Research on adverse event classification algorithm of da Vinci surgical robot based on Bert-BiLSTM model DOI Creative Commons

Tianchun Li,

Wanting Zhu, Wenke Xia

et al.

Frontiers in Computational Neuroscience, Journal Year: 2024, Volume and Issue: 18

Published: Dec. 16, 2024

This study aims to enhance the classification accuracy of adverse events associated with da Vinci surgical robot through advanced natural language processing techniques, thereby ensuring medical device safety and protecting patient health. Addressing issues incomplete inconsistent event records, we employed a deep learning model that combines BERT BiLSTM predict whether reports resulted in harm. We developed Bert-BiLSTM-Att_dropout specifically for text tasks small datasets, optimizing model's generalization ability key information capture integration dropout attention mechanisms. Our demonstrated exceptional performance on dataset comprising 4,568 collected from 2013 2023, achieving an average F1 score 90.15%, significantly surpassing baseline models such as GRU, LSTM, BiLSTM-Attention, BERT. achievement not only validates effectiveness within this specific domain but also substantially improves usability reporting, contributing prevention incidents reduction Furthermore, our research experimentally confirmed performance, alleviating data analysis burden healthcare professionals. Through comparative analysis, highlighted potential combining tasks, particularly datasets field. findings advance development monitoring technologies devices provide critical insights future enhancements.

Language: Английский

Improving Clothing Product Quality and Reducing Waste Based on Consumer Review Using RoBERTa and BERTopic Language Model DOI Creative Commons
Andry Alamsyah,

Nadhif Ditertian Girawan

Big Data and Cognitive Computing, Journal Year: 2023, Volume and Issue: 7(4), P. 168 - 168

Published: Oct. 25, 2023

The disposability of clothing has emerged as a critical concern, precipitating waste accumulation due to product quality degradation. Such consequences exert significant pressure on resources and challenge sustainability efforts. In response, this research focuses empowering companies elevate excellence by harnessing consumer feedback. Beyond insights, extends providing suggestions refining improving material handling, gradually mitigating production, cultivating longevity, therefore decreasing discarded clothes. Managing vast influx diverse reviews necessitates sophisticated natural language processing (NLP) techniques. Our study introduces Robustly optimized BERT Pretraining Approach (RoBERTa) model calibrated for multilabel classification BERTopic topic modeling. adeptly distills vital themes from reviews, exhibiting astounding accuracy in projecting concerns across various dimensions quality. NLP’s potential lies endowing with insights into review, augmented the facilitate immersive exploration harvested review topics. This presents thorough case integrating machine learning foster reduction. contribution is notable its integration RoBERTa tasks modeling fashion industry. results indicate that exhibits remarkable performance, demonstrated macro-averaged F1 score 0.87 micro-averaged 0.87. Likewise, achieves coherence 0.67, meaning can form an insightful topic.

Language: Английский

Citations

16
Intelligent Countermeasures Analysis in Oil and Gas Projects Utilizing Topic Modeling DOI

Ehab Elhosary,

Osama Moselhi

Published: Jan. 1, 2025

The oil and gas industry is inherently complex high-risk, with potential fires, explosions, releases of hazardous substances posing significant safety challenges. Despite robust management systems, accidents persist, highlighting the importance learning from past incidents hazard reports. Historical Hazard Operability (HAZOP) reports generate valuable countermeasures—safeguards recommendations—that inform design protection systems to enhance management. However, sheer volume countermeasures produced makes addressing each one prohibitively expensive time-consuming. Additionally, current HAZOP literature software tools lack automation these countermeasures, impeding efficient dissemination information appropriate departments for detailed design. This paper introduces categorizing utilizing BERTopic algorithm in natural language processing (NLP). methodology comprises data preprocessing, SBERT (a modification Bidirectional Encoder Representations Transformers) generating embeddings, Uniform manifold approximation projection (UMAP) dimensionality reduction, hierarchical density-based spatial clustering applications noise (HDBSCAN) clustering, KeyBERT topic representation. Applied 1,574 records a report an pump station, model achieved 84.6% coherence score 90.7% diversity score, resulting 15 final topics, outperforming Latent Dirichlet Allocation (LDA) (45.3% 84.7%). study identified included excluded topics node most frequent by risk rate. generated (SS) were validated against API RP 750 752 standards Countermeasures Breakdown Structure (CBS) was introduced organize hierarchically. research benefits participants improving identification, emphasizing key preventative actions, assigning them relevant design-stage deployment.

