TD Model Research Articles

English-Chinese Translation Quality Assessment Based on Phrase Statistical Machine Translation Decoding Algorithm

Machine learning translation is the automated process of translating text from one language to another using computational algorithms and statistical models. neural network-based approaches, particularly using models like sequence-to-sequence (Seq2Seq) with attention mechanisms, have shown remarkable performance improvements in translation quality. This paper proposes a Statistical Stochastic Gradient Machine Translation Decoding (SSGM-TD) algorithm for the English–Chinese translation for the quality assessment. The proposed SSGM-TD model uses statistical analysis for the estimation and evaluation of the features for the computation of variables. The proposed SSGM – TD model estimates the stochastic gradient with the regression analysis for the feature estimation. The developed SSGM-TD model is implemented with the machine learning model for the automated translation of the English–Chinese languages. The simulation analysis is performed for the evaluation of the quality assessment in the translation process. The detailed evaluation is conducted using various metrics, including BLEU and METEOR scores, offering quantitative insights into the algorithm's performance. The classification process of the SSGM-TD algorithm is examined, revealing its proficiency in correctly classifying positive and negative instances. Precision, recall, and F1 score metrics provide a significant evaluation of the algorithm's classification capabilities. The decoding results and quality assessments are presented with providing a comprehensive view of the algorithm's strengths and potential areas for improvement. The quality assessments incorporate both quantitative metrics and human evaluations, ensuring a holistic understanding of the algorithm's translation capabilities. The consistency between automated metrics and human assessments underscores the algorithm's commendable performance in maintaining semantic accuracy and linguistic coherence.

Resin screening and process optimization for erythritol mother liquor chromatographic separation

In order to improve the utilization value of the erythritol mother liquor, the separation and purification of the erythritol mother liquor was selected in this study. The selected chromatographic separation programme for erythritol crystallizing mother liquor is as follows: Firstly, erythritol is resolved from mannitol and arabitol with DTF-01Ca (Suqing Group) resin and then mannitol is resolved from arabitol with 99Ca/320 (Dowex) resin. At the same time, the chromatographic conditions of the DTF-01Ca (Suqing Group) and 99Ca/320 (Dowex) resins were optimized, resulting in an optimal separation temperature and mobile phase flow rate of 70 °C, 10 ml/min. On this basis, a single-column chromatographic model was used to calculate the TD model parameter ( N ) and the mass transfer coefficient ( k m ) of the separation of erythritol mother liquor by DTF-01Ca (Suqing Group) and 99Ca/320 (Dowex) resins. The adsorption isotherms, TD model parameter ( N ) and the mass transfer coefficient ( k m ) provides data references for the design and operation of the simulated moving beds (SMB) separation system for the industrial-scale separation of erythritol crystallizing mother liquor.

Translational PK-PD/TD modeling of antitumor effects and peripheral neuropathy in gemcitabine and nab-paclitaxel chemotherapy from xenograft mice to patients for optimal dose and schedule.

Gemcitabine and nab-paclitaxel (GnP) treatment, the standard first-line chemotherapy for unresectable pancreatic cancer, often causes peripheral neuropathy (PN). To develop alternative dosing strategies to avoid severe PN, understanding the relationship between pharmacokinetics (PK) and pharmacodynamics/toxicodynamics (PD/TD) is necessary. We established a PK-PD/TD model of GnP treatment to develop an optimal dose schedule. A mouse xenograft model of human pancreatic cancer was generated to measure drug concentrations in the plasma and tumor, antitumor effects, and PN after GnP treatment. The Simeoni tumor growth inhibition model with tumor concentrations and empirical indirect response models were used for the PD and TD models, respectively. Clinical outcomes were predicted with reported population estimates of PK parameters in cancer patients. The PK-PD/TD model simultaneously described the observed tumor volume and paw withdrawal frequency in the von Frey test. For the standard GnP regimen, the model predicted clinical overall response (75.1%), which was overestimated compared to that in a recent phase II study (42.1%) but lower than the observed disease control rate (96.5%). Model simulation showed that dose reduction to less than 40% GnP dose was not effective; a change of dose schedule from every week for 3weeks to every 2weeks was a more favorable approach than dose reduction to 60% every week. The PK-PD/TD model-based translational approach provides a guide for optimal dose determination to avoid severe PN while maintaining antitumor effects during GnP chemotherapy. Further research is needed to enhance its applicability and potential for combination chemotherapy regimens.

