Signature Matrix Research Articles

Remaining Useful Life Prediction of Aircraft Flap Control System with Mode Transition

Hybrid system is a dynamic system with mode transition, and the degradation behavior of its components is varied with different operating modes. However, the recent prognosis methodologies treat the degradation of different operating modes as the same and ignore their differences or even do not consider degradation in idle mode. These methods have a significant error on remaining useful life (RUL) prediction of aircraft flap control system, which is a typical hybrid system and also vital for flight safety. In response to this problem, this paper proposes a method based on bond graph, which integrates the mode transition information into global analytical redundancy relations. Here, the mode-dependent fault signature matrix is constructed and different degradation models are identified. Finally, simulation experiments are done, and the RUL prediction value of the proposed method is closer to the preset value than that of the method without considering degradation of the idle mode.

Computational Intelligence-Based Prognosis for Hybrid Mechatronic System Using Improved Wiener Process

In this article, a fast krill herd algorithm is developed for prognosis of hybrid mechatronic system using the improved Wiener degradation process. First, the diagnostic hybrid bond graph is used to model the hybrid mechatronic system and derive global analytical redundancy relations. Based on the global analytical redundancy relations, the fault signature matrix and mode change signature matrix for fault and mode change isolation can be obtained. Second, in order to determine the true faults from the suspected fault candidates after fault isolation, a fault estimation method based on adaptive square root cubature Kalman filter is proposed when the noise distributions are unknown. Then, the improved Wiener process incorporating nonlinear term is developed to build the degradation model of incipient fault based on the fault estimation results. For prognosis, the fast krill herd algorithm is proposed to estimate unknown degradation model coefficients. After that, the probability density function of remaining useful life is derived using the identified degradation model. Finally, the proposed methods are validated by simulations.

A novel method for predicting cell abundance based on single-cell RNA-seq data

BackgroundIt is important to understand the composition of cell type and its proportion in intact tissues, as changes in certain cell types are the underlying cause of disease in humans. Although compositions of cell type and ratios can be obtained by single-cell sequencing, single-cell sequencing is currently expensive and cannot be applied in clinical studies involving a large number of subjects. Therefore, it is useful to apply the bulk RNA-Seq dataset and the single-cell RNA dataset to deconvolute and obtain the cell type composition in the tissue.ResultsBy analyzing the existing cell population prediction methods, we found that most of the existing methods need the cell-type-specific gene expression profile as the input of the signature matrix. However, in real applications, it is not always possible to find an available signature matrix. To solve this problem, we proposed a novel method, named DCap, to predict cell abundance. DCap is a deconvolution method based on non-negative least squares. DCap considers the weight resulting from measurement noise of bulk RNA-seq and calculation error of single-cell RNA-seq data, during the calculation process of non-negative least squares and performs the weighted iterative calculation based on least squares. By weighting the bulk tissue gene expression matrix and single-cell gene expression matrix, DCap minimizes the measurement error of bulk RNA-Seq and also reduces errors resulting from differences in the number of expressed genes in the same type of cells in different samples. Evaluation test shows that DCap performs better in cell type abundance prediction than existing methods.ConclusionDCap solves the deconvolution problem using weighted non-negative least squares to predict cell type abundance in tissues. DCap has better prediction results and does not need to prepare a signature matrix that gives the cell-type-specific gene expression profile in advance. By using DCap, we can better study the changes in cell proportion in diseased tissues and provide more information on the follow-up treatment of diseases.

Deconvolution of the Immune Microenvironment Can Predict the Outcome of Myeloma Patients and Inform Potential Intervention Strategies

