Software For Classification Research Articles

DEVELOPMENT AND RESEARCH OF SOFTWARE SOLUTION FOR BUSINESS PROCESS MODEL CORRECTNESS ANALYSIS USING MACHINE LEARNING

Poorly designed business process models are a source of errors and the subsequent costs associated with these errors, such as monetary costs, lost time, or even some harmful impact on people or the environment if the erroneous business process models are associated with critical industries. The BPM (Business Process Management) lifecycle usually consists of designing, implementing, monitoring, and controlling the business process execution, but it lacks continuous quality control of the created BPMN (Business Process Model and Notation) models. Thus, this paper considers the problem of business process models classification based on their correctness, which solution will ensure quality control of the designed business process models. Thus, this study aims to improve the quality of business process models by developing a software solution for business process models classification based on their correctness. The subject of the study is the process of business process models classification based on their correctness, which uses quality measures and thresholds, usually, complexity measures. The subject of the study is a software solution for business process models classification based on their correctness. Therefore, in this study, the algorithm to solve the problem of BPMN models classification using logistic regression, interface complexity, and modularity measures is proposed, the software requirements are determined, the software development tools are selected, the software for business process models classification based on their correctness is designed, the corresponding software components are developed, the use of a software solution for solving the problem of business process models classification based on their correctness is demonstrated, the obtained results are analyzed and discussed. The developed software indicates high performance of BPMN models classification based on their correctness, achieving high accuracy (99.14 %), precision (99.88 %), recall (99.23 %), and F-score (99.56 %), highlighting the high performance of modeling errors detection.

Multiplexed immunolabelling of cancer using bioconjugated plasmonic gold-silver alloy nanoparticles.

Reliable protein detection methods are vital for advancing biological research and medical diagnostics. While immunohistochemistry and immunofluorescence are commonly employed, their limitations underscore the necessity for alternative approaches. This study introduces immunoplasmonic labelling, utilizing plasmonic nanoparticles (NPs), specifically designed gold and gold-silver alloy NPs (Au:Ag NPs), for multiplexed and quantitative protein detection. These NPs, when coupled with antibodies targeting proteins of interest, enable accurate counting and evaluation of protein expression levels while overcoming issues such as autofluorescence. In this study, we compare two nanoparticle functionalization strategies-one coating based on thiolated PEG and one coating based on calix[4]arenes-on gold and gold-silver alloy nanoparticles of varying sizes. Overall results tend to demonstrate a greater versatility for the calix[4]arene-based coating. With this coating and using the classical EDC/sulfo-NHS cross-linking procedure, we also demonstrate the successful multiplexed immunolabelling of Her2, CD44, and EpCAM in breast cancer cell lines (SK-BR-3 and MDA-MB-231). Furthermore, we introduce a user-friendly software for automatic NP detection and classification by colour, providing a promising proof-of-concept for the practical application of immunoplasmonic techniques in the quantitative analysis of biopsies in the clinical setting.

Software for classification of banana ripening stage using machine learning

Software for classification of banana ripening stage using machine learning

Представяне на „Софтуер за потенциална относителна оценка и класификация по пригодност за земеползване на земеделските земи в България“

In 2023, I developed “Software for potential land use suitability classification of agricultural land in Bulgaria” (Radovanova, 2023). It works in MS Office terms, programmed entirely through Microsoft Excel. The software with the corresponding methodology is essentially an update of the legally adopted “Methodology for work on the cadastre of agricultural land in Bulgaria” (Petrov E. et al., 1988), and this was achieved by a newly developed “Methodology for potential assessment of agricultural land in Bulgaria” (Radovanova, 2020). Actual and potential assessment are fully comparable in terms of approach, scheme, parameters, etc. The difference between actual and potential assessment, as in the case of software, is that in the case of potential assessment of agricultural land, potential characteristics (subject to improvement by external intervention) are used. These characteristics are: depth of the humus horizon, soil reaction, humus content, groundwater level, stoniness of the plough soil and degree of salinity and/or alkalinity. For these characteristics, 86 newly developed Bonite scales and algorithms have been applied (Radovanova, 2020). The software uses the built-in functions of Microsoft Excel. The mechanism of the existing final land use suitability categorisation of the assessed land is fully maintained, but is now considered in a “potential” aspect and uses the new potential algorithms. The software allows the printing of an official document of the potential assessment of the agricultural land concerned.