Implementation evaluation of a teledermatology virtual clinic at an academic medical center

BackgroundTeledermatology (TD) is an evidence-based practice that may increase access to dermatologic care. We sought to use the Exploration, Preparation, Implementation, and Sustainment (EPIS) and the Reach, Efficacy, Adoption, Implementation, and Maintenance (RE-AIM) frameworks to evaluate implementation of TD at Duke.MethodsThe EPIS and RE-AIM frameworks were deployed to design and implement a TD program that leveraged the strengths of the Duke University Health System and addressed previously reported barriers to implementation of store-and-forward and synchronous TD models. In the resultant hybrid TD model, trained primary care providers (PCPs) sent e-comm referrals with clinical and dermatoscopic images to dermatology. These e-consults were reviewed asynchronously and patients were scheduled for a synchronous video visit with dermatology within days. Dermatologists managed the patient plan. This hybrid TD model was piloted at four primary care clinics. Pertinent outcomes from a TD-adapted RE-AIM framework were tracked using electronic health record data. Patient satisfaction was assessed using a post-video visit survey (n = 18). Implementation barriers and facilitators were also collected through provider surveys (n = 24 PCPs, n = 10 dermatologists, n = 10 dermatology residents).ResultsAt four PCP clinics throughout 9/1/2021–4/30/2022, there were 218 TD referrals. Video visits occurred on average 7.5 ± 0.5 days after referral and 18/18 patients completing the post-visit survey were satisfied. Adoption varied between clinics, with one placing 22% of all dermatology referrals as TD and another placing 2%. The primary PCP barriers to TD were time burdens, lack of fit in clinic flow, and discomfort with image taking. Top-endorsed potential facilitating interventions included allowing for rash referrals without dermoscopy and assurance for clinical evaluation within 3 days.ConclusionsThe use of implementation science frameworks allowed for identification of system and contextual strengths which informed the hybrid TD pilot. Barriers and facilitating interventions will provide guidance for expansion and ongoing maintenance of TD.

Shared GABA transmission pathology in dopamine agonist- and antagonist-induced dyskinesia

Dyskinesia is involuntary movement caused by long-term medication with dopamine-related agents: the dopamine agonist 3,4-dihydroxy-L-phenylalanine (L-DOPA) to treat Parkinson's disease (L-DOPA-induced dyskinesia [LID]) or dopamine antagonists to treat schizophrenia (tardive dyskinesia [TD]). However, it remains unknown why distinct types of medications for distinct neuropsychiatric disorders induce similar involuntary movements. Here, we search for a shared structural footprint using magnetic resonance imaging-based macroscopic screening and super-resolution microscopy-based microscopic identification. We identify the enlarged axon terminals of striatal medium spiny neurons in LID and TD model mice. Striatal overexpression of the vesicular gamma-aminobutyric acid transporter (VGAT) is necessary and sufficient for modeling these structural changes; VGAT levels gate the functional and behavioral alterations in dyskinesia models. Our findings indicate that lowered type 2 dopamine receptor signaling with repetitive dopamine fluctuations is a common cause of VGAT overexpression and late-onset dyskinesia formation and that reducing dopamine fluctuation rescues dyskinesia pathology via VGAT downregulation.

