Linear Transformation Of Data Research Articles

Advancing chirality analysis through enhanced enantiomer characterization and quantification via fast Fourier transform capacitance voltammetry

The exploration of the chiral configurations of enantiomers represents a highly intriguing realm of scientific inquiry due to the distinct roles played by each enantiomer (D and L) in chemical reactions and their practical utilities. This study introduces a pioneering analytical methodology, termed fast Fourier transform capacitance voltammetry (FFT-CPV), in conjunction with principal component analysis (PCA), for the identification and quantification of the chiral forms of tartaric acid (TA), serving as a representative model system for materials exhibiting pronounced chiral characteristics. The proposed methodology relies on the principle of chirality, wherein the capacitance signal generated by the adsorption of D-TA and L-TA onto the surface of a platinum electrode (Pt-electrode) in an acidic solution is harnessed. The capacitance voltammograms were meticulously recorded under optimized experimental conditions. To compile the final dataset for the analyte, the average of the FFT capacitance voltammograms of the acidic solution (without the presence of the analyte) was subtracted from those containing the analyte. A distinct arrangement was obtained by employing PCA as a linear data transformation method, representing D-TA and L-TA in a two/three-dimensional space. The outcomes of the study reveal the successful detection of the two chiral forms of TA with a considerable degree of precision and reproducibility. Moreover, the proposed method facilitated the establishment of two linear response ranges for the concentration values of each enantiomer, spanning from 1 to 20 µM, and 50 to 500 µM. The respective detection limits were also determined to be 0.4 µM for L-TA and 1.3 µM for D-TA. These findings underscore the satisfactory sensitivity and efficiency of the proposed method in both qualitative and quantitative assessments of the chiral forms of TA.

A robust cascaded deep neural network for image reconstruction of single plane wave ultrasound RF data

Reconstruction of ultrasound data from single plane wave Radio Frequency (RF) data is a challenging task. The traditional Delay and Sum (DAS) method produces an image with low resolution and contrast, if employed with RF data from only a single plane wave. A Coherent Compounding (CC) method that reconstructs the image by coherently summing the individual DAS images was proposed to enhance the image quality. However, CC relies on a large number of plane waves to accurately sum the individual DAS images, hence it produces high quality images but with low frame rate that may not be suitable for time-demanding applications. Therefore, there is a need for a method that can create a high quality image with higher frame rates. Furthermore, the method needs to be robust against the input transmission angle of the plane wave. To reduce the method’s dependence on the input angle, we propose to unify the RF data at different angles by learning a linear data transformation from different angled data to a common, 0° data. We further propose a cascade of two independent neural networks to reconstruct an image, similar in quality to CC, by making use of a single plane wave. The first network, denoted as “PixelNet”, is a fully Convolutional Neural Network (CNN) which takes in the transformed time-delayed RF data as input. PixelNet learns optimal pixel weights that get element-wise multiplied with the single angle DAS image. The second network is a conditional Generative Adversarial Network (cGAN) which is used to further enhance the image quality. Our networks were trained on the publicly available PICMUS and CPWC datasets and evaluated on a completely separate, CUBDL dataset obtained from different acquisition settings than the training dataset. The results thus obtained on the testing dataset, demonstrate the networks’ ability to generalize well on unseen data, with frame rates better than the CC method. This paves the way for applications that require high-quality images reconstructed at higher frame rates.

