Research
Classification of imbalanced ECGs through segmentation models and augmented by conditional diffusion model
Electrocardiograms (ECGs) provide essential data for diagnosing arrhythmias, which can potentially cause serious health complications. Early detection through continuous monitoring is crucial for timely intervention. The Massachusetts Institute of Technology-Beth Israel Hospital (MIT-BIH) arrhythmia dataset employed for arrhythmia analysis research comprises imbalanced data. It is necessary to create a robust model independent of data imbalances to classify arrhythmias accurately. To mitigate the pronounced class imbalance in the MIT-BIH arrhythmia dataset, this study employs advanced augmentation techniques, specifically variational autoencoder (VAE) and conditional diffusion, to augment the dataset. Furthermore, accurately segmenting the continuous heartbeat dataset into individual heartbeats is crucial for confidently detecting arrhythmias. This research compared a model that employed annotation-based segmentation, utilizing R-peak labels, and a model that utilized an automated segmentation method based on a deep learning model to segment heartbeats. In our experiments, the proposed model, utilizing MobileNetV2 along with annotation-based segmentation and conditional diffusion augmentation to address minority class, demonstrated a notable 1.23% improvement in the F1 score and 1.73% in the precision, compared to the model classifying arrhythmia classes with the original imbalanced dataset.
A criteria-based classification model using augmentation and contrastive learning for analyzing imbalanced statement data
Criteria Based Content Analysis (CBCA) is a forensic tool that analyzes victim statements. It involves the categorization of victims' statements into 19 distinct criteria classifications, playing a crucial role in evaluating the authenticity of testimonies by discerning whether they are rooted in genuine experiences or fabricated accounts. The exclusion of subjective opinions becomes imperative to assess statements through this forensic tool objectively. This study proposes developing an objective classification model for CBCA-based statement analysis using natural language processing techniques. Nevertheless, achieving optimal classification performance proves challenging due to imbalances in data distribution among the various criterion classifications. To enhance the accuracy and reliability of the classification model, this research employs data augmentation techniques and dual contrastive learning methods for fine-tuning the RoBERTa language model.
Intent Classification and Named Entity Recognition Using Augmentation
In cases of child sexual abuse, interviewing and obtaining trustworthy statements from victims and witnesses is essential because their statements are the only evidence. It is crucial to ascertain objectively the credibility of the victim’s statements, which may vary based on the nature of the questions posed by the forensic interviewer. Therefore, interview skills that eliminate subjective opinions require a high level of training for forensic interviewers. To reduce high-risk subjective interviews, objectively analyzing statements is essential. Understanding the victim’s intent and named entity recognition (NER) in the statements is necessary to give the victim open-ended questions and memory recall. Therefore, the system provides an intent classification and NER method that follows the National Institute of Child Health and Human Development Investigative Interview Protocol, which outlines the collection of objective statements. Large language models such as BERT and KoBERT, along with data augmentation techniques, were proposed using a restricted training dataset of limited size to achieve effective intent classification and NER performance. Additionally, a system that can collect objective statements with the proposed model was developed and it was confirmed that it could assist statement analysts. The verification results showed that the model achieved average F1-scores of 95.5% and 97.8% for intent classification and NER, respectively, which improved the results of the limited data by 3.4% and 3.7%, respectively.
ASL
American Sign Language (ASL) images can be used as a communication tool by determining numbers and letters using the shape of the fingers. Particularly, ASL can have an key role in communication for hearing-impaired persons and conveying information to other persons, because sign language is their only channel of expression. Representative ASL recognition methods primarily adopt images, sensors, and pose-based recognition techniques, and employ various gestures together with hand-shapes. This study briefly reviews these attempts at ASL recognition and provides an improved ASL classification model that attempts to develop a deep learning method with meta-layers. In the proposed model, the collected ASL images were clustered based on similarities in shape, and clustered group classification was first performed, followed by reclassification within the group. The experiments were conducted with various groups using different learning layers to improve the accuracy of individual image recognition. After selecting the optimized group, we proposed a meta-layered learning model with the highest recognition rate using a deep learning method of image processing. The proposed model exhibited an improved performance compared with the general classification model.
cPlot
Advances in the next-generation sequencing technology have led to a dramatic decrease in read-generation cost and an increase in read output. Reconstruction of short DNA sequence reads generated by next-generation sequencing requires a read alignment method that reconstructs a reference genome. In addition, it is essential to analyze the results of read alignments for a biologically meaningful inference. However, read alignment from vast amounts of genomic data from various organisms is challenging in that it involves repeated automatic and manual analysis steps. We, here, devised cPlot software for read alignment of nucleotide sequences, with automated read alignment and position analysis, which allows visual assessment of the analysis results by the user. cPlot compares sequence similarity of reads by performing multiple read alignments, with FASTA format files as the input. This application provides a web-based interface for the user for facile implementation, without the need for a dedicated computing environment. cPlot identifies the location and order of the sequencing reads by comparing the sequence to a genetically close reference sequence in a way that is effective for visualizing the assembly of short reads generated by NGS and rapid gene map construction.