IntroductionThe microenvironment of cells around multiple myeloma (MM) tumors strongly influences patient outcomes, as evidenced by the success of immunomodulatory therapies. To develop precision immunotherapeutic approaches, it is essential to identify and enumerate pro- and anti-myelomatous cell types within the bone marrow.MethodsThe population of immune and other non-tumor cell types was estimated from RNA microarray data derived from paired whole bone marrow (WBM) core biopsies and CD138-enriched (CD138+) bone marrow aspirates at presentation from 405 newly diagnosed MM patients who underwent similar treatments. Expression profiles from CD138+ cells were used to infer the residual transcriptome contribution of immune and stromal cells in the WBM array data. We performed unsupervised clustering of the estimated microenvironment gene expression profiles and calculated the statistical significance of cluster splitting using the R packages ConsensusClusterPlus and sigclust, respectively. Individual cell type composition of these microenvironments was estimated using the DCQ algorithm and a gene expression signature matrix based on the published LM22 leukocyte matrix (Newman et al., 2015) that we augmented with phenotypic profiles of 5 bone marrow-specific cell types (myelomatous plasma cells, plasma memory cells, osteoclasts, osteoblasts, and adipocytes).ResultsThis deconvolution approach accurately estimated the percent of tumor cells in WBM core biopsy as estimated by both microscopy and flow cytometry (PCC = 0.57, RMSE = 10.9%). Unsupervised consensus clustering of whole bone marrow microenvironment gene expression data revealed 5 significantly different subtypes. The two largest subtypes, Cluster 1 (N=136) and Cluster 2 (N=128), had the most divergent outcomes with median PFS values of 78 months and 53 months, respectively. Cluster 1 was significantly enriched for ISS-I patients (FDR < 0.02) while Cluster 2 was enriched for ISS-III patients (FDR < 7x10^-5). Cluster 2 showed a significant increase in estimated M1 macrophages, CD8+ T cells, adipocytes, and dendritic cells with a decrease in estimated osteoclasts versus all other clusters combined (FDRs < 0.05). Cluster 5 (N=63) had the second worst PFS at 55 months, however, it had 44% fewer than expected ISS-III patients (FDR = 0.04). Cluster 5 had fewer estimated T cells and adipocytes, and it was the only cluster with a significant (3-fold) increase in estimated NK cells compared to the other clusters combined (FDRs < 0.05). No clusters were significantly enriched for older individuals, hyperdiploidy or common chromosomal translocations (t(4;14), t(11;14), t(14;16)). No clusters were significantly enriched for CD138+ gene expression-defined high risk patients (UAMS GEP-70) or CD138+ gene expression-based molecular subtypes (Zhan et al., 2006).ConclusionsWe demonstrated the technical feasibility to accurately delineate CD138+ expression profiles from a bulk transcriptome admixture and to use the residual microenvironment transcriptome to cluster patients into distinct microenvironment subtypes. Within these subtypes, the clinical outcomes are not explained by CD138+ molecular features nor ISS stage. For example, Cluster 5 has patients with poor prognosis compared to the rest of the cohort, yet it is not significantly enriched for high risk translocations, ISS-III patients, nor GEP-70 high risk patients. Cluster 5 patients show significant decrease in both CD8+ T cell infiltrate and activated dendritic cells, perhaps reflective of a high risk immune state that would not be responsive to checkpoint inhibition. Moreover, Cluster 5 patients are the only ones to have an increase in NK cells, suggesting one could try to stimulate these NK cells to reinvigorate their anti-tumor immunity. Cluster 2, with the poorest prognosis patients and highest tumor burden (by ISS stage), shows increased CD8+ T cells and dendritic cells (where all other clusters are neutral or decreased in these cell types). Cluster 2 patients may thus benefit from checkpoint blockade, and we plan to further explore relevant checkpoint-related pathways in this cohort. Adding this layer of immune-microenvironment to the analysis of myeloma patient samples enables us to formulate biological hypotheses independent of CD138+ biology and may eventually guide more tailored therapeutic interventions and improve outcomes for patients. [Display omitted] DisclosuresDanziger:Celgene Corporation: Employment. Dervan:Twinstrand Biosciences: Equity Ownership; Celgene Corporation: Employment, Equity Ownership. Schmitz:Celgene Corporation: Employment. Copeland:Celgene Corporation: Employment. Fitch:Celgene Corporation: Employment. Newhall:Celgene Corporation: Employment, Equity Ownership. Barlogie:Celgene Corporation: Consultancy, Research Funding; Millenium Pharmaceuticals: Consultancy, Research Funding. Foy:Celgene Corporation: Employment, Equity Ownership. Trotter:Celgene Corporation: Equity Ownership; Celgene Institute for Translational Research Europe: Employment. Hershberg:Adaptive Biotechnologies: Membership on an entity's Board of Directors or advisory committees; Fraizer Healthcare Partners: Consultancy; NanoString Technologies: Membership on an entity's Board of Directors or advisory committees; Celgene Corporation: Employment, Equity Ownership. Ratushny:Celgene Corporation: Employment. Morgan:Takeda: Consultancy, Honoraria; Bristol Myers: Consultancy, Honoraria; Celgene: Consultancy, Honoraria, Research Funding.