Interpretation of experimental findings on the structure of glass in the CaO–MoO3–P2O5 system using a thermodynamic model including oxidation–reduction equilibria

AbstractThis work extends the thermodynamic model of associated solutions used in the past to describe the structure and properties of glasses to the area of complex multicomponent glasses with polyvalent elements, where it has not been applied until now either due to the absence of Gibbs energies of formation of the necessary compounds or due to oxidation–reduction equilibrium in the presence of a gas phase containing oxygen. While the fitting of unknown Gibbs energies based on experimental data has already been applied to some extent in our previous work, the implementation of redox is, to the best of our knowledge, new. Four concentration series were taken from the published data from the glass-forming ternary system CaO–MoO3−P2O5: A) xMoO3−(0.5–0.75x)CaO−(0.5–0.25x)P2O5; B) xMoO3−(0.5–0.875x)CaO−(0.5–0.125x)P2O5; C) xMoO3−(0.5−x)CaO−0.5P2O5; M) xMoO3−(1−x)P2O, for which the distributions of Qn units were also published (Q denotes the PO4 tetrahedral unit with n bridging oxygens) by the 31P MAS NMR method and the Mo[V]/ΣMo fraction by the ESR method [Černošek et al. (J Solid State Chem 303:122522, 2021); Holubová et al., (J Non-Cryst Solids 607:122222, 2023)]. The following compounds were considered in the TD model: P2O5, CaO, Mo[VI]O3, Ca(PO3)2, Ca2P2O7, (Mo[VI]O2)(PO3)2, (Mo[VI]O2)2(P2O7), (Mo[VI]O2)3(PO4)2, (Mo[V]O)2(PO3)2(P2O7), (Mo[V]O)PO4. All except the hypothetical compound (Mo[VI]O2)3(PO4)2 exist, and their structure is known. Binary phosphate compounds with molybdenum lack Gibbs energies of formation. Therefore, one of the series, namely A, was used to determine these energies by nonlinear regression with the help of a genetic algorithm, without/with redox, and then the distribution of Qn units and the fraction of Mo[V]/ΣMo was predicted for the remaining series. It was found that the distribution of Qn units can be described by the TD model with redox only. During the reduction of molybdenum, the distribution of Qn unit’s changes, and thus also the connectivity of the phosphate network, for example, according to the reactions: (MoO2)2(P2O7)—> 2(MoO)PO4 + 1/2O2, in which Q1—> Q0 and 2(MoO2)(PO3)2—> (MoO)2(PO3)2(P2O7) + 1/2O2 in which Q2—> Q1. Despite the fact that the TD model with redox gives excellent agreement in the case of the Qn distribution, the agreement with the ESR measurements of the Mo[V]/ΣMo ratio is not good. The TD model predicts significantly more pentavalent molybdenum in the glass.

Model-Agnostic Meta-Learning for Fast Text-Dependent Speaker Embedding Adaptation

By constraining the lexical content of input speech, text-dependent speaker verification (TD-SV) offers more reliable performance than text-independent speaker verification (TI-SV) when dealing with short utterances. Because speech with constrained lexical content is harder to collect, often TD models are fine-tuned from a TI model using a small target phrase dataset. However, sometimes the target phrase dataset is too tiny for fine-tuning, which is the main obstacle for deploying TD-SV. One solution is to fine-tune the model using medium-size multi-phrase TD data and then deploy the model on the target phrase. Although this strategy does help in some cases, the performance is still sub-optimal because the model is not optimized for the target phrase. Inspired by the recent progress in meta-learning, we propose a three-stage pipeline for adapting a TI model to a TD model for the target phrase. Firstly, a TI model is trained using a large amount of speech data. Then, we use a multi-phrase TD dataset to tune the TI model via model-agnostic meta-learning. Finally, we perform fast adaptation using a small target phrase dataset. Results show that the three-stage pipeline consistently outperforms multi-phrase and target phrase fine-tuning.

