planning strategies were optimized with the goal of increasing the minimum LETd in the tumor without compromising its homogeneous dose coverage. Beam configurations investigated included the two-beam in-house clinical standard (2-SFOPhys, 2-SFORBE and 2-MFORBE-LETopt), a three-beam configuration (3-MFORBE and 3-MFORBE-LETopt) and SHArc (SHArcPhys, SHArcRBE and SHArcRBE-LETopt) using step-and-shoot delivery. The different plans were verified using an anthropomorphic pancreas phantom at HIT and compared to treatment planning system (TPS) predictions. All investigated LETd-boosting strategies altered the LETd distribution while meeting optimization goals and constraints, resulting in varying degrees of LETd enhancement. For the cylindrical volume, the SHArc plan resulted in the highest LETd concentration in the tumor core, with the minimum LETd in the GTV scaling up to 91keV/µm. For the pancreas-shaped volume, however, the 3-MFORBE-LETopt achieved a higher minimum LETd in the GTV than SHArcRBE (75.6 and 62.3keV/µm, respectively). When combining SHArc with LETd optimization, a minimum LETd of 76.3keV/µm was achieved, suggesting a potential benefit from this combined approach. Most dosimetric verifications showed dose deviations to the TPS within a 5% range, for both beam-per-beam and total dose. LETd-optimized and SHArc plans exhibited slightly higher mean dose deviations (2.0%-4.6%) compared to the standard RBE-based plans (<1.5%). This study demonstrated the feasibility of enhancing LETd in pancreatic tumors using carbon ion arc delivery coupled with LETd optimization. The possibility of delivering these plans was verified through irradiation of an anthropomorphic pancreas phantom, which showed agreement between dose measurements and predictions.

Evaluation of plant-based coagulants for turbidity removal and coagulant dosage prediction using machine learning

ABSTRACT This study investigates the use of six plant-based coagulants – Acacia erioloba, Ricinodendron rautanenii, Schinziophyton rautanenii, Peltophorum africanum, Delonix regia, and Maerua angolensis for the removal of turbidity from wastewater effluent. The coagulants were characterized using Scanning Electron Microscopy (SEM) to determine morphological structure, X-ray fluorescence (XRF) to assess chemical composition, and X-ray diffraction to analyse the molecular structure. The coagulation process was evaluated using jar tests with varying coagulant dosages and pH levels. SEM images revealed irregular, rough surfaces, with all materials being amorphous and non-crystalline. Significant levels of essential elements, including iron (Fe), calcium (Ca), sulphur (S), and potassium (K) were revealed. Turbidity removal efficiency fluctuated with pH, showing optimal results under alkaline conditions. Notably, strong negative correlations between pH and turbidity were observed for all coagulants except Peltophorum africanum at a dosage of 20 g/L. Doubling the coagulant volume achieved turbidity reductions between 59% and 92.24%, except for Acacia erioloba and Ricinodendron rautanenii at a dosage of 40 g/L, which showed increased turbidity. The study also employed machine learning techniques to analyse the data and predict the most effective coagulant dosage under different pH conditions. These findings suggest that plant-based coagulants could be viable alternatives to chemical coagulants, with machine learning providing accurate predictions of coagulation performance. Further research is recommended to explore the capabilities of these natural coagulants fully.

Estimation of heart dose in left breast cancer radiotherapy: Assessment of vDIBH feasibility using the supervised machine learning algorithm.

The volunteer deep inspiration breath hold (vDIBH) technique is used to reduce the heart dose in left breast cancer radiotherapy. Many times, it is faced that despite rigorous exercise and training, not all patients get benefited as expected. The primary objective of this study was to develop a machine learning program for prediction of mean heart dose before left breast radiotherapy under vDIBH. The present work is based on the dosimetric parameters of eighty-two left breast cancer patients, who have undergone modified radical mastectomy, enrolled for their radiation treatment. The trained machine learning algorithm employed linear regression to establish a correlation between Haller Index and heart mean dose (HMD) received during the ca left breast cancer radiotherapy. Subsequently, HMD values were used to model the regression relationship with maximum heart distance (MHD). The method adopted is beneficial in patient selection and assessment for suitability of patients' radiotherapy planning under vDIBH treatment technique. For data from 21 test patients, the mean of HMD obtained from the treatment planning system (TPS) and the mean of predicted HMD by developed program were found to be 468.76 cGy and 464.66 cGy, respectively. The present work facilitates precise HMD prediction in left breast cancer radiation therapy even before starting the treatment planning process. Additionally, this program offers suggestions in terms of modifications in treatment settings for even better results of vDIBH techniques if not matches with the anticipated results.