Elucidating the immune infiltration in acne and its comparison with rosacea by integrated bioinformatics analysis.

Acne vulgaris and rosacea are common inflammatory complications of the skin, both characterized by abnormal infiltration of immune cells. The two diseases can be differentiated based on characteristic profile of the immune cell infiltrates at the periphery of disease lesions. In addition, dysregulated infiltration of immune cells not only occur in the acne lesions but also in non-lesional areas of patients with the disease, thus characterizing the immune infiltration in these sites can further enhance our understanding on the pathogenesis of acne. Five microarray data-sets (GSE108110, GSE53795, GSE65914, GSE14905 and GSE78097) were downloaded from Gene Expression Omnibus. After removing the batch effects and normalizing the data, we applied the CIBERSORT algorithm combined with signature matrix LM22, to describe 22 types of immune cells' infiltration in acne less than 48 hour (H) old, in comparation with non-lesional skin of acne patients, healthy skin and rosacea (including erythematotelangiectatic rosacea, papulopustular rosacea and phymatous rosacea) and we compared gene expression of Th1 and Th17-related molecules in acne, rosacea and healthy control. Compared with the non-lesional skin of acne patients, healthy individuals and rosacea patients, there is a significant increase in infiltration of neutrophils, monocytes and activated mast cells around the acne lesions, less than 48 H after their development. Contrarily, few naive CD4+ T cells, plasma cells, memory B cells and resting mast cells infiltrate acne sites compared to the aforementioned groups of individuals. Moreover, the infiltration of Regulatory T cells (Tregs) in acne lesions is substantially lower, relative to non-lesional sites of acne patients and skin of healthy individuals. In addition, non-lesional sites of acne patients exhibit lower infiltration of activated memory CD4+ T cells, plasma cells, memory B cells, M0 macrophages, neutrophils, resting mast cells but higher infiltration of Tregs and resting dendritic cells relative to skin of healthy individuals. Intriguingly, we found that among the 3 rosacea subtypes, the immune infiltration profile of papulopustular rosacea is the closest to that of acne lesions. In addition, through gene expression analysis of acne, rosacea and skin tissues of healthy individuals, we found a higher infiltration of Th1 and Th17 cells in acne lesions, relative to non-lesional skin areas of acne patients. Our study provides new insights into the inflammatory pathogenesis of acne, and the difference between acne and rosacea, which helps in differentiating the two diseases. Our findings also guide on appropriate target therapy of the immune cell infiltrates in the two disease conditions.

Hybrid condition monitoring of nonlinear mechatronic system using biogeography-based optimization particle filter and optimized extreme learning machine

This paper proposes a hybrid condition monitoring approach, which integrates bond graph model-based diagnostic technique and data-driven remaining useful life (RUL) prediction, for a nonlinear mechatronic system. In this approach, various degrading faults can be considered and the physical degradation model is not required for RUL prediction. Firstly, an integrated fault signature matrix is proposed by the causal path of bicausal-bond graph model to improve fault isolation performance. After that, a biogeography-based optimization (BBO)-particle filter is developed for fault identification. For prognosis, an optimized extreme learning machine (OELM) is proposed where the hidden layer biases and input weights are optimized by BBO. The fault identification results provide data set to train the OELM for prognosis. Finally, the effectiveness of the approach is verified by simulation and experiment results.

Dimensionality reduction using singular vectors

A common problem in machine learning and pattern recognition is the process of identifying the most relevant features, specifically in dealing with high-dimensional datasets in bioinformatics. In this paper, we propose a new feature selection method, called Singular-Vectors Feature Selection (SVFS). Let D= [A mid mathbf {b}] be a labeled dataset, where mathbf {b} is the class label and features (attributes) are columns of matrix A. We show that the signature matrix S_A=I-A^{dagger }A can be used to partition the columns of A into clusters so that columns in a cluster correlate only with the columns in the same cluster. In the first step, SVFS uses the signature matrix S_D of D to find the cluster that contains mathbf {b}. We reduce the size of A by discarding features in the other clusters as irrelevant features. In the next step, SVFS uses the signature matrix S_A of reduced A to partition the remaining features into clusters and choose the most important features from each cluster. Even though SVFS works perfectly on synthetic datasets, comprehensive experiments on real world benchmark and genomic datasets shows that SVFS exhibits overall superior performance compared to the state-of-the-art feature selection methods in terms of accuracy, running time, and memory usage. A Python implementation of SVFS along with the datasets used in this paper are available at https://github.com/Majid1292/SVFS.