Language: Английский

Citations

0
Dynamic Insights: Unraveling Public Demand Evolution in Health Emergencies Through Integrated Language Models and Spatial-Temporal Analysis DOI Creative Commons
Yuan Zhang,

Lin Fu,

Xingyu Guo

et al.

Risk Management and Healthcare Policy, Journal Year: 2024, Volume and Issue: Volume 17, P. 2443 - 2455

Published: Oct. 1, 2024

In public health emergencies, rapid perception and analysis of demands are essential prerequisites for effective crisis communication. Public serve as the most instinctive response to current state a crisis. Therefore, government must promptly grasp leverage information enhance effectiveness efficiency emergency management, that is planned better deal with outbreak meet medical public.

Language: Английский

Citations

0
Mapping microfinance research to sustainable development goals: Insights from Scientometrics and BERTopic analysis DOI
Debidutta Pattnaik, M. Kabir Hassan

Journal of Economic Surveys, Journal Year: 2024, Volume and Issue: unknown

Published: Nov. 12, 2024

Abstract Microfinance research plays a pivotal role in addressing global development challenges, yet comprehensive assessments of its alignment with sustainable goals (SDGs) remain scarce. This study fills this gap by systematically mapping the landscape microfinance literature to corresponding SDG contributions. Leveraging data from Scopus and SCIval, we analyzed 1004 articles spanning between 2014 2023. Our findings reveal substantial body focused on poverty alleviation (SDG 1) economic empowerment 8), notable attention gender equality 5) reduced inequalities 10). Key thematic clusters include performance impact institutions (MFIs), microinsurance innovations, Islamic microfinance. Notably, top‐cited underscored sector's commitment growth, nuanced exploration dynamics rural household impacts. Furthermore, our BERTopic analysis unveils multidimensional nature research, highlighting prevalent themes such as MFI impact. Geographically, efforts are concentrated United States, India, France, reflecting SDG‐aligned interventions. The paper's theoretical contributions lie framework development, interdisciplinary engagement, understanding themes, perspective, methodological advancements, all which enhance scholarly discourse microfinance's achieving SDGs.

Language: Английский

Citations

0
Evaluating Software Quality Through User Reviews: The ISOftSentiment Tool DOI
Fang Hou, Feng Liang, Siamak Farshidi

et al.

Lecture notes in computer science, Journal Year: 2024, Volume and Issue: unknown, P. 75 - 91

Published: Nov. 27, 2024

Language: Английский

Citations

0
Research on adverse event classification algorithm of da Vinci surgical robot based on Bert-BiLSTM model DOI Creative Commons

Tianchun Li,

Wanting Zhu, Wenke Xia

et al.

Frontiers in Computational Neuroscience, Journal Year: 2024, Volume and Issue: 18

Published: Dec. 16, 2024

This study aims to enhance the classification accuracy of adverse events associated with da Vinci surgical robot through advanced natural language processing techniques, thereby ensuring medical device safety and protecting patient health. Addressing issues incomplete inconsistent event records, we employed a deep learning model that combines BERT BiLSTM predict whether reports resulted in harm. We developed Bert-BiLSTM-Att_dropout specifically for text tasks small datasets, optimizing model's generalization ability key information capture integration dropout attention mechanisms. Our demonstrated exceptional performance on dataset comprising 4,568 collected from 2013 2023, achieving an average F1 score 90.15%, significantly surpassing baseline models such as GRU, LSTM, BiLSTM-Attention, BERT. achievement not only validates effectiveness within this specific domain but also substantially improves usability reporting, contributing prevention incidents reduction Furthermore, our research experimentally confirmed performance, alleviating data analysis burden healthcare professionals. Through comparative analysis, highlighted potential combining tasks, particularly datasets field. findings advance development monitoring technologies devices provide critical insights future enhancements.

Language: Английский

Citations

0