A site-specific earthquake ground response analysis using a fault-based approach and nonlinear modeling: The Case Pente site (Sulmona, Italy)

In this paper we present the ground response analyses (GRA) of a site where an industrial facility is planned. Due to its location on an active normal fault system known as a relevant seismic gap, the Mt. Morrone Fault system (MMF), and at the edge of a basin filled with slow velocity continental deposits, a inter-disciplinary and non-standard approach has been applied to assess the seismic input of the dynamic numerical analyses. It includes geological, seismological, geotechnical and engineering contributions. Two fault scenarios, MMF1 and MMF2, were considered and scenario-based (SSHA) and probabilistic (time-dependent, TD, and time-independent, TI) seismic hazard (PSHA) analyses were implemented. Comparison among the spectra corresponding to the 90th percentile of the SSHA statistical distribution and the PSHA average ones shows that the SSHA MMF2 has values similar to the PSHA TD model. The SSHA 90th percentile distribution was selected as target spectra to retrieve the seismic input for GRA. Nonlinear numerical simulations of seismic wave propagation were implemented to derive surface ground motion parameters. GRA acceleration response spectra and their PGA are notably higher, and thus on the side of safety, than those obtained following the Italian code approach for seismic resistant buildings. These results confirm that a scenario-based methodology can better capture the shaking effect in near-field conditions, avoiding possibly unconservative underestimations of the seismic actions and in view of a more robust performance-based approach used by engineers for either new design and/or assessment/retrofit purposes of the built environment.

Explicit Consideration of Temperature Improves Predictions of Toxicokinetic-Toxicodynamic Models for Flupyradifurone and Imidacloprid in Gammarus pulex.

In the face of global climate change, where temperature fluctuations and the frequency of extreme weather events are increasing, it is needed to evaluate the impact of temperature on the ecological risk assessment of chemicals. Current state-of-the-art mechanistic effect models, such as toxicokinetic-toxicodynamic (TK-TD) models, often do not explicitly consider temperature as a modulating factor. This study implemented the effect of temperature in a widely used modeling framework, the General Unified Threshold model for Survival (GUTS). We tested the model using data from toxicokinetic and toxicity experiments with Gammarus pulex exposed to the insecticides imidacloprid and flupyradifurone. The experiments revealed increased TK rates with increasing temperature and increased toxicity under chronic exposures. Using the widely used Arrhenius equation, we could include the temperature influence into the modeling. By further testing of different model approaches, differences in the temperature scaling of TK and TD model parameters could be identified, urging further investigations of the underlying mechanisms. Finally, our results show that predictions of TK-TD models improve if we include the toxicity modulating effect of temperature explicitly.

Effect of dissolved organic matter on the bioavailability and toxicity of cadmium in zebrafish larvae: Determination based on toxicokinetic–toxicodynamic processes

The presence of dissolved organic matter (DOM) strongly influences the bioavailability of metals in aquatic environments; however, the association between the binding activities and the concentrations of DOM compositions is not well documented, leading to uncertainties in metal toxicity assessment. We creatively quantify the mitigation and acceleration effects of DOM compositions on cadmium (Cd) bioaccumulation and toxicity in zebrafish larvae using abiotic ligand (ABLs) and biotic ligand (BLs) in a toxicokinetic–toxicodynamic (TK–TD) model. The BL–TK–TD model could accurately predict the protective effect of fulvic acid while overestimating the complexing capacity of citric acid. The model also could successfully simulate the protective effects of native DOM in most cases from 32 natural water bodies in China. The observed LC50 values of Cd showed a peak effect for the native DOM fraction comprising hydrophilic acidic contents (3.55 ± 0.44 mg L − 1) in natural water from 32 sites. The BL–TK–TD model provides practically useful information to identify the effect of different DOM compositions on metal bioavailability and toxicity in aquatic environments and guides future water management policies aimed at controlling aquatic heavy metal pollution.