Coupling data-driven and model-based methods to improve fault diagnosis

Monitoring a system is often not an easy task and the best approach to address it would be to develop a monitoring system that uses data, expert knowledge, and mathematical models. Combining these three sources of information on the system is often unpractical of various reasons such as in complex systems. In this paper, a hybrid method for diagnosing single and multiple simultaneous faults, while considering unknown operating conditions, is proposed. This method consists of a Bayesian classifier combining statistical decisions and fault signature matrix. Several scenarios of operating conditions are simulated for illustrative purposes. The results, in terms of classification rates, show the interest of the hybrid classifier. It demostrates higher capabilities to isolating multiple simultaneous faults that the purely data-driven classifier fails to accomplish.

Development of a Methodology Using Artificial Neural Network in the Detection and Diagnosis of Faults for Pneumatic Control Valves.

To satisfy the market, competition in the industrial sector aims for productivity and safety in industrial plant control systems. The appearance of a fault can compromise the system’s proper functioning process. Therefore, Fault Detection and Diagnosis (FDD) methods contribute to avoiding any undesired events, as there are techniques and methods that study the detection, isolation, identification and, consequently, fault diagnosis. In this work, a new methodology that uses faults emulation to obtain parameters similar to the Development and Application of Methods for Diagnosis of Actuators in Industrial Control Systems (DAMADICS) benchmark model will be developed. This methodology uses previous information from tests on sensors with and without faults to detect and classify the situation of the plant and, in the presence of faults, perform the diagnosis through a process of elimination in a hierarchical manner. In this way, the definition of residue signature is used as well as the creation of a decision tree. The whole process is carried out incorporating FDD techniques, through the Non-Linear Auto-Regressive Neural Network Model With Exogenous Inputs (NARX), in the diagnosis of the behavioral prediction of the signals to generate the residual values. Then, it is applied to the construction of the decision tree based on the most significant residue of a certain signal, enabling the process of acquisition and formation of the signature matrix. With the procedures in this article, it is possible to demonstrate a practical and systematic method of how to emulate faults for control valves and the possibility of carrying out an analysis of the data to acquire signatures of the fault behavior. Finally, simulations resulting from the most sensitized variables for the production of residuals that is generated by neural networks are presented, which are used to obtain signatures and isolate the flaws. The process proves to be efficient in computational time and makes it easy to present a fault diagnosis strategy that can be reproduced in other processes.

The signature matrix for 6-Pfaffian graphs

Given a graph G, an orientation D of G and a perfect matching M of G, it is possible to define the sign (-1 or +1) of M in D. Given k orientations of a graph G, a signature matrix A has rows corresponding to perfect matchings of G and columns corresponding to each of these k orientations such that each entry aij is the sign of the i-th perfect matching on the j-th orientation. A graph is k-Pfaffian if there is a set of k orientations whose signature matrix A is such that the linear system Ax = 1 has a solution. The Pfaffian number of a graph G is the smallest k such that G is k-Pfaffian. We present in this paper a characterization of the signature matrices of graphs with Pfaffian number 6.

Is My System of ODEs k-Cooperative?

A linear dynamical system is called positive if its flow maps the non-negative orthant to itself. More precisely, it maps the set of vectors with zero sign variations to itself. A linear dynamical system is called $k$-positive if its flow maps the set of vectors with up to $k-1$ sign variations to itself. A nonlinear dynamical system is called $k$-cooperative if its variational system, which is a time-varying linear dynamical system, is $k$-positive. These systems have special asymptotic properties. For example, it was recently shown that strongly $2$-cooperative systems satisfy a strong Poincar\'{e}-Bendixson property. Positivity and~$k$-positivity are easy to verify in terms of the sign-pattern of the matrix in the dynamics. However, these sign conditions are not invariant under a coordinate transformation. A natural question is to determine if a given~$n$-dimensional system is $k$-positive up to a coordinate transformation. We study this problem for two special kinds of transformations: permutations and scaling by a signature matrix. For any $n\geq 4$ and~$k\in\{2,\dots, n-2\}$, we provide a graph-theoretical necessary and sufficient condition for $k$-positivity up to such coordinate transformations. We describe an application of our results to a specific class of Lotka-Volterra systems.