A new tailored developed software for the risk classification of bridges according to the Italian Guidelines

The “Italian guidelines for the maintenance of bridges” propose a method for the management and classification of risk of infrastructures. The method is structured in different levels. The first three ones (from level 0 to level 2) are applied to all the structures, and the final aim is to determine the “Warning Class” (WaC) of the structure by adopting a qualitative method for the determination of structural, seismic, hydraulic and geotechnical risks (the higher is the WaC, the more critical are the level of risks obtained). The other levels (from level 3 to level 5) require a deeper level of analysis (the assessment of the level of safety of the structure and of the resilience of the network are involved), but are applied only to those structures which are characterized by higher WaC s. Focusing on the first three levels, they are based on: collecting all the data for deepening the knowledge of the bridge, especially focusing on those necessary for the evaluation of the WaC (Level 0); performing the visual inspections for all the risks to assess the conservation status of the structure and of its context (Level 1); determination of the WaC by applying the logical flows proposed by the method (Level 2).In this framework, this paper shows a new software that can support all the operations and gather the relevant data acquired and determined for each bridge and for each level of the method (from level 0 to level 2). This unique software contains all together: the important pieces of information collected in level 0 and 1 for the determination of the WaC; support the technician in the execution of the visual inspections of level 1 for all the risks; automatically determine the WaC, starting from the previously collected data and give the base for a final report about the risk classification of the structure. The input and output data of this software can be used to gather all the key pieces of information about the actual and current status of all the structures and can be used as a powerful instrument to support the organization of the operations and decisions after the determination of the WaC.This is achieved through an eXplainable Artificial Intelligence (XAI) algorithm, a new expert system that overcomes the usual pitfalls of conventional “black box” approaches. This improves transparency and trust in its results and consequently in the decision-making processes.

The g3mclass is a practical software for multiclass classification on biomarkers

The analytes qualified as biomarkers are potent tools to diagnose various diseases, monitor therapy responses, and design therapeutic interventions. The early assessment of the diverseness of human disease is essential for the speedy and cost-efficient implementation of personalized medicine. We developed g3mclass, the Gaussian mixture modeling software for molecular assay data classification. This software automates the validated multiclass classifier applicable to single analyte tests and multiplexing assays. The g3mclass achieves automation using the original semi-constrained expectation–maximization (EM) algorithm that allows inference from the test, control, and query data that human experts cannot interpret. In this study, we used real-world clinical data and gene expression datasets (ERBB2, ESR1, PGR) to provide examples of how g3mclass may help overcome the problems of over-/underdiagnosis and equivocal results in diagnostic tests for breast cancer. We showed the g3mclass output’s accuracy, robustness, scalability, and interpretability. The user-friendly interface and free dissemination of this multi-platform software aim to ease its use by research laboratories, biomedical pharma, companion diagnostic developers, and healthcare regulators. Furthermore, the g3mclass automatic extracting information through probabilistic modeling is adaptable for blending with machine learning and artificial intelligence.

Finite Element Analysis and Near-Infrared Hyperspectral Reflectance Imaging for the Determination of Blueberry Bruise Grading.

Bruising of the subcutaneous tissues of blueberries is an important form of mechanical damage. Different levels of bruising have a significant effect on the post-harvest marketing of blueberries. To distinguish different grades of blueberry bruises and explore the effects of different factors, explicit dynamic simulation and near-infrared hyperspectral reflectance imaging were employed without harming the blueberries in this study. Based on the results of the compression experiment, an explicit dynamic simulation of blueberries was performed to measure the potential locations of bruises and preliminarily divide the bruise stages. A near-infrared hyperspectral reflectance imaging system was used to detect the actual blueberry bruises. According to the blueberry photos taken by the near-infrared hyperspectral reflectance imaging system, the actual bruise rates of blueberries were obtained by using the Environment for Visualizing Images software for training and classification. Bruise grades of blueberries were divided accordingly. Response surface methodology was used to determine the effects of ripeness, loading speed and loading location on the blueberry bruising rate. Under the optimized parameters, the actual damage rate of blueberries was 1.1%. The results provide an important theoretical basis for the accurate and rapid identification and classification of blueberry bruise damage.