Modeling Crossing Conflicts at Unsignalized T-Intersections under Heterogeneous Traffic Conditions

The safety of unsignalized intersections is evaluated by correlating the number of crashes with traffic volume and intersection geometry. However, crash-based safety assessment has known drawbacks related to data quality and coverage. Further, the crash-based safety analysis does not account that not all vehicles interact unsafely. Therefore, the present study develops crossing conflict-based safety performance functions (C-SPFs) for eight urban unsignalized T-intersections with varying intersection geometry. Initially, the crossing conflicts were analyzed using post encroachment time (PET); based on that, they are bifurcated into critical and noncritical conflicts. The C-SPFs were modeled as a function of traffic volume and intersection geometry using the generalized estimating equations with the Tweedie distribution (GEE_TD) regression approach. The results revealed the time of the day, intersection geometry, vehicular composition, and traffic volume of both offending and conflicting approaches as significant variables influencing the number of critical and noncritical crossing conflicts. Further, to check the predictive power of the GEE_TD model, the model errors are compared with those obtained using the negative binomial (NB) model. The result revealed that for both critical and noncritical conflicts, the GEE_TD model has better predictivity (lesser error) than the NB model.

Efficient terahertz absorption gas sensor with Gaussian process regression in time- and frequency-domain

The terahertz region presents promising opportunities for quantitative gas sensing in practical environments thanks to its access to molecular fingerprints and immunity to particulate scattering. With Terahertz time-domain spectroscopy (THz-TDS), molecular absorption and dispersion spectra can be rendered using femtosecond-laser-based interferometric system. However, procurement of accurate spectroscopic parameters in this uncharted spectral region and quantitative reduction of the broadband data remain challenging. We propose and validate two end-to-end machine learning models based on Gaussian process regression (GPR) for efficient THz-TDS sensing in both time- and frequency-domain (TD and FD). Quantitative CO sensing is accomplished for proof-of-concept demonstration of the method. Both TD- and FD-GPR models demonstrate accurate predictions in the presence of measurement noise and ambient interference, and the root-mean-square errors are below 0.6 % with both simulated and experimental data. It is concluded that while the FD model benefits from immunity to dispersion distortions, the TD model noses out for its baseline-free nature and better elimination of temporally introduced noises. This work presents the first experimental demonstration of THz-TDS for quantitative gas sensing exploiting both TD and FD signals, showing promising prospects for real-time, multi-species, multi-parameter sensing under challenging scenarios such as chemically reactive or open-path atmospheric environments.

Metronomic dose-finding approach in oral chemotherapy by experimentally-driven integrative mathematical modeling

In conventional chemotherapy, maximum tolerated dose approach is considered as a first-line medication for cancer treatment in clinics. In contrast to the conventional chemotherapy which has heavy tumor burdens arising from high dose treatment, metronomic chemotherapy (MCT) engages relatively low dose without drug-free breaks, and is recognized as a promising strategy for a long-term management of the disease. Although doxorubicin (DOX), an anthracycline anti-cancer drug, showed a potential of maintenance effect in vitro, further study on in vivo-relevant concentration to achieve tumor suppression with no toxicity is required to apply the MCT in clinicals. Therefore, the objective of this study was to identify an optimal MCT regimen of DOX by determining concentration-response relationships of tumor suppression (pharmacodynamic; PD) and cardiac toxicity (toxicodynamic; TD). Utilizing an oral DOX formulation complexed with deoxycholic acid (DOX/DOCA complex) which has enhanced bioavailability, physiologically-based pharmacokinetic (PBPK) model was linked to TD and PD models to generate drug profiles from the combined PK, TD, and PD parameters. The integrated model was validated for various scenarios of administration route, formulation, dose, and frequency. The established mathematical model facilitated calculations of adequate in vivo-relevant dosages and intervals, suggesting the optimum oral metronomic regimen of DOX. It is expected to serve as a useful guideline for the design and evaluation of oral DOX formulations in future preclinical/clinical studies.