Hypergraph Model for Wireless Sensor Networks Supervision Design

Security is an important condition to ensure the proper functioning of wireless sensor networks (WSNs). The prevention-based security approaches as cryptographic mechanisms must be enriched by a reactive schemes such as intrusion detection systems (IDSs). But, for safety and reliability reasons, moreover to IDSs, WSN supervision mechanisms are also required. Two activities are concerned: fault detection and isolation (FDI) and fault-tolerant control (FTC). All FDI systems consist of comparing the real behaviour of system with the nominal behaviour given by a model: the difference leads to a fault indicator named residual. Graphical modelling is an important step in FDI realisation. The innovative interest of this paper is the use of hypergraph formalism to model the interconnection between different Components WSN network (CNs). In our proposed approach, the hypergraph model is built from WSNs missions, such that each hyperedge is associated with a particular mission. Based on the structural properties of this model, an online WSN supervision strategy can be defined. Moreover, a fault signature matrix (FSM), as a logic decision, is used to provide the list of faults, which can be detected and isolated.

Model-Based Diagnosis of Telecommunication Cooling Systems Malfunctioning

A model is developed that allows simulating the most-likely failures possibly occurring in freecooling (FC) systems of telecommunication (TLC) switching rooms. Main aim is to provide an effective and online implementable diagnosis method, which in turn will allow fulfilling the threefold function of safeguarding electronic equipment, ensuring desired air quality in case of human presence and reducing malfunction-related waste of energy. Specifically in this work, obstruction (reduction of the volumetric flow of air introduced into the room) and loss of efficiency (degradation of the fan) are deepened. Two black-box sub-models were developed to simulate the above described faulty functioning of the free-coolers. Subsequently, the fault signature matrix was developed, through which the “symptoms”, calculated as residuals between the “faulty” and “non faulty” conditions of the monitored variables, are associated to the corresponding faults. The peculiarity of the telecommunication sector, where nowadays data acquisition and monitoring platforms are significantly spreading to monitor most significant energy consumptions, including cooling loads, was proved essential in guaranteeing effective isolation of different faults. The simulation results highlight the reliability of the developed diagnostic tool, expected to be versatile and easy to implement enough for being extended to air-handling unit diagnosis, as well as other industrial sectors.

GR decompositions and their relations to Cholesky‐like factorizations

AbstractFor a given matrix, we are interested in computing GR decompositions A = GR, where G is an isometry with respect to given scalar products. The orthogonal QR decomposition is the representative for the Euclidian scalar product. For a signature matrix, a respective factorization is given as the hyperbolic QR decomposition. Considering a skew‐symmetric matrix leads to the symplectic QR decomposition. The standard approach for computing GR decompositions is based on the successive elimination of subdiagonal matrix entries. For the hyperbolic and symplectic case, this approach does in general not lead to a satisfying numerical accuracy. An alternative approach computes the QR decomposition via a Cholesky factorization, but also has bad stability. It is improved by repeating the procedure a second time. In the same way, the hyperbolic and the symplectic QR decomposition are related to the LDLT and a skew‐symmetric Cholesky‐like factorization. We show that methods exploiting this connection can provide better numerical stability than elimination‐based approaches.

Compound Fault Diagnosis and Sequential Prognosis for Electric Scooter with Uncertainties

This paper addresses diagnosis and prognosis problems for an electric scooter subjected to parameter uncertainties and compound faults (i.e., permanent fault and intermittent fault with non-monotonic degradation). First, the diagnostic bond graph in linear fractional transformation form is used to model the uncertain electric scooter and derive the analytical redundancy relations incorporating the nominal part and uncertain part, based on which the adaptive thresholds for robust fault detection and the fault signature matrix for fault isolation can be obtained. Second, an adaptive enhanced unscented Kalman filter is proposed to identify the fault magnitudes and distinguish the fault types where an auxiliary detector is introduced to capture the appearing and disappearing moments of intermittent fault. Third, a dynamic model with usage dependent degradation coefficient is developed to describe the degradation process of intermittent fault under various usage conditions. Due to the variation of degradation coefficient and the presence of non-monotonic degradation characteristic under some usage conditions, a sequential prognosis method is proposed where the reactivation of the prognoser is governed by the reactivation events. Finally, the proposed methods are validated by experiment results.