A spatial distribution – Principal component analysis (SD-PCA) model to assess pollution of heavy metals in soil

With the rapid development of urbanization, heavy metal pollution of soil has received great attention. Over-enrichment of heavy metals in soil may endanger human health. Assessing soil pollution and identifying potential sources of heavy metals are crucial for prevention and control of soil heavy metal pollution. This study introduced a spatial distribution - principal component analysis (SD-PCA) model that couples the spatial attributes of soil pollution with linear data transformation by the eigenvector-based principal component analysis. By evaluating soil pollution in the spatial dimension it identifies the potential sources of heavy metals more easily. In this study, soil contamination by eight heavy metals was investigated in the Lintong District, a typical multi-source urban area in Northwest China. In general, the soils in the study area were lightly contaminated by Cr and Pb. Pearson correlation analysis showed that Cr was negatively correlated with other heavy metals, whereas the spatial autocorrelation analysis revealed that there was strong association in the spatial distribution of eight heavy metals. The aggregation forms were more varied and the correlation between Cr contamination and other heavy metals was lower. The aggregation forms of Mn and Cu, Zn and Pb, on the other hand, were remarkably comparable. Agriculture was the largest pollution source, contributing 65.5 % to soil pollution, which was caused by the superposition of multiple heavy metals. Additionally, traffic and natural pollution sources contributed 17.9 % and 11.1 %, respectively. The ability of this model to track pollution of heavy metals has important practical significance for the assessment and control of multi-source soil pollution.

Saliency-Based Multilabel Linear Discriminant Analysis.

Linear discriminant analysis (LDA) is a classical statistical machine-learning method, which aims to find a linear data transformation increasing class discrimination in an optimal discriminant subspace. Traditional LDA sets assumptions related to the Gaussian class distributions and single-label data annotations. In this article, we propose a new variant of LDA to be used in multilabel classification tasks for dimensionality reduction on original data to enhance the subsequent performance of any multilabel classifier. A probabilistic class saliency estimation approach is introduced for computing saliency-based weights for all instances. We use the weights to redefine the between-class and within-class scatter matrices needed for calculating the projection matrix. We formulate six different variants of the proposed saliency-based multilabel LDA (SMLDA) based on different prior information on the importance of each instance for their class(es) extracted from labels and features. Our experiments show that the proposed SMLDA leads to performance improvements in various multilabel classification problems compared to several competing dimensionality reduction methods.

Blind source separation for non-stationary random fields

Regional data analysis is concerned with the analysis and modeling of measurements that are spatially separated by specifically accounting for typical features of such data. Namely, measurements in close proximity tend to be more similar than the ones further separated. This might hold also true for cross-dependencies when multivariate spatial data is considered. Often, scientists are interested in linear transformations of such data which are easy to interpret and might be used as dimension reduction. Recently, for that purpose spatial blind source separation (SBSS) was introduced which assumes that the observed data are formed by a linear mixture of uncorrelated, weakly stationary random fields. However, in practical applications, it is well-known that when the spatial domain increases in size the weak stationarity assumptions can be violated in the sense that the second order dependency is varying over the domain which leads to non-stationary analysis. In our work we extend the SBSS model to adjust for these stationarity violations, present three novel estimators and establish the identifiability and affine equivariance property of the unmixing matrix functionals defining these estimators. In an extensive simulation study, we investigate the performance of our estimators and also show their use in the analysis of a geochemical dataset which is derived from the GEMAS geochemical mapping project.

An enhanced fault detection method for centrifugal chillers using kernel density estimation based kernel entropy component analysis

In building automation system, timely detecting the operational faults in water chillers is crucial to system operation management. As a linear data transformation technique, the performance of principal component analysis (PCA) based chiller fault detection method is limited, especially for the incipient ones, due to the nonlinearities in chillers. In addition, the control limit for the monitoring statistic, such as squared prediction error (SPE), is determined under the Gaussian assumption of the score variables, which can hardly be satisfied in water chillers. Therefore, an enhanced fault detection method with the application of kernel density estimation (KDE) and kernel entropy component analysis (KECA) algorithm is reported in this paper. Cauchy-Schwarz (CS) divergence was evaluated as monitoring statistic to measure the cosine of the angle between two data sets after being projected onto a dominated subspace, and then adopted as an index for dissimilarity. KDE with its bandwidth being optimized was also applied to estimate the distribution of the CS divergence, so that the control limit for fault monitoring could be determined. The proposed KDE based KECA-CS method was validated using the experimental data from ASHRAE RP-1043, and further compared to the PCA -SPE, kernel principal component analysis (KPCA)-SPE, and KECA-SPE methods. Results showed that the best performance could be realized when using the KDE based KECA-CS method. The reported fault detection ratio was over 68% for the seven typical chiller faults even at their corresponding least severity level. The average fault detection accuracy was over 90%.