PSXII-7 Milk somatic cells monitoring in Russian Holstein cattle population as a base for determining genetic and genomic variability

Abstract Health traits in dairy cattle have crucial meaning to produce high quality milk. Despite of fertility problems and metabolic disorders in cows, the mastitis has a bigger economic losses influence to include it as selection criteria in cattle breeding. Somatic cell count (SCC) in that case are the good predictor for monitoring udder health cows under whole population level or separate herd. The aim of our study was to assess genetic and genomic components for SCC and their scores (SCS) using experimental dataset by seven herds with the subsequent QTL identification. For six-month observation the 5824 cows with 19786 test-day records were included into analysis. Then EBVs by offspring assessing of 139 genotyped Holstein sires were calculated trough TD Model (BLUPF90) and then it adopted as pseudo-phenotypes for GWAS. After quality control using Plink 1.90, we used ≈39K SNP (Illumina 50K). The average values for SCC and SCS were 351±7 thousands cells/ml and 2.86±0.02 score respectively. Heritability coefficients revealed low genetic variation for SCC – 0.119 and moderate for SCS – 0.211. Daily yield for cows with SCC >1000×103 cells/ml was low by -4.0 kg milk to compare individuals with SCC < 100×103 cells/ml. At the same time lactose content and freezing point were decreasing by 4.93 to 4.69% and -0.635 to -0.618°C. By Cattle QTLdb we identified some causal genes for SCC on BTA3 (ROR1), BTA9 (EZR), BTA13 (OSBPL2,DNAJC5,ZBTB46,MTG2), BTA14 (KHDRBS3) and BTA22 (RBMS3). But more relevant GWAS calls were found for SCS by BTA14 (KCNB2, ZFAT) as QTL associated to the milking speed that has unfavorable genetic correlation with clinical mastitis or SCS. Thereby, genes detected under experimental study, are the valuable and informative markers to implementation genomic selection methods for cattle health in creating Russian bulls’ reference population. The study was funded by RFBR within project No. 20-316-90050

The Seismic Hazard Implications of Declustering and Poisson Assumptions Inferred from a Fully Time-Dependent Model

ABSTRACT We use the Third Uniform California Earthquake Rupture Forecast (UCERF3) epidemic-type aftershock sequence (ETAS) model (UCERF3-ETAS) to evaluate the effects of declustering and Poisson assumptions on seismic hazard estimates. Although declustering is necessary to infer the long-term spatial distribution of earthquake rates, the question is whether it is also necessary to honor the Poisson assumption in classic probabilistic seismic hazard assessment. We use 500,000 yr, M ≥ 2.5 synthetic catalogs to address this question, for which UCERF3-ETAS exhibits realistic spatiotemporal clustering effects (e.g., aftershocks). We find that Gardner and Knopoff (1974) declustering, used in the U.S. Geological Survey seismic hazard models, lowers 2% in 50 yr and risk-targeted ground-motion hazard metrics by about 4% on average (compared with the full time-dependent [TD] model), with the reduction being 5% at 40% in 50 yr ground motions. Keeping all earthquakes and treating them as a Poisson process increases these same hazard metrics by about 3%–12%, on average, due to the removal of relatively quiet time periods in the full TD model. In the interest of model simplification, bias minimization, and consideration of the probabilities of multiple exceedances, we agree with others (Marzocchi and Taroni, 2014) that we are better off keeping aftershocks and treating them as a Poisson process rather than removing them from hazard consideration via declustering. Honoring the true time dependence, however, will likely be important for other hazard and risk metrics, and this study further exemplifies how this can now be evaluated more extensively.

Gravitational interaction through a feedback mechanism

We study the models of Kafri, Taylor and Milburn (KTM) and Tilloy and Di\'osi (TD), both of which implement gravity between quantum systems through a continuous measurement and feedback mechanism. The first model is for two particles, moving in one dimension, where the Newtonian potential is linearized. The second is applicable to any quantum system, within the context of Newtonian gravity. We address the issue of how to generalize the KTM model for an arbitrary finite number of particles. We find that the most straightforward generalisations are either inconsistent or are ruled out by experimental evidence. We also show that the TD model does not reduce to the KTM model under the approximations which define the latter model. We then argue that under the simplest conditions, the TD model is the only viable implementation of a full-Newtonian interaction through a continuous measurement and feedback mechanism.

Multiscale and Translational Quantitative Systems Toxicology, Pharmacokinetic-Toxicodynamic Modeling Analysis for Assessment of Doxorubicin-Induced Cardiotoxicity.