Robust unsupervised anomaly detection via multi-time scale DCGANs with forgetting mechanism for industrial multivariate time series

Detecting anomalies in time series is a vital technique in a wide variety of industrial application in which sensors monitor expensive machinery. The complexity of this task increases when heterogeneous sensors provide information of different attributes, scales and characteristics from the same machine. Actually, the challenges of anomaly detection for industrial time series are to design effective pre-processing, feature extraction and overcome the lack of abnormal samples. Recent deep learning models have shown prominent abilities on raw multivariate time series, alleviating these previous works. In this work, a novel framework named multi-time scale deep convolutional generative adversarial network (MTS-DCGAN) is proposed to deal with anomaly detection of industrial time series. Firstly, multivariate time series are transformed into the multi-channel signature matrices via sliding window based cross-correlation computation, and therein forgetting mechanism is introduced to effectively avoid false alarms due to excessive influence of old sequences. Then the framework conducts an unsupervised adversarial training of multi-channel signature matrices and capture their hidden features via deep convolution structure. Besides, a novel threshold setting strategy is proposed to optimize anomaly detection performance under imbalance of normal and abnormal data. Finally, the proposed framework is assessed against the experiments on four datasets. Within this study, experimental results show our framework outperforms the comparison algorithms in terms of model performance and robustness, providing an effective anomaly detection method for industrial multivariate time series.

A review of digital cytometry methods: estimating the relative abundance of cell types in a bulk of cells.

Due to the high cost of flow and mass cytometry, there has been a recent surge in the development of computational methods for estimating the relative distributions of cell types from the gene expression profile of a bulk of cells. Here, we review the five common 'digital cytometry' methods: deconvolution of RNA-Seq, cell-type identification by estimating relative subsets of RNA transcripts (CIBERSORT), CIBERSORTx, single sample gene set enrichment analysis and single-sample scoring of molecular phenotypes deconvolution method. The results show that CIBERSORTx B-mode, which uses batch correction to adjust the gene expression profile of the bulk of cells ('mixture data') to eliminate possible cross-platform variations between the mixture data and the gene expression data of single cells ('signature matrix'), outperforms other methods, especially when signature matrix and mixture data come from different platforms. However, in our tests, CIBERSORTx S-mode, which uses batch correction for adjusting the signature matrix instead of mixture data, did not perform better than the original CIBERSORT method, which does not use any batch correction method. This result suggests the need for further investigations into how to utilize batch correction in deconvolution methods.

Multi-label fault diagnosis of rolling bearing based on meta-learning

In practical applications, it is difficult to acquire sufficient fault samples for training deep learning fault diagnosis model of rolling bearing. Aiming at the few-shot issue and multi-label attributes of single-point faults, a novel fault diagnosis method of rolling bearing based on time–frequency signature matrix (T–FSM) feature and multi-label convolutional neural network with meta-learning (MLCML) is proposed in this paper. At the beginning, the T–FSM features sensitive to few-shot fault diagnosis of measured vibration signal are extracted. Subsequently, a designed multi-label convolutional neural network (MLCNN) with a specific architecture is employed to identify faults. Crucially, the meta-learning strategy of learning initial network parameters susceptive to task changes is incorporated to MLCNN for addressing the few-shot problem. Ultimately, the publicly available rolling bearing dataset is utilized to demonstrate the effectiveness of the proposed method. The experimental results exhibit that the trained MLCML has the capability of learning to learn few-shot fault attributes with outstanding diagnosis accuracy and generalization. More concretely, the model can adapt to new fault categories rapidly owing to that only a few samples and update steps are required to fine-tune the network.

Development and application of a comprehensive model-based methodology for fault mitigation of fuel cell powered systems

