Artificial Intelligence and Biosensors in Healthcare and