Search results for: labels

Authors: George Zhou, Yunchan Chen, Candace Chien

Abstract:

Kidney replacement therapy is the current standard of care for end-stage renal diseases. In-center or home hemodialysis remains an integral component of the therapeutic regimen. Arteriovenous fistulas (AVF) make up the vascular circuit through which blood is filtered and returned. Naturally, AVF patency determines whether adequate clearance and filtration can be achieved and directly influences clinical outcomes. Our aim was to build a deep learning model for automated AVF stenosis screening based on the sound of blood flow through the AVF. A total of 311 patients with AVF were enrolled in this study. Blood flow sounds were collected using a digital stethoscope. For each patient, blood flow sounds were collected at 6 different locations along the patient’s AVF. The 6 locations are artery, anastomosis, distal vein, middle vein, proximal vein, and venous arch. A total of 1866 sounds were collected. The blood flow sounds are labeled as “patent” (normal) or “stenotic” (abnormal). The labels are validated from concurrent ultrasound. Our dataset included 1527 “patent” and 339 “stenotic” sounds. We show that blood flow sounds vary significantly along the AVF. For example, the blood flow sound is loudest at the anastomosis site and softest at the cephalic arch. Contextualizing the sound with location metadata significantly improves classification performance. How to encode and incorporate categorical metadata is an active area of research1. Herein, we study ordinal (i.e., integer) encoding schemes. The numerical representation is concatenated to the flattened feature vector. We train a vision transformer (ViT) on spectrogram image representations of the sound and demonstrate that using scalar multiples of our integer encodings improves classification performance. Models are evaluated using a 10-fold cross-validation procedure. The baseline performance of our ViT without any location metadata achieves an AuROC and AuPRC of 0.68 ± 0.05 and 0.28 ± 0.09, respectively. Using the following encodings of Artery:0; Arch: 1; Proximal: 2; Middle: 3; Distal 4: Anastomosis: 5, the ViT achieves an AuROC and AuPRC of 0.69 ± 0.06 and 0.30 ± 0.10, respectively. Using the following encodings of Artery:0; Arch: 10; Proximal: 20; Middle: 30; Distal 40: Anastomosis: 50, the ViT achieves an AuROC and AuPRC of 0.74 ± 0.06 and 0.38 ± 0.10, respectively. Using the following encodings of Artery:0; Arch: 100; Proximal: 200; Middle: 300; Distal 400: Anastomosis: 500, the ViT achieves an AuROC and AuPRC of 0.78 ± 0.06 and 0.43 ± 0.11. respectively. Interestingly, we see that using increasing scalar multiples of our integer encoding scheme (i.e., encoding “venous arch” as 1,10,100) results in progressively improved performance. In theory, the integer values do not matter since we are optimizing the same loss function; the model can learn to increase or decrease the weights associated with location encodings and converge on the same solution. However, in the setting of limited data and computation resources, increasing the importance at initialization either leads to faster convergence or helps the model escape a local minimum.

Keywords: arteriovenous fistula, blood flow sounds, metadata encoding, deep learning

Procedia PDF Downloads 52

Authors: Manish Saha, Manish Kumar Niranjan, Saket Asthana

Abstract:

The K₀.₅Na₀.₅NbO₃ (KNN) system has emerged as one of the most promising lead-free piezoelectric over the years. In this work, we perform a comprehensive investigation of electronic structure, lattice dynamics and dielectric/ferroelectric properties of the room temperature phase of KNN by combining ab-initio DFT-based theoretical analysis and experimental characterization. We assign the symmetry labels to KNN vibrational modes and obtain ab-initio polarized Raman spectra, Infrared (IR) reflectivity, Born-effective charge tensors, oscillator strengths etc. The computed Raman spectrum is found to agree well with the experimental spectrum. In particular, the results suggest that the mode in the range ~840-870 cm-¹ reported in the experimental studies is longitudinal optical (LO) with A_1 symmetry. The Raman mode intensities are calculated for different light polarization set-ups, which suggests the observation of different symmetry modes in different polarization set-ups. The electronic structure of KNN is investigated, and an optical absorption spectrum is obtained. Further, the performances of DFT semi-local, metal-GGA and hybrid exchange-correlations (XC) functionals, in the estimation of KNN band gaps are investigated. The KNN bandgap computed using GGA-1/2 and HSE06 hybrid functional schemes are found to be in excellant agreement with the experimental value. The COHP, electron localization function and Bader charge analysis is also performed to deduce the nature of chemical bonding in the KNN. The solid-state reaction and hydrothermal methods are used to prepare the KNN ceramics, and the effects of grain size on the physical characteristics these ceramics are examined. A comprehensive study on the impact of different synthesis techniques on the structural, electrical, and photocatalytic properties of ferroelectric ceramics KNN. The KNN-S prepared by solid-state method have significantly larger grain size as compared to that for KNN-H prepared by hydrothermal method. Furthermore, the KNN-S is found to exhibit higher dielectric, piezoelectric and ferroelectric properties as compared to KNN-H. On the other hand, the increased photocatalytic activity is observed in KNN-H as compared to KNN-S. As compared to the hydrothermal synthesis, the solid-state synthesis causes an increase in the relative dielectric permittivity (ε^') from 2394 to 3286, remnant polarization (P_r) from 15.38 to 20.41 μC/cm^², planer electromechanical coupling factor (k_p) from 0.19 to 0.28 and piezoelectric coefficient (d_33) from 88 to 125 pC/N. The KNN-S ceramics are also found to have a lower leakage current density, and higher grain resistance than KNN-H ceramic. The enhanced photocatalytic activity of KNN-H is attributed to relatively smaller particle sizes. The KNN-S and KNN-H samples are found to have degradation efficiencies of RhB solution of 20% and 65%, respectively. The experimental study highlights the importance of synthesis methods and how these can be exploited to tailor the dielectric, piezoelectric and photocatalytic properties of KNN. Overall, our study provides several bench-mark important results on KNN that have not been reported so far.

Keywords: lead-free piezoelectric, Raman intensity spectrum, electronic structure, first-principles calculations, solid state synthesis, photocatalysis, hydrothermal synthesis

Procedia PDF Downloads 18

Authors: Rodrigo Arnaldo Scarpel

Abstract:

Trend following is a widely accepted investment strategy that adopts a rule-based trading mechanism that rather than striving to predict market direction or on information gathering to decide when to buy and when to sell a stock. Thus, in trend following one must respond to market’s movements that has recently happen and what is currently happening, rather than on what will happen. Optimally, in trend following strategy, is to catch a bull market at its early stage, ride the trend, and liquidate the position at the first evidence of the subsequent bear market. For applying the trend following strategy one needs to find the trend and identify trade signals. In order to avoid false signals, i.e., identify fluctuations of short, mid and long terms and to separate noise from real changes in the trend, most academic works rely on moving averages and other technical analysis indicators, such as the moving average convergence divergence (MACD) and the relative strength index (RSI) to uncover intelligible stock trading rules following trend following strategy philosophy. Recently, some works has applied machine learning techniques for trade rules discovery. In those works, the process of rule construction is based on evolutionary learning which aims to adapt the rules to the current environment and searches for the global optimum rules in the search space. In this work, instead of focusing on the usage of machine learning techniques for creating trading rules, a time series trend classification employing a semi-supervised approach was used to early identify both the beginning and the end of upward and downward trends. Such classification model can be employed to identify trade signals and the decision-making procedure is that if an up-trend (down-trend) is identified, a buy (sell) signal is generated. Semi-supervised learning is used for model training when only part of the data is labeled and Semi-supervised classification aims to train a classifier from both the labeled and unlabeled data, such that it is better than the supervised classifier trained only on the labeled data. For illustrating the proposed approach, it was employed daily trade information, including the open, high, low and closing values and volume from January 1, 2000 to December 31, 2022, of the São Paulo Exchange Composite index (IBOVESPA). Through this time period it was visually identified consistent changes in price, upwards or downwards, for assigning labels and leaving the rest of the days (when there is not a consistent change in price) unlabeled. For training the classification model, a pseudo-label semi-supervised learning strategy was used employing different technical analysis indicators. In this learning strategy, the core is to use unlabeled data to generate a pseudo-label for supervised training. For evaluating the achieved results, it was considered the annualized return and excess return, the Sortino and the Sharpe indicators. Through the evaluated time period, the obtained results were very consistent and can be considered promising for generating the intended trading signals.