Dose-dependent life-threatening doxorubicin-induced cardiotoxicity (DIC) is a major clinical challenge that needs to be addressed. Here, we developed an integrated multiscale and translational quantitative systems toxicology and pharmacokinetic-toxicodynamic (QST-PK/TD) model for optimization of doxorubicin dosing regimens for early monitoring and minimization of DIC. A QST model was established by exposing human cardiomyocytes, AC16 cells, to doxorubicin over a time course, and measuring the dynamics of intracellular signaling proteins, AC16 cell viability and released biomarkers of cardiomyocyte injury such as the B-type natriuretic peptide (BNP). Experiments were scaled up to a three-dimensional and dynamic (3DD) cell culture system to evaluate DIC under various dosing regimens. The PK determinants of doxorubicin influencing DIC were identified in vitro and then translated to the in vivo setting through hybrid physiologically based PK (PBPK)/TD models using preclinical- and clinical-level data extracted from literature. The developed cellular-level QST model captured well the observed dynamics of intracellular proteins, AC16 cell viability and BNP kinetics. In the 3DD setting, dose fractionation of doxorubicin displayed a significant reduction in cardiotoxicity compared to single intravenous doses with equal exposure, implying doxorubicin peak concentrations as the PK determinant for DIC. The in vivo hybrid PBPK/TD models captured well doxorubicin PK and DIC. Peak doxorubicin concentrations correlated well with acute DIC for dose-fractionated regimens, while maximum 48-h moving average concentrations correlated with DIC for dose-fractionated and long-term infusion regimens in vivo. The developed multiscale and translational QST-PK/TD modeling platform may serve as an in silico tool for assessment of early toxicity and/or efficacy of developmental drugs in vitro.

Comparison of light scattering properties of porous dust particle with connected and unconnected dipoles

In the present work, we propose a comparative study of optical properties in the visual spectral regime of porous dust particles having porosities ranging from 0% (compact particle) to 50% (as much matter as void in a same particle), generated using two different models considering particle as an ensemble of dipoles much smaller than wavelength. One of the models (the touching-dipoles model, named below: ‘‘TD model”) considers a homogeneous structure made up of touching dipoles (that is neighbouring); the dipoles are removed randomly one by one from a compact structure in such a way that the remaining structure is left connected. The other model (the non-touching-dipoles model, named below:’‘non-TD model”) generates porous particles by randomly removing dipoles without constraining the ensemble of dipoles to remain connected. The computations of the optical properties of these disordered particles are performed using Discrete Dipole Approximation (DDA) code. Linear polarization profiles and color (i.e. the ratio between the intensities of the light scattered by particles in the comet at 0.485 ​μm and 0.684 ​μm wavelengths expressed in log scale) curves are shown vs. the scattering angle, and compared for the two models at different porosities. The variation of scattering efficiency factor as a function of the size parameter (X) is also studied to point out sensitivity of light scattering to different pore structures (shape and size), comparing data from particles of same porosity but generated through different processes. The results are compared with Mie results where the effective refractive index for each porosity is calculated using Bruggemann mixing rule. It is observed that light scattering properties of the TD model is not much different from the EMT-Mie model, but the non-TD model differs significantly. These difference could be due to the fact that non-TD model contains a higher number of non-Rayleigh inclusions, as compared to the TD model.

Analysis of Adverse Drug Reactions Identified in Nursing Notes Using Reinforcement Learning.