SprayNPray: user-friendly taxonomic profiling of genome and metagenome contigs

BackgroundShotgun sequencing of cultured microbial isolates/individual eukaryotes (whole-genome sequencing) and microbial communities (metagenomics) has become commonplace in biology. Very often, sequenced samples encompass organisms spanning multiple domains of life, necessitating increasingly elaborate software for accurate taxonomic classification of assembled sequences.ResultsWhile many software tools for taxonomic classification exist, SprayNPray offers a quick and user-friendly, semi-automated approach, allowing users to separate contigs by taxonomy (and other metrics) of interest. Easy installation, usage, and intuitive output, which is amenable to visual inspection and/or further computational parsing, will reduce barriers for biologists beginning to analyze genomes and metagenomes. This approach can be used for broad-level overviews, preliminary analyses, or as a supplement to other taxonomic classification or binning software. SprayNPray profiles contigs using multiple metrics, including closest homologs from a user-specified reference database, gene density, read coverage, GC content, tetranucleotide frequency, and codon-usage bias.ConclusionsThe output from this software is designed to allow users to spot-check metagenome-assembled genomes, identify, and remove contigs from putative contaminants in isolate assemblies, identify bacteria in eukaryotic assemblies (and vice-versa), and identify possible horizontal gene transfer events.

Development and Validation of an AI-driven Mammographic Breast Density Classification Tool Based on Radiologist Consensus.

Mammographic breast density (BD) is commonly visually assessed using the Breast Imaging Reporting and Data System (BI-RADS) four-category scale. To overcome inter- and intraobserver variability of visual assessment, the authors retrospectively developed and externally validated a software for BD classification based on convolutional neural networks from mammograms obtained between 2017 and 2020. The tool was trained using the majority BD category determined by seven board-certified radiologists who independently visually assessed 760 mediolateral oblique (MLO) images in 380 women (mean age, 57 years ± 6 [SD]) from center 1; this process mimicked training from a consensus of several human readers. External validation of the model was performed by the three radiologists whose BD assessment was closest to the majority (consensus) of the initial seven on a dataset of 384 MLO images in 197 women (mean age, 56 years ± 13) obtained from center 2. The model achieved an accuracy of 89.3% in distinguishing BI-RADS a or b (nondense breasts) versus c or d (dense breasts) categories, with an agreement of 90.4% (178 of 197 mammograms) and a reliability of 0.807 (Cohen κ) compared with the mode of the three readers. This study demonstrates accuracy and reliability of a fully automated software for BD classification. Keywords: Mammography, Breast, Convolutional Neural Network (CNN), Deep Learning Algorithms, Machine Learning Algorithms Supplemental material is available for this article. © RSNA, 2022.

Seawall detection in Florida coastal area from high-resolution imagery using machine learning and OBIA

A methodology and framework were presented to detect the seawalls accurately and efficiently in low coastal areas and were then evaluated in the study area of Hallandale Beach City, Broward County, Florida. Aerial images collected by the Florida Department of Transportation were processed using eCognition Developer software for multiresolution segmentation and classification of objects. Two classification approaches, pixel-based image analysis and the object-based image analysis method, were applied for image classification. However, pixel-based classification was discarded for having less accuracy in output. Three techniques within object-based classification-machine learning (ML) technique, knowledge-based (KB) technique, and ML followed by KB technique were used to compare the most efficient method of classification. While performing the ML technique, three algorithms, such as random forest, support vector machine, and decision tree, were applied to test the best algorithm. Of all the approaches used, the combination of ML and a KB method was able to map the seawall effectively with an overall accuracy of 94%.

Convolutional neural network model and software for classification of typical pests

A model of a convolutional neural network, a dataset for neural network training, and a software tool for the classification of typical insect pests have been developed, which allows recognizing the class of insect pests from an image. The structure of the neural network model was optimized to improve the classification results. In addition, the user interface, authentication, and authorization, data personalization, the presence of user roles and the appropriate distribution of functionality by role, the ability to view statistics on classified insects in a certain period of time were developed. Functional testing of the developed software application on a heterogeneous set of images of insects of 20 different classes was performed.