Incipient Fault Detection for Air Brake System of High-Speed Trains

The ideal air brake system is to provide accurate brake pressures, which ensures the normal braking performance of high-speed trains. For the high-speed trains in China, a univariate monitoring strategy has been adopted to monitor the state of each brake cylinder of the air brake system separately. In practice, this kind of strategy is effective for faults with large amplitude, but it is always ineffective for all kinds of incipient faults with small amplitude. Due to nonstationary and multimode properties inherent in air brake systems, conventional detection methods are difficult to be used. In this article, we develop novel detection strategies for three kinds of incipient faults in the air brake systems, including sensor deviation fault, leakage fault, and brake cylinder component fault. Starting from the characteristics of actual data from air brake systems, some variables with high level of consistency are selected. Then, a linear data transformation is conducted to obtain approximate stationary properties, and all the data in multiple modes are mapped into a compact domain. In addition, a fault-to-signal ratio (FSR) is proposed to reflect the sensitivity of the proposed detection statistic to the fault. In particular, the multistage control limits and sliding window strategies are integrated into the detection statistic, which effectively lowers the control limits and enhances robustness to noise, respectively. Finally, experiments on the braking test platform are provided to show the effectiveness of our proposed strategies.

Predicting the likelihood of need for future keratoplasty intervention using artificial intelligence

ObjectiveTo apply artificial intelligence (AI) for automated identification of corneal condition and prediction of the likelihood of need for future keratoplasty intervention from optical coherence tomography (OCT)-based corneal parameters. DesignCohort study. ParticipantsWe collected 12,242 corneal OCT images from 3162 subjects using CASIA OCT Imaging Systems (Tomey, Japan). We included 3318 measurements collected at the baseline visit of each patient. A total of 333 eyes had post-operative penetrating keratoplasty (PKP), lamellar keratoplasty (LKP), deep anterior keratoplasty (DALK), descemet's stripping automated endothelial keratoplasty (DSAEK) or descemet's membrane endothelial keratoplasty (DMEK) intervention. MethodWe developed a pipeline including linear and nonlinear data transformations followed by unsupervised machine learning and applied on corneal parameters from the baseline visit of each patient. Five non-overlapping clusters of eyes were identified. Post hoc analyses revealed that clusters corresponded to different likelihoods of need for future keratoplasty. These clusters on a 2-dimensional map can be used by clinicians and surgeons to identify patients with higher risk of need for future keratoplasty intervention. Main outcome measuresThe likelihood of the need for future surgery. ResultsThe mean age of participants was 69.7 (standard deviation; SD = 16.1) and 59% were female. The normalized likelihood of need for future corneal keratoplasty intervention for eyes mapped onto clusters one to five were 2.2%, 1.0%, 33.1%, 32.7%, and 31.0%, respectively. ConclusionsThe AI system can assist the (cornea) surgeon in identifying those patients who may be at higher risk for future keratoplasty using comprehensive corneal shape, thickness, and elevation parameters. Future research utilizing independent datasets is necessary to validate the proposed system.

Analysis of neuronal ensemble activity reveals the pitfalls and shortcomings of rotation dynamics