ObjectivesElectronic Health Records (EHRs)-based surveillance systems are being actively developed for detecting adverse drug reactions (ADRs), but this is being hindered by the difficulty of extracting data from unstructured records. This study performed the analysis of ADRs from nursing notes for drug safety surveillance using the temporal difference method in reinforcement learning (TD learning).MethodsNursing notes of 8,316 patients (4,158 ADR and 4,158 non-ADR cases) admitted to Ajou University Hospital were used for the ADR classification task. A TD(λ) model was used to estimate state values for indicating the ADR risk. For the TD learning, each nursing phrase was encoded into one of seven states, and the state values estimated during training were employed for the subsequent testing phase. We applied logistic regression to the state values from the TD(λ) model for the classification task.ResultsThe overall accuracy of TD-based logistic regression of 0.63 was comparable to that of two machine-learning methods (0.64 for a naïve Bayes classifier and 0.63 for a support vector machine), while it outperformed two deep learning-based methods (0.58 for a text convolutional neural network and 0.61 for a long short-term memory neural network). Most importantly, it was found that the TD-based method can estimate state values according to the context of nursing phrases.ConclusionsTD learning is a promising approach because it can exploit contextual, time-dependent aspects of the available data and provide an analysis of the severity of ADRs in a fully incremental manner.

Abstract 3632: Combination of EGFR antibody with PD-1 pathway inhibitors improves anti-tumor efficacy and enhances intra-tumor immune response in preclinical mouse tumor models

Abstract Necitumumab (EGFR inhibitor) in combination with chemotherapy provides a modest, yet a significant improvement in overall survival over chemotherapy alone in patients with advanced squamous non-small cell lung carcinoma (NSCLC). However, combination therapies targeting EGFR and PD-1 pathway blockers may represent a better way to extend clinical benefit to more cancer patients given that PD-(L)1 antibodies have emerged as a standard of care in NSCLC. Antibodies targeting EGFR have the potential to promote an inflamed tumor microenvironment through engagement of Fc-gamma receptors (FcγR) on innate immune cells resulting in an improved antigen presentation and T cell priming. Therefore, the present study was initiated to understand the combinatorial effect of immune checkpoint (PD-(L)1) inhibitors with necitumumab. Preclinical modeling of EGFR/PD-(L)1 mAb combination in mice is challenging due to the lack of cross-reactivity of necitumumab with mouse EGFR. Syngeneic mouse tumor models widely used to study effects of immunomodulatory agents express low or no EGFR. To overcome this limitation, we used two immunocompetent model systems to study the combination effect of EGFR mAb with PD-1 (RMP1-14) or PD-L1 (178G7) mAbs: 1) genetically engineered mouse model of lung adenocarcinoma (TD model) driven by mutant forms of human EGFR (exon 19 deletion and T790M mutation) and 2) CT26 syngeneic mouse tumor model with ectopic expression of human EGFR (CT26-hEGFR). To engage mouse immune cells more efficiently, a murinized version of necitumumab was generated through antibody engineering, with human EGFR binding Fabs and a mouse Fc backbone. Intratumor immune response was evaluated by immunohistochemistry and a custom-made immune profiling Quantigene Plex (QGP) gene expression panel. In both models, targeting EGFR and PD-1 pathway resulted in the combinatorial antitumor efficacy exemplified by decreased tumor burden compared to the monotherapy groups. QGP analysis of CT26-hEGFR tumor tissue revealed that the combination treatment enhanced intra-tumor immune response exemplified by an upregulation of immune-related genes indicative of T cell infiltration (Cd3e, Cd4, Cd8b1), T cell activation (Ifng, Cd274, Pdcd1lg2, Icos, Tnfrsf4, Tnfrsf18, Cd69, Ido1, Havcr2, Lag3), myeloid cell infiltration (Cd86, Timd4, Vista, Cd68, Mpo, Nos2). Histopathological analysis confirmed an increase in T cell infiltration indicating an improved immune response in the combination therapy group. Taken together, these results provide a rationale for further evaluation of EGFR and PD-(L)1 mAbs in clinical setting. Citation Format: Veena Kandaswamy, Amelie Forest, Marianne Deroose, David A. Schaer, Ting Chen, Shengwu Liu, David Surguladze, Yung-mae Yao, Thompson Doman, Gerald Hall, Kwok-Kin Wong, Michael Kalos, Ruslan D. Novosiadly. Combination of EGFR antibody with PD-1 pathway inhibitors improves anti-tumor efficacy and enhances intra-tumor immune response in preclinical mouse tumor models [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 3632.