VA-GCN: A point cloud analysis network used to mine local aggregation information

<h2>Abstract</h2> In recent years, point cloud analysis models have made breakthroughs due to the development of local aggregation operators. In this article, we propose a new point cloud analysis network, Vector Attention Graph Convolution Network (VA-GCN) based on Vector Attention Convolution (VAConv) modules. VA-GCN can be easily embedded into other complex point cloud analysis models, which can promote the development and application of point cloud models in the field of artificial intelligence. In addition, we designed an open-source software for point cloud classification based on VA-GCN. Its ease of use and efficiency will allow non-professionals to quickly get started.

Validation of a Commercial Collar-Based Sensor for Monitoring Eating and Ruminating Behaviour of Dairy Cows.

Simple SummaryThis paper aims to validate a recently released commercial sensor (AFICollar®, Afimilk, Israel) to monitor dairy cows’ feeding and ruminating behavior. In order to evaluate the sensor’s performance under different feeding scenarios, the cows involved were divided into groups and fed different types of feed, including grazed pasture. Further, two version of the software used to convert raw data into behavioral information were tested and compared. Sensor data have been validated against visual observations, which served as the gold standard. Various statistical methods have been employed to assess sensor precision and accuracy. The results indicate that the sensor is adequately accurate for both feeding and ruminating time. However, the precision of the system appeared somewhat limited and should be improved.The use of sensor technologies to monitor cows’ behavior is becoming commonplace in the context of dairy production. This study aimed at validating a commercial collar-based sensor system, the AFICollar® (Afimilk, Kibbutz Afikim, Israel), designed to monitor dairy cattle feeding and ruminating behavior. Additionally, the performances of two versions of the software for behavior classification, the current software AFIfarm® 5.4 and the updated version AFIfarm® 5.5, were compared. The study involved twenty Holstein-Friesian cows fitted with the collars. To evaluate the sensor performance under different feeding scenarios, the animals were divided into four groups and fed three different types of feed (total mixed ration, long hay, animals allowed to graze). Recordings of hourly rumination and feeding time produced by the sensor were compared with visual observation by scan sampling at 1 minute intervals using Spearman correlation, concordance correlation coefficient (CCC), Bland–Altman plots and linear mixed models for assessing the precision and accuracy of the system. The analyses confirmed that the updated software version V5.5 produced better detection performance than the current V5.4. The updated software version produced high correlations between visual observations and data recorded by the sensor for both feeding (r = 0.85, CCC = 0.86) and rumination (r = 0.83, CCC = 0.86). However, the limits of agreement for both behaviors remained quite wide (feeding: −19.60 min/h, 17.46 min/h; rumination: −15.80 min/h, 15.00 min/h). Type of feed did not produce significant effects on the agreement between visual observations and sensor recordings. Overall, the results indicate that the system can provide farmers with adequately accurate data on feeding and rumination time, and can be used to support herd management decisions. Despite all this, the precision of the system remained relatively limited, and should be improved with further developments in the classification algorithm.

HumGut: a comprehensive human gut prokaryotic genomes collection filtered by metagenome data

BackgroundA major bottleneck in the use of metagenome sequencing for human gut microbiome studies has been the lack of a comprehensive genome collection to be used as a reference database. Several recent efforts have been made to re-construct genomes from human gut metagenome data, resulting in a huge increase in the number of relevant genomes. In this work, we aimed to create a collection of the most prevalent healthy human gut prokaryotic genomes, to be used as a reference database, including both MAGs from the human gut and ordinary RefSeq genomes.ResultsWe screened > 5,700 healthy human gut metagenomes for the containment of > 490,000 publicly available prokaryotic genomes sourced from RefSeq and the recently announced UHGG collection. This resulted in a pool of > 381,000 genomes that were subsequently scored and ranked based on their prevalence in the healthy human metagenomes. The genomes were then clustered at a 97.5% sequence identity resolution, and cluster representatives (30,691 in total) were retained to comprise the HumGut collection. Using the Kraken2 software for classification, we find superior performance in the assignment of metagenomic reads, classifying on average 94.5% of the reads in a metagenome, as opposed to 86% with UHGG and 44% when using standard Kraken2 database. A coarser HumGut collection, consisting of genomes dereplicated at 95% sequence identity—similar to UHGG, classified 88.25% of the reads. HumGut, half the size of standard Kraken2 database and directly comparable to the UHGG size, outperforms them both.ConclusionsThe HumGut collection contains > 30,000 genomes clustered at a 97.5% sequence identity resolution and ranked by human gut prevalence. We demonstrate how metagenomes from IBD-patients map equally well to this collection, indicating this reference is relevant also for studies well outside the metagenome reference set used to obtain HumGut. All data and metadata, as well as helpful code, are available at http://arken.nmbu.no/~larssn/humgut/.124P2wWccPtMynGKpmueTbVideo