Back in 2012, Churchland and his colleagues proposed that “rotational dynamics”, uncovered through linear transformations of multidimensional neuronal data, represent a fundamental type of neuronal population processing in a variety of organisms, from the isolated leech central nervous system to the primate motor cortex. Here, we evaluated this claim using Churchland’s own data and simple simulations of neuronal responses. We observed that rotational patterns occurred in neuronal populations when (1) there was a temporal sequence in peak firing rates exhibited by individual neurons, and (2) this sequence remained consistent across different experimental conditions. Provided that such a temporal order of peak firing rates existed, rotational patterns could be easily obtained using a rather arbitrary computer simulation of neural activity; modeling of any realistic properties of motor cortical responses was not needed. Additionally, arbitrary traces, such as Lissajous curves, could be easily obtained from Churchland’s data with multiple linear regression. While these observations suggest that temporal sequences of neuronal responses could be visualized as rotations with various methods, we express doubt about Churchland et al.’s bold assessment that such rotations are related to “an unexpected yet surprisingly simple structure in the population response”, which “explains many of the confusing features of individual neural responses”. Instead, we argue that their approach provides little, if any, insight on the underlying neuronal mechanisms employed by neuronal ensembles to encode motor behaviors in any species.

Learning a metric when clustering data points in the presence of constraints

Learning an appropriate distance measure under supervision of side information has become a topic of significant interest within machine learning community. In this paper, we address the problem of metric learning for constrained clustering by considering three important issues: (1) considering importance degree for constraints, (2) preserving the topological structure of data, and (3) preserving some natural distribution properties in the data. This work provides a unified way to handle different issues in constrained clustering by learning an appropriate distance measure. It has modeled the first issue by injecting the importance degree of constraints directly into an objective function. The topological structure of data is preserved by minimizing the reconstruction error of data in the target space. Finally we addressed the issue of preserving natural distribution properties in the data by using the proximity information of data. We have proposed two different methods to address the above mentioned issues. The first approach learns a linear transformation of data into a target space (linear-model) and the second one uses kernel functions to learn an appropriate distance measure (non-linear-model). Experiments show that considering these issues significantly improves clustering accuracy.

Detection of short-term urban land use changes by combining SAR time series images and spectral angle mapping

Rapid urban sprawl and re-construction of old towns have been leading to great changes of land use in cities of China. To witness short-term urban land use changes, rapid or real time remote sensing images and effective detection methods are required. With the availability of short repeat cycle, relatively high spatial resolution, and weather-independent Synthetic Aperture Radar (SAR) remotely sensed data, detection of short-term urban land use changes becomes possible. This paper adopts newly released Sentinel-1 SAR data for urban change detection in Tianhe District of Guangzhou City in Southern China, where dramatic urban redevelopment practices have been taking place in past years. An integrative method that combines the SAR time series data and a spectral angle mapping (SAM) was developed and applied to detect the short-term land use changes. Linear trend transformations of the SAR time series data were first conducted to reveal patterns of substantial changes. Spectral mixture analysis was then conducted to extract temporal endmembers to reflect the land development patterns based on the SAR backscattering intensities over time. Moreover, SAM was applied to extract the information of significant increase and decrease patterns. The results of validation and method comparison showed a significant capability of both the proposed method and the SAR time series images for detecting the short-term urban land use changes. The method received an overall accuracy of 78%, being more accurate than that using a bi-temporal image change detection method. The results revealed land use conversions due to the removal of old buildings and their replacement by new construction. This implies that SAR time series data reflects the spatiotemporal evolution of urban constructed areas within a short time period and this study provided the potential for detecting changes that requires continuously short-term capability, and could be potential in other landscapes.

Clutter Parameter Estimation Based on Indirect Algorithms

Indirect algorithms for clutter parameter estimation using linear transformations of initial data have been synthesized by the maximum likelihood method. The likelihood function was introduced for the input samples transformed in accordance with the sum-difference algorithms. The estimation algorithms of the interperiod correlation and Doppler phase shift coefficients derived by solving the appropriate likelihood equations do not contain the traditional complex multiplication operation over input data. The functional block diagram of the estimator of appropriate clutter parameters that can be used in adaptive rejection filters is presented. The estimation accuracy analysis of required parameters of clutter is performed depending on the size of training sample and correlation properties of clutter. The comparison of simulation and calculation results for indirect and direct algorithms revealed their complete match and confirmed the asymptotic efficiency of resultant estimates and the equivalence of indirect and direct algorithms.