Keywords: evolutionary learning, semi-supervised classification, time series data, trading signals generation

Procedia PDF Downloads 54

Authors: Z. Gharaee

Abstract:

Recognizing actions performed by others is important in our daily lives since it is necessary for communicating with others in a proper way. We perceive an action by observing the kinematics of motions involved in the performance. We use our experience and concepts to make a correct recognition of the actions. Although building the action concepts is a life-long process, which is repeated throughout life, we are very efficient in applying our learned concepts in analyzing motions and recognizing actions. Experiments on the subjects observing the actions performed by an actor show that an action is recognized after only about two hundred milliseconds of observation. In this study, hierarchical action recognition architecture is proposed by using growing grid layers. The first-layer growing grid receives the pre-processed data of consecutive 3D postures of joint positions and applies some heuristics during the growth phase to allocate areas of the map by inserting new neurons. As a result of training the first-layer growing grid, action pattern vectors are generated by connecting the elicited activations of the learned map. The ordered vector representation layer receives action pattern vectors to create time-invariant vectors of key elicited activations. Time-invariant vectors are sent to second-layer growing grid for categorization. This grid creates the clusters representing the actions. Finally, one-layer neural network developed by a delta rule labels the action categories in the last layer. System performance has been evaluated in an experiment with the publicly available MSR-Action3D dataset. There are actions performed by using different parts of human body: Hand Clap, Two Hands Wave, Side Boxing, Bend, Forward Kick, Side Kick, Jogging, Tennis Serve, Golf Swing, Pick Up and Throw. The growing grid architecture was trained by applying several random selections of generalization test data fed to the system during on average 100 epochs for each training of the first-layer growing grid and around 75 epochs for each training of the second-layer growing grid. The average generalization test accuracy is 92.6%. A comparison analysis between the performance of growing grid architecture and self-organizing map (SOM) architecture in terms of accuracy and learning speed show that the growing grid architecture is superior to the SOM architecture in action recognition task. The SOM architecture completes learning the same dataset of actions in around 150 epochs for each training of the first-layer SOM while it takes 1200 epochs for each training of the second-layer SOM and it achieves the average recognition accuracy of 90% for generalization test data. In summary, using the growing grid network preserves the fundamental features of SOMs, such as topographic organization of neurons, lateral interactions, the abilities of unsupervised learning and representing high dimensional input space in the lower dimensional maps. The architecture also benefits from an automatic size setting mechanism resulting in higher flexibility and robustness. Moreover, by utilizing growing grids the system automatically obtains a prior knowledge of input space during the growth phase and applies this information to expand the map by inserting new neurons wherever there is high representational demand.