ReGSP
The massively parallel nature of next-generation sequencing technologies has contributed to the generation of massive sequence data in the last two decades. Deciphering the meaning of each generated sequence requires multiple analysis tools, at all stages of analysis, from the reads stage all the way up to the whole-genome level. Homology-based approaches based on related reference sequences are usually the preferred option for gene and transcript prediction in newly sequenced genomes, resulting in the popularity of a variety of BLAST and BLAST-based tools. For organelle genomes, a single-reference–based gene finding tool that uses grouping parameters for BLAST results has been implemented in Genome Search Plotter (GSP). However, this tool does not accept multiple and user-customized reference sequences required for a broad homology search. Here, we present multiple-Reference–based Gene Search and Plot (ReGSP), a simple and convenient web tool that accepts multiple reference sequences for homology-based gene search. The tool incorporates cPlot, a novel dot plot tool, for illustrating nucleotide sequence similarity between the query and the reference sequences. ReGSP has an easy-to-use web interface and is freely accessible at
geneCo
In comparative and evolutionary genomics, a detailed comparison of common features between organisms is essential to evaluate genetic distance. However, identifying differences in matched and mismatched genes among multiple genomes is difficult using current comparative genomic approaches due to complicated methodologies or the generation of meager information from obtained results. This study describes a visualized software tool, geneCo (gene Comparison), for comparing genome structure and gene arrangements between various organisms. User data are aligned, gene information is recognized, and genome structures are compared based on user-defined GenBank files. Information regarding inversion, gain, loss, duplication, and gene rearrangement among multiple organisms being compared is provided by geneCo, which uses a web-based interface that users can easily access without any need to consider the computational environment.
AGORA
Next-generation sequencing (NGS) technologies have led to the accumulation of high-throughput sequence data from various organisms in biology. To apply gene annotation of organellar genomes for various organisms, more optimized tools for functional gene annotation are required. Almost all gene annotation tools are mainly focused on the chloroplast genome of land plants or the mitochondrial genome of animals. We have developed a web application AGORA for the fast, user-friendly and improved annotations of organellar genomes. Annotator for Genes of Organelle from the Reference sequence Analysis (AGORA) annotates genes based on a basic local alignment search tool (BLAST)-based homology search and clustering with selected reference sequences from the NCBI database or user-defined uploaded data. AGORA can annotate the functional genes in almost all mitochondrion and plastid genomes of eukaryotes. The gene annotation of a genome with an exon–intron structure within a gene or inverted repeat region is also available. It provides information of start and end positions of each gene, BLAST results compared with the reference sequence and visualization of gene map by OGDRAW.
Genome Search Plotter
Big data research on genomic sequence analysis has accelerated considerably with the development of next-generation sequencing. Currently, research on genomic sequencing has been conducted using various methods, ranging from the assembly of reads consisting of fragments to the annotation of genetic information using a database that contains known genome information. According to the development, most tools to analyze the new organelles’ genetic information requires different input formats such as FASTA, GeneBank (GB) and tab separated files. The various data formats should be modified to satisfy the requirements of the gene annotation system after genome assembly. In addition, the currently available tools for the analysis of organelles are usually developed only for specific organisms, thus the need for gene prediction tools, which are useful for any organism, has been increased. The proposed method—termed the genome_search_plotter—is designed for the easy analysis of genome information from the related references without any file format modification. Anyone who is interested in intracellular organelles such as the nucleus, chloroplast, and mitochondria can analyze the genetic information using the assembled contig of an unknown genome and a reference model without any modification of the data from the assembled contig.
An LSTM method for natural pronunciation expression of foreign words in sentences
Korea language has postpositions such as eul, reul, yi, ga, wa, and gwa, which are attached to nouns and add meaning to the sentence. When foreign notations or abbreviations are included in sentences, the appropriate postposition for the pronunciation of the foreign words may not be used. Sometimes, for natural expression of the sentence, two postpositions are used with one in parentheses as in “eul(reul)” so that both postpositions can be acceptable. This study finds examples of using unnatural postpositions when foreign words are included in Korean sentences and proposes a method for using natural postpositions by learning the final consonant pronunciation of nouns. The proposed method uses a recurrent neural network model to naturally express postpositions connected to foreign words. Furthermore, the proposed method is proven by learning and testing with the proposed method. It will be useful for composing perfect sentences for machine translation by using natural postpositions for English abbreviations or new foreign words included in Korean sentences in the future.