Surface Roughness Estimation for Turning Operation Based on Different Regression Models Using Vibration Signals

On machined parts, major indication of surface quality is surface roughness and also surface quality is one of the most specified customer requirements. In the turning process, the importance of machining parameter choice is enhancing, as it controls the required surface quality. To obtain the better surface quality, the most essential control parameters are tool overhang and tool geometry in turning operations. The goal of this study was to develop an empirical multiple regression models for prediction of surface roughness (Ra) from the input variables in finishing turning of 42CrMo4 steel. The main input parameters of this model are tool overhang and tool geometry such as tool nose radius, approaching angle, and rake angle in negative direction. Regression analysis with linear, quadratic and exponential data transformation is applied so as to find the best suitable model. The best results according to comparison of models considering determination coefficient (R 2 ) are achieved with quadratic regression model. In addition, tool nose radius was determined as the most effective parameter on turning by variance analysis (ANOVA). Cutting experiments and statistical analysis demonstrate that the model developed in this work produces smaller errors than those from some of the existing models and have a satisfactory goodness in all three models construction and verification .

Keratoconus severity identification using unsupervised machine learning.

We developed an unsupervised machine learning algorithm and applied it to big corneal parameters to identify and monitor keratoconus stages. A big dataset of corneal swept source optical coherence tomography (OCT) images of 12,242 eyes acquired from SS-1000 CASIA OCT Imaging Systems in multiple centers across Japan was assembled. A total of 3,156 eyes with valid Ectasia Status Index (ESI) between zero and 100% were selected for the downstream analysis. Four hundred and twenty corneal topography, elevation, and pachymetry parameters (excluding ESI Keratoconus indices) were selected. The algorithm included three major steps. 1) Principal component analysis (PCA) was used to linearly reduce the dimensionality of the input data from 420 to eight significant principal components. 2) Manifold learning was used to further reducing the selected principal components nonlinearly to two eigen-parameters. 3) Finally, a density-based clustering was applied to the eigen-parameters to identify eyes with keratoconus. Visualization of clusters in 2-D space was used to validate the quality of learning subjectively and ESI was used to assess the accuracy of the identified clusters objectively. The proposed method identified four clusters; I: a cluster composed of mostly normal eyes (224 eyes with ESI equal to zero, 23 eyes with ESI between five and 29, and nine eyes with ESI greater than 29), II: a cluster composed of mostly healthy eyes and eyes with forme fruste keratoconus (1772 eyes with ESI equal to zero, 698 eyes with ESI between five and 29, and 117 eyes with ESI greater than 29), III: a cluster composed of mostly eyes with mild keratoconus stage (184 eyes with ESI greater than 29, 74 eyes with ESI between five and 29, and 6 eyes with ESI equal to zero), and IV: a cluster composed of eyes with mostly advanced keratoconus stage (80 eyes had ESI greater than 29 and 1 eye had ESI between five and 29). We found that keratoconus status and severity can be well identified using unsupervised machine learning algorithms along with linear and non-linear corneal data transformation. The proposed method can better identify and visualize the keratoconus stages.

Simultaneous determination of alpha-, beta- and gamma-hydroxybutyric acids in micro-pulverized human hair by GC-MS: Method development, validation and application