Breast Cancer Classification Using Deep Convolutional Neural Networks

Breast cancer remains the primary causes of death for women and much effort has been depleted in the form of screening series for prevention. Given the exponential growth in the number of mammograms collected, computer-assisted diagnosis has become a necessity. Histopathological imaging is one of the methods for cancer diagnosis where Pathologists examine tissue cells under different microscopic standards but disagree on the final decision. In this context, the use of automatic image processing techniques resulting from deep learning denotes a promising avenue for assisting in the diagnosis of breast cancer. In this paper, an android software for breast cancer classification using deep learning approach based on a Convolutional Neural Network (CNN) was developed. The software aims to classify the breast tumors to benign or malignant. Experimental results on histopathological images using the BreakHis dataset shows that the DenseNet CNN model achieved high processing performances with 96% of accuracy in the breast cancer classification task when compared with state-of-the-art modelsKeywords— Breast cancer classification, Convolutional Neural Network (CNN), deep learning, DenseNet, histopathological images

Artificial Intelligence in Plasma Cell Myeloma: Neural Networks and Support Vector Machines in the Classification of Plasma Cell Myeloma Data at Diagnosis

Background: Plasma cell neoplasm and/or plasma cell myeloma (PCM) is a mature B-cell lymphoproliferative neoplasm of plasma cells that secrete a single homogeneous immunoglobulin called paraprotein or M-protein. Plasma cells accumulate in the bone marrow (BM) leading to bone destruction and BM failure. Diagnosis of PCM is based on clinical, radiologic, and pathological characteristics. The percent of plasma cells by manual differential (bone marrow morphology), the white blood cell (WBC) count, cytogenetics, fluorescence in situ hybridization (FISH), microarray, and next-generation sequencing of BM are used in the risk stratification of newly diagnosed PCM patients. The genetics of PCM is highly complex and heterogeneous with several genetic subtypes that have different clinical outcomes. National Comprehensive Cancer Network guidelines recommend targeted FISH analysis of plasma cells with specific DNA probes to detect genetic abnormalities for the staging of PCM (4.2021). Recognition of risk categories through training software for classification of high-risk PCM and a novel way of addressing the current approaches through bioinformatics will be a significant step toward automation of PCM analysis. Methods: A new artificial neural network (ANN) classification model was developed and tested in Python programming language with a first data set of 301 cases and a second data set of 176 cases for a total of 477 cases of PCM at diagnosis. Classification model was also developed with support vector machines (SVM) algorithm in R studio and interactive data visuals using Tableau. Results: The resulting ANN algorithm had 94% accuracy for the first and second data sets with a classification summary of precision (PPV): 0.97, recall (sensitivity): 0.76, f1 score: 0.83, and accuracy of logistic regression of 1.0. SVM of plasma cells versus TP53 revealed a 95% accuracy level. Conclusion: A novel classification model based only on specific morphological and genetic variables was developed using a machine learning algorithm, the ANN. ANN identified an association of WBC and BM plasma cell percentage with two of the high-risk genetic categories in the diagnostic cases of PCM. With further training and testing of additional data sets that include morphologic and additional genetic rearrangements, the newly developed ANN model has the potential to develop an accurate classification of high-risk categories of PCM.

Digital Image Analysis Using FloCIA Software for Ornamental Sunflower Ray Floret Color Evaluation.