Keywords: action recognition, growing grid, hierarchical architecture, neural networks, system performance

Procedia PDF Downloads 133

Authors: Leander Van Neste, Kirk Wojno

Abstract:

The innate immune system reacts to foreign insult in several unique ways, one of which is phagocytosis of perceived threats such as cancer, bacteria, and viruses. The goal of this study was to look for evidence of phagocytosed RNA from tumor cells in circulating monocytes. While all monocytes possess phagocytic capabilities, the non-classical CD14+/FCGR3A+ monocytes and the intermediate CD14++/FCGR3A+ monocytes most actively remove threatening ‘external’ cellular materials. Purified CD14-positive monocyte samples from fourteen patients recently diagnosed with clinically localized prostate cancer (PCa) were investigated by single-cell RNA sequencing using the 10X Genomics protocol followed by paired-end sequencing on Illumina’s NovaSeq. Similarly, samples were processed and used as controls, i.e., one patient underwent biopsy but was found not to harbor prostate cancer (benign), three young, healthy men, and three men previously diagnosed with prostate cancer that recently underwent (curative) radical prostatectomy (post-RP). Sequencing data were mapped using 10X Genomics’ CellRanger software and viable cells were subsequently identified using CellBender, removing technical artifacts such as doublets and non-cellular RNA. Next, data analysis was performed in R, using the Seurat package. Because the main goal was to identify differences between PCa patients and ‘control’ patients, rather than exploring differences between individual subjects, the individual Seurat objects of all 21 patients were merged into one Seurat object per Seurat’s recommendation. Finally, the single-cell dataset was normalized as a whole prior to further analysis. Cell identity was assessed using the SingleR and cell dex packages. The Monaco Immune Data was selected as the reference dataset, consisting of bulk RNA-seq data of sorted human immune cells. The Monaco classification was supplemented with normalized PCa data obtained from The Cancer Genome Atlas (TCGA), which consists of bulk RNA sequencing data from 499 prostate tumor tissues (including 1 metastatic) and 52 (adjacent) normal prostate tissues. SingleR was subsequently run on the combined immune cell and PCa datasets. As expected, the vast majority of cells were labeled as having a monocytic origin (~90%), with the most noticeable difference being the larger number of intermediate monocytes in the PCa patients (13.6% versus 7.1%; p<.001). In men harboring PCa, 0.60% of all purified monocytes were classified as harboring PCa signals when the TCGA data were included. This was 3-fold, 7.5-fold, and 4-fold higher compared to post-RP, benign, and young men, respectively (all p<.001). In addition, with 7.91%, the number of unclassified cells, i.e., cells with pruned labels due to high uncertainty of the assigned label, was also highest in men with PCa, compared to 3.51%, 2.67%, and 5.51% of cells in post-RP, benign, and young men, respectively (all p<.001). It can be postulated that actively phagocytosing cells are hardest to classify due to their dual immune cell and foreign cell nature. Hence, the higher number of unclassified cells and intermediate monocytes in PCa patients might reflect higher phagocytic activity due to tumor burden. This also illustrates that small numbers (~1%) of circulating peripheral blood monocytes that have interacted with tumor cells might still possess detectable phagocytosed tumor RNA.

Keywords: circulating monocytes, phagocytic cells, prostate cancer, tumor immune response

Procedia PDF Downloads 133

Authors: Jelena Maletkovic

Abstract:

Background: Endogenous Cushing’s syndrome is a rare disease, but iatrogenic or drug related Cushing syndrome from glucocorticoid products is commonly seen in clinical practice. With high dose and long term use of glucocorticoids, patients can develop isolated hypothalamic-pituitary-adrenal (HPA) suppression, or HPA axis suppression can be accompanied by overt iatrogenic Cushing’s syndrome. This is a rare case where severe Cushing’s syndrome developed from an unknown medication and was followed by severe and prolonged adrenal insufficiency and multiple potentially fatal complications. Case: This is a 37-year-old woman who is presented to Emergency Room (ER) with shortness of breath and chest pain. Four months prior to this presentation the patient was a generally healthy woman who was looking for improvement in her appearance and visited local Rejuvenation Clinic. After initial consultation with a nurse, she was contacted by a physician over the phone and was advised to start taking multiple injectable medications that will arrive by mail. Medications without any labels on bottles were delivered and the patient started daily intramuscular injections. Over the next two months, she noticed rounding of her face and swelling around her eyes. She gained 20 pounds, mostly abdominal fat and became extremely fatigued. Her muscles on legs were visibly decreasing in size and she felt significant muscle weakness. Unexplained bruising occurred. She started growing hair on face and developed secondary amenorrhea. New severe back pain started. She developed depression and headaches. Finally, over a few days, a number of red-purple stretch marks that were sensitive and painful appeared over her abdomen, upper part of arms and legs. She then became suspicious that these dramatic symptoms are caused by injectable medication and she discontinued injections. Over the next few days she presented to ER with low blood pressure and oxygen saturation of 75%. Studies revealed extensive pneumonia as well as multiple pulmonary emboli. Her white blood count was elevated with 32 000 and she also had acute kidney failure on admission. She was treated for sepsis and was also given stress dose steroids. Steroids were tapered over 48 hours and discontinued. After being discharged to home, on her first visit to endocrinology clinic she had undetectable ACTH of < 2pg/mL and undetectable 8am cortisol of < 0.2mcg/dL. She did not respond to an intramuscular injection of cosyntropin 250mcg and her repeated cortisol after 60 minutes was only 1mcg/dL. The patient was diagnosed with adrenal insufficiency and was started on hydrocortisone 20mg+10mg. It took close to 2 years of slow tapering for recovery of this patient’s HPA axis and resolve all the sequelae from Cushing’s syndrome. Conclusion: Misuse and abuse of glucocorticoids have been present almost since these medications were discovered. This is a rare case where not only severe Cushing’s syndrome in full clinical picture developed but also the patient suffered multiple potentially fatal complications and prolonged adrenal insufficiency. Visits to herbal, rejuvenation, esthetic, and similar clinics are becoming more and more popular and physicians need to be aware of possible non-benign nature of medications that their patients may be using.

Keywords: iatrogenic, Cushing's syndrome, adrenal crisis, steroid abuse

Procedia PDF Downloads 141

Authors: Francesca Crivellari, Nicholas Mavrogiannis, Zachary Gagnon

Abstract:

Portable diagnostic methods have become increasingly important for a number of different purposes: point-of-care screening in developing nations, environmental contamination studies, bio/chemical warfare agent detection, and end-user use for commercial health monitoring. The cheapest and most portable methods currently available are paper-based – lateral flow and dipstick methods are widely available in drug stores for use in pregnancy detection and blood glucose monitoring. These tests are successful because they are cheap to produce, easy to use, and require minimally invasive sampling. While adequate for their intended uses, in the realm of blood-borne pathogens and numerous cancers, these paper-based methods become unreliable, as they lack the nM/pM sensitivity currently achieved by clinical diagnostic methods. Clinical diagnostics, however, utilize techniques involving surface plasmon resonance (SPR) and enzyme-linked immunosorbent assays (ELISAs), which are expensive and unfeasible in terms of portability. To develop a better, competitive biosensor, we must reduce the cost of one, or increase the sensitivity of the other. Electric fields are commonly utilized in microfluidic devices to manipulate particles, biomolecules, and cells. Applications in this area, however, are primarily limited to interfaces formed between immiscible interfaces. Miscible, liquid-liquid interfaces are common in microfluidic devices, and are easily reproduced with simple geometries. Here, we demonstrate the use of electrical fields at liquid-liquid electrical interfaces, known as fluidic dielectrophoresis, (fDEP) for biodetection in a microfluidic device. In this work, we apply an AC electric field across concurrent laminar streams with differing conductivities and permittivities to polarize the interface and induce a discernible, near-immediate, frequency-dependent interfacial tilt. We design this aqueous electrical interface, which becomes the biosensing “substrate,” to be intelligent – it “moves” only when a target of interest is present. This motion requires neither labels nor expensive electrical equipment, so the biosensor is inexpensive and portable, yet still capable of sensitive detection. Nanoparticles, due to their high surface-area-to-volume ratio, are often incorporated to enhance detection capabilities of schemes like SPR and fluorimetric assays. Most studies currently investigate binding at an immobilized solid-liquid or solid-gas interface, where particles are adsorbed onto a planar surface, functionalized with a receptor to create a reactive substrate, and subsequently flushed with a fluid or gas with the relevant analyte. These typically involve many preparation and rinsing steps, and are susceptible to surface fouling. Our microfluidic device is continuously flowing and renewing the “substrate,” and is thus not subject to fouling. In this work, we demonstrate the ability to electrokinetically detect biomolecules binding to functionalized nanoparticles at liquid-liquid interfaces using fDEP. In biotin-streptavidin experiments, we report binding detection limits on the order of 1-10 pM, without amplifying signals or concentrating samples. We also demonstrate the ability to detect this interfacial motion, and thus the presence of binding, using impedance spectroscopy, allowing this scheme to become non-optical, in addition to being label-free.