An analytical method was developed and validated for the simultaneous determination of alpha, beta and gamma-hydroxybutyric acids (AHB, BHB and GHB, respectively) in human hair. Hair samples (10 mg) were pulverized and submitted to extraction using methanol (1 mL) for 10 min. The extract was centrifuged, filtered, and evaporated to dryness at 40 °C under a gentle N2 flow, dissolved in ethyl acetate (0.050 mL) and derivatized using 0.050 mL of BSTFA containing 1% TMCS, at °C for 40 min. The derivatives were determined by GC/MS. Aliquots of natural hair (“blank“) samples after water extraction (3 steps) were pulverized, spiked with AHB, BHB, GHB or d6-GHB and used for method validation. Figures of merit showed the method’s feasibility. The LOQ values were 1 ng mg−1 for BHB and AHB and 0.6 ng mg−1 for GHB, and mean recoveries were 54%, 69% and 75% for AHB, BHB and GHB respectively. Precision and bias were better than 15% and 20% respectively. Calibration curves (LOQ to 20 ng mg−1 for AHB and BHB and LOQ to 16 ng mg−1 for GHB) obtained in 5 days were pooled for statistical analysis, but the data were heteroscedastic, as demonstrated by the White test. Therefore, weighted linear regression and data transformations were applied and the best results were obtained using the Box-Cox transformation [c′ = (cλ-1)/λ] for GHB and log transformation for AHB and BHB. Authentic hair samples were collected from the posterior head vertex of 23 volunteers (16 females and 7 males). Four of them declared having type-2 diabetes (T2D). Only female volunteers reported hair treatments such as dyeing, flat ironing, straightening and/or bleaching. Dyeing appeared not to affect GHB levels in hair, but no data were found in the literature about other cosmetic treatments. GHB levels varied from < LOQ to 1.5 ng mg−1. BHB levels varied from < LOQ to 3.4 ng mg−1 and from < LOQ to 1.8 ng mg−1 in non-diabetic volunteers with or without family history of T2D, respectively. However, in diabetics, BHB levels varied from 3.4 to 5.2 ng mg−1. The levels of AHB varied from < LOQ to 1.2 ng mg−1 in non-diabetic individuals, and from 1.1 to 6.2 ng mg−1 in diabetics. Our results showed that AHB levels in hair have potential to as a biomarker for T2D, but this needs to be further studied with larger groups of volunteers. To our knowledge, this is the first published method for the determination of AHB and BHB in human hair.

Graph autoencoder-based unsupervised feature selection with broad and local data structure preservation

Abstract Feature selection is a dimensionality reduction technique that selects a subset of representative features from high-dimensional data by eliminating irrelevant and redundant features. Recently, feature selection combined with sparse learning has attracted significant attention due to its outstanding performance compared with traditional feature selection methods that ignores correlation between features. These works first map data onto a low-dimensional subspace and then select features by posing a sparsity constraint on the transformation matrix. However, they are restricted by design to linear data transformation, a potential drawback given that the underlying correlation structures of data are often non-linear. To leverage a more sophisticated embedding, we propose an autoencoder-based unsupervised feature selection approach that leverages a single-layer autoencoder for a joint framework of feature selection and manifold learning. More specifically, we enforce column sparsity on the weight matrix connecting the input layer and the hidden layer, as in previous work. Additionally, we include spectral graph analysis on the projected data into the learning process to achieve local data geometry preservation from the original data space to the low-dimensional feature space. Extensive experiments are conducted on image, audio, text, and biological data. The promising experimental results validate the superiority of the proposed method.

Classification of micro-calcification in mammograms using scalable linear Fisher discriminant analysis

Breast cancer is one of the major causes of death in women. Computer Aided Diagnosis (CAD) systems are being developed to assist radiologists in early diagnosis. Micro-calcifications can be an early symptom of breast cancer. Besides detection, classification of micro-calcification as benign or malignant is essential in a complete CAD system. We have developed a novel method for the classification of benign and malignant micro-calcification using an improved Fisher Linear Discriminant Analysis (LDA) approach for the linear transformation of segmented micro-calcification data in combination with a Support Vector Machine (SVM) variant to classify between the two classes. The results indicate an average accuracy equal to 96% which is comparable to state-of-the art methods in the literature.Graphical Classification of Micro-calcification in Mammograms using Scalable Linear Fisher Discriminant Analysis

Testing for Panel Unit Roots under General Cross-sectional Dependence

In this article, we generalize four tests of multivariate linear hypothesis to panel data unit root testing. The test statistics are invariant to certain linear transformations of data and therefore simulated critical values may conveniently be used. It is demonstrated that all four tests remains well behaved in cases of where there are heterogeneous alternatives and cross-correlations between marginal variables. A Monte Carlo simulation is included to compare and contrast the tests with two well-established ones.