As an esthetic trait, ray floret color has a high importance in the development of new sunflower genotypes and their market value. Standard methodology for the evaluation of sunflower ray florets is based on International Union for the Protection of New Varieties of Plants (UPOV) guidelines for sunflower. The major deficiency of this methodology is the necessity of high expertise from evaluators and its high subjectivity. To test the hypothesis that humans cannot distinguish colors equally, six commercial sunflower genotypes were evaluated by 100 agriculture experts, using UPOV guidelines. Moreover, the paper proposes a new methodology for sunflower ray floret color classification – digital UPOV (dUPOV), that relies on software image analysis but still leaves the final decision to the evaluator. For this purpose, we created a new Flower Color Image Analysis (FloCIA) software for sunflower ray floret digital image segmentation and automatic classification into one of the categories given by the UPOV guidelines. To assess the benefits and relevance of this method, accuracy of the newly developed software was studied by comparing 153 digital photographs of F2 genotypes with expert evaluator answers which were used as the ground truth. The FloCIA enabled visualizations of segmentation of ray floret images of sunflower genotypes used in the study, as well as two dominant color clusters, percentages of pixels belonging to each UPOV color category with graphical representation in the CIE (International Commission on Illumination) L∗a∗b∗ (or simply Lab) color space in relation to the mean vectors of the UPOV category. Precision (repeatability) of ray flower color determination was greater between dUPOV based expert color evaluation and software evaluation than between two UPOV based evaluations performed by the same expert. The accuracy of FloCIA software used for unsupervised (automatic) classification was 91.50% on the image dataset containing 153 photographs of F2 genotypes. In this case, the software and the experts had classified 140 out of 153 of images in the same color categories. This visual presentation can serve as a guideline for evaluators to determine the dominant color and to conclude if more than one significant color exists in the examined genotype.

A Prototype of EEG System for IoT.

In this work, we develop open source hardware and software for eye state classification and integrate it with a protocol for the Internet of Things (IoT). We design and build the hardware using a reduced number of components and with a very low-cost. Moreover, we propose a method for the detection of open eyes (oE) and closed eyes (cE) states based on computing a power ratio between different frequency bands of the acquired signal. We compare several real- and complex-valued transformations combined with two decision strategies: a threshold-based method and a linear discriminant analysis. Simulation results show both classifier accuracies and their corresponding system delays.

РОЗРОБКА ІНТЕРФЕЙСНОГО МОДУЛЯ ДЛЯ ПРОГРАМНО-АПАРАТНОГО КОМПЛЕКСУ АВТОМАТИЧНОЇ КЛАСИФІКАЦІЇ ТЕКСТІВ

The amount of information presented in text form is constantly increasing. A feature of a large amount of such information in most cases is the lack of its structuredness, which complicates the analysis process. At the stage of document processing in the information and analytical system, an automatic classifier is used – as a program, or as a software and hardware complex that accelerates automatic text processing. In this paper, a software and hardware complex is considered, in which the software part functions on the user's computer, and the hardware part is implemented on the FPGA. Thus, the problem of organizing the connection between its two parts is urgent. While FPGAs have high internal parallelism and very high internal bandwidth, the low bandwidth of the interface between the accelerator and the rest of the system, as well as restrictions on the conditions of its use, is a bottleneck in the field of data transfer. This problem is aggravated in the conditions of the need to create a network system for constructing a classifier, where several user computers connected to the network need access to the hardware-implemented part of the complex (accelerator). The aim of this work is to develop an internal interface module with network communication, in accordance with the technical and functional requirements of the software and hardware complex for automatic text classification. Considering the need to provide network access to the accelerator, as a result of the analysis, the Fast Ethernet interface was chosen as the base one. This interface is the most common LAN standard used by most businesses. For the development of the printed circuit board of the interface module, a comprehensive computer-aided design system Altium Designer was used. After the development of the electrical schematic diagram of the project, a simulation of the controller part was created in the Proteus digital-to-analog simulation program. For the convenience of testing the developed interface module, the Arduino UNO module and the SPI serial interface were used. A feature of the developed module is the support of network-level protocols, it makes it possible to organize access from any personal computer of the local network to an expansion board built on the basis of FPGA. The developed printed circuit board of the interface module was analyzed for possible defects and a list of recommendations for the operating conditions of the device was developed. The interface board is designed taking into account the balanced ratio of price-performance criteria, and is intended for use in any institution where there is a need for automatic classification of text documents.