Keywords: biodetection, dielectrophoresis, microfluidics, nanoparticles

Procedia PDF Downloads 360

Authors: Amberly Brigden, Ali Heawood, Emma C. Anderson, Richard Morris, Esther Crawley

Abstract:

Background: Paediatric chronic fatigue syndrome (CFS)/myalgic encephalomyelitis (ME) is characterised by persistent, disabling fatigue. Health services see patients below the age of 12. This age group experience high levels of disability, with low levels of school attendance, high levels of fatigue, anxiety, functional disability and pain. CFS/ME interventions have been developed for adolescents, but the developmental needs of younger children suggest treatment should be tailored to this age group. Little is known about how intervention should be delivered to this age group, and further work is needed to explore this. Qualitative research aids patient-centered design of health intervention. Methods: Five to 11-year-olds and their parents were recruited from a specialist CFS/ME service. Semi-structured interviews explored the families’ experience of the condition and perspectives on treatment. Interactive and arts-based methods were used. Interviews were audio-recorded, transcribed and analysed thematically. Qualitative Results: 14 parents and 7 children were interviewed. Early analysis of the interviews revealed the importance of the social-ecological setting of the child, which led to themes being developed in the context of Systems Theory. Theme one relates to the level of the child, theme two the family system, theme three the organisational and societal systems, and theme four cuts-across all levels. Theme1: The child’s capacity to describe, understand and manage their condition. Younger children struggled to describe their internal experiences, such as physical symptoms. Parents felt younger children did not understand some concepts of CFS/ME and did not have the capabilities to monitor and self-regulate their behaviour, as required by treatment. A spectrum of abilities was described; older children (10-11-year-olds) were more involved in clinical sessions and had more responsibility for self-management. Theme2: Parents’ responsibility for managing their child’s condition. Parents took responsibility for regulating their child’s behaviour in accordance with the treatment programme. They structured their child’s environment, gave direct instructions to their child, and communicated the needs of their child to others involved in care. Parents wanted their child to experience a 'normal' childhood and took steps to shield their child from medicalization, including diagnostic labels and clinical discussions. Theme3: Parental isolation and the role of organisational and societal systems. Parents felt unsupported in their role of managing the condition and felt negative responses from primary care health services and schools were underpinned by a lack of awareness and knowledge about CFS/ME in younger children. This sometimes led to a protracted time to diagnosis. Parents felt that schools have the potential important role in managing the child’s condition. Theme4: Complexity and uncertainty. Many parents valued specialist treatment (which included activity management, physiotherapy, sleep management, dietary advice, medical management and psychological support), but felt it needed to account for the complexity of the condition in younger children. Some parents expressed uncertainty about the diagnosis and the treatment programme. Conclusions: Interventions for younger children need to consider the 'systems' (family, organisational and societal) involved in the child’s care. Future research will include interviews with clinicians and schools supporting younger children with CFS/ME.

Keywords: chronic fatigue syndrome (CFS)/myalgic encephalomyelitis (ME), pediatric, qualitative, treatment

Procedia PDF Downloads 109