Transformation of feature space based on Fisher’s linear discriminant

Linear transformation of data in multidimensional feature space based on Fisher's criterion is considered. The case of two classes with arbitrary distributions is studied. We derived expressions for recurrent calculation of weight vectors which form new features. Example offered shows that the newly found features which represent the data more accurately make it possible to achieve linear separability of classes which remains impossible using the technique of principal components and the classic Fisher's linear discriminant.

Abstract B1-36: Population pharmacodynamics: Mechanism-based modeling of receptor tyrosine kinase networks in cancer

Abstract The purpose of our study was to improve our ability to predict response to therapy by integrating mechanistic kinetic data of protein interactions with patient-specific gene and protein expression data. The genetic heterogeneity of cancer results in patient-to-patient variability that makes it difficult to predict whether the patient will respond to a treatment. Molecular biomarkers such as the expression levels of genes or proteins have proven useful in limited cases, for example ErbB2 and trastuzumab, particularly in the epidermal growth factor (EGF) family, but these kinds of markers are still unable to accurately predict all responsive patients. These biomarkers are typically univariate and linear, whereas multivariate, nonlinear biomarkers are needed to adequately describe the network of molecular interactions targeted by a drug. Here, we present a method for building multivariate biomarkers that incorporate BOTH patient-specific transcriptomic/proteomic data AND detailed mechanistic models of the nonlinear interactions between ligands and receptors. The mechanistic models that we use are computational pharmacodynamic models, with multiple compartments, each with multiple cell types that express the ligands and receptors under investigation. The models incorporate detailed protein-protein interaction networks to simulate the complex dynamics of growth factor families and their receptors. By integrating this molecular detail into whole-body simulations with tumors, we can evaluate many different therapeutic approaches – different drugs, doses, schedules, and routes of administration. Our models make predictions of the dynamics of receptor tyrosine kinase activity and of key blood-borne biomarkers following therapeutic intervention. These predictions can and have been validated against clinical experimental data. We applied the method to the EGFR/ErbB family in breast cancer, using individualized data from The Cancer Genome Atlas (TCGA), and showed that the personalized models were able to capture the observed variability in receptor phosphorylation. Before the addition of drugs, the models behaved in a relatively monotonic fashion, with signaling outputs closely following the expression of the key ligands. However, the response to the addition of drugs was much more complex; the baseline expression of genes/proteins was not as good a predictor of the response. We simulated the addition of three antibody drugs that each target one of EGFR, HER2, and HER3. We applied principal component analysis to characterize the output of our simulations – post-therapy time course response. We derived metrics that accounted for both target-specific and off-target effects. We then used multivariate supervised learning methods to develop predictive biomarkers. We found that biomarkers derived from gene expression data were outperformed by biomarkers derived from simulated baseline tumor behavior (i.e. that combined quantitative mechanistic information with gene expression data). This suggested that linear transformations of transcriptomic and proteomic data may not be adequate for predicting drug response; instead, the nonlinear mechanism-based transformation that is central to the computational model is more predictive. In addition, for each of the antibodies investigated, the incorporation of mechanistic protein interactions resulted in the identification of off-target effects with high inter-individual heterogeneity that have the potential to significantly blunt the response. In conclusion, transforming individualized expression data through a detailed kinetic model of molecular interactions improves predictiveness of treatment response. Citation Format: Robert Joseph Bender, Feilim Mac Gabhann. Population pharmacodynamics: Mechanism-based modeling of receptor tyrosine kinase networks in cancer. [abstract]. In: Proceedings of the AACR Special Conference on Computational and Systems Biology of Cancer; Feb 8-11 2015; San Francisco, CA. Philadelphia (PA): AACR; Cancer Res 2015;75(22 Suppl 2):Abstract nr B1-36.