Impact of Antibiotics on Gut Microbiota Diversity and the Results of a Prospective Dietary Assessment in Patients with Multiple Myeloma Undergoing Autologous Hematopoietic Stem Cell Transplantation

Introduction: The gut microbiome is an important mediator of the host immune system, and studies in allogeneic transplantation demonstrate that loss of microbiome diversity is associated with inferior outcomes (Taur et al, Blood 2014). Loss of gut microbial diversity following autologous stem cell transplantation (ASCT) in multiple myeloma (MM) patients has been associated with reductions in progression-free survival but lacks clear validation from independent transplant centers, and the factors driving these changes are poorly understood (Khan et al, Blood 2018). Evidence from non-oncologic studies suggest that dietary factors and antibiotic exposure may play a key role in understanding the gut microbiota changes observed in the peri-transplant setting (Falony et. al., Science, 2016). Methods We performed a prospective study of changes to the gut microbiome following ASCT in 30 patients with MM using melphalan 200mg/m2 conditioning. Stool samples were collected at 4 timepoints: pre-transplant, post-transplant at engraftment, day +100, and day +365. Samples were processed within 48 hours and stored at -80C until undergoing DNA extraction and 16S rRNA sequencing. Microbiota analysis was performed using qiime and R programming software for species classification (using a pre-trained Silva database) and diversity calculations. Dietary analysis was performed prior to ASCT and at day +100 using the online diet history questionnaire version 3.0 (DHQ-3), a NIH-validated, web-based dietary survey that calculates daily nutrient consumption based on intake. We present dietary and demographic results of the entire cohort prior to transplant and microbiota changes in the pre-transplant and engraftment samples for the first 7 subjects. Results We enrolled 30 patients between 10/2018-6/2019. All patients received uniform infectious prophylaxis including levofloxacin from onset of neutropenia to engraftment or neutropenic fever, acyclovir, and fluconazole. Neutropenic fever was treated with combination vancomycin and cefepime until culture results could guide directed therapy or neutrophil engraftment. Demographics are displayed in table one and reflect a typical transplant-eligible myeloma cohort. DHQ-3 analysis (table 2) revealed deficiencies in fiber, vegetable, and whole grain intake below national guidelines, and female patients demonstrated significantly lower mean caloric intake compared to males (1679 vs 2015 calories/day, P= 0.047). Younger patients aged less than 60 demonstrated higher fiber intake compared to those aged 60 or older (19 vs 16 g/day P= 0.07) although this was not statistically significant. Microbiota changes revealed a significant loss of alpha diversity (P = 0.02) following transplant (figure 1), which was confirmed in a beta-diversity analysis demonstrating that pre- and post-transplant engraftment samples feature distinct microbiota communities (P = 0.002, figure 2). Diversity changes appeared to vary by exposure to antibiotics used for neutropenic fever, although this result was not statistically significant likely due to the limited sample analyzed at this time ( P = 0.068, figure 1). Two out of the 7 patients with available microbiota data demonstrated early disease progression before day 100, and analysis of the taxa changes at a genus level demonstrates a significant bloom of Bacteroides species at engraftment in these subjects (figure 3). Conclusions MM patients undergoing ASCT develop a significant loss of gut microbiome diversity following transplant, which confirms findings observed in other transplant centers. Further analysis of our cohort may determine if antibiotics significantly impact post-transplant diversity. To our knowledge, this is the first investigation to report pre-transplant dietary information in a microbiome study in ASCT, which may be important in understanding microbiome differences between patients and therapies and may guide future investigations. Prior to ASCT, most subjects were deficient in fiber intake, which may have important consequences to the gut microbiome as fiber is a key nutrient source for beneficial microbial species. Updated results from longitudinal study of our cohort will determine the potential impact of dietary intake on microbiome changes, and in turn, how these changes may affect clinical outcomes. Disclosures No relevant conflicts of interest to declare.