The present paper describes an innovative and generalizable approach to apply fault mitigation strategies to fuel cell powered systems, upon information on system State of Health and Remaining Useful Life. Model-based approach is proposed to derive useful performance-related indicators for each system component. The model comprises two main parts: a nominal part, providing the key variables behavior in nominal conditions, and a faulty part that can be used for fault identification purposes. The framework of the algorithm firstly addresses a monitoring phase, through which residuals are computed, and if one or more residuals overcome defined thresholds, a fault detection is triggered. Afterwards, fault isolation is performed by means of a Fault Signature Matrix and the fault identification (i.e., its magnitude and time-behavior definition) is performed thanks to the faulty sub-models. Once characterized the fault, several strategies (according to different fault magnitudes) are considered, and the most suitable one can be chosen and applied. A case study is then presented to validate the methodology on a fuel starvation fault caused in a 6-cells solid oxide fuel cell stack by a fuel leakage in the anode pipeline. Once applied the mitigation strategy, it has been verified that the power output of the system safely bounds within 20% of its nominal value, whereas stack efficiency variation is negligible. The methodology herein proposed could substantially help the commercial success of solid oxide fuel cell technology, allowing increasing lifetime, with a much focused control of the main variable for diagnostic and maintenance-oriented applications. Indeed, if used in real applications, the proposed approach will speed up maintenance actions even setting the system in a soft condition to be properly prepared for replacement as well.

Abstract 3779: Multiomic single cell analysis of normal human bone marrow identifies a unique stem and progenitor population that expands in AML

Abstract Hematopoietic stem and progenitor cells (HSPCs) can generate a diversity of blood cells throughout the human lifespan. Although these cells have been evaluated using both sorted (Corces et al. Nat. Genet. 2016) and single cell assays (Pellin et al Nat. Commun. 2019), there remains uncertainty in the degree of heterogeneity within HSPC subpopulations and their associated differentiation trajectories. The phenotypic diversity within HSPCs needs to be better characterized in order to understand the pathogenesis of blood disorders including hematologic malignancies. To address this need, we characterized healthy bone marrow mononuclear cells (BMMCs) with whole transcriptome analysis (WTA) and surface marker evaluation. We hypothesized that by utilizing concurrent RNA and surface marker analysis (n=35), we can improve HSPC clustering and characterize specific phenotypic states along unique hematopoietic differentiation trajectories. Three healthy BMMC samples were stained with antibody conjugated oligonucleotides (BD® Abseq) and analyzed using BD Rhapsody™. We filtered 8,070 high quality cells for 2,508 HSPCs, myeloid cells, and lymphocyte precursors. The antibody-derived tags (ADTs) obtained from BD® Abseq comprised 33 of the most informative features (n=2000) and resulted in more stable clustering as determined by within sum of squares (WSS = 898 verses 934 for WTA alone for 20 clusters). Additionally, we designed a targeted HSPC panel (n=500 genes) with BD® Abseq which identified similar cell clusters compared to the WTA alone and WTA plus ADT data (rand index = 0.88). HSPC clustering identified putative hematopoietic stem cell (HSC), common myeloid progenitor (CMP), and megakaryocyte-erythroid progenitor (MEP) clusters that expressed canonical surface markers. Of interest, we identified candidate novel myeloid, T-cell, and B-cell primed precursors and new surface marker expression gradients that align with specific differentiation trajectories. The results of this analysis will be presented. The HSPC clusters were converted into a signature matrix using Cibersortx (Newman et al. Nat. Biotechnol. 2019), and bulk acute myeloid leukemia (AML) and healthy samples were deconvolved into respective healthy cell clusters. We subsequently performed multivariate Cox proportional hazard analysis, and observed that high levels of healthy cluster 8 (H8; HR 3.40, 95% confidence interval 1.24-9.34), the candidate lymphoid-primed multipotent progenitor (LMPP), and low levels of healthy cluster 6 (H6; HR 0.26, 95% confidence interval 0.12-0.58), a candidate erythrocyte precursor, at diagnosis were associated with worse overall survival. Deconvolution using sorted healthy sub-populations (Corces et al. Nat. Genet. 2016) only identified erythrocyte precursors as statistically relevant (HR 0.27, 95% confidence interval 0.12-0.61). Of note, H8 had a distinct gene expression profile compared to that identified for the sorted LMPP sub-population using differential gene expression analysis. In summary, we identified novel cell type clusters and surface marker associations using combined single cell WTA and surface marker analysis (BD® Abseq). We were able to correlate cell types with both canonical and novel surface markers, and deconvolution analysis provided preliminary insights into their clinical relevance in AML. Citation Format: Asiri Ediriwickrema, Sreejith Ramakrishnan, Margaret Nakamoto, Smita Ghanekar, Bogdan Luca, Aaron Newman, Andrew Gentles, Ravindra Majeti. Multiomic single cell analysis of normal human bone marrow identifies a unique stem and progenitor population that expands in AML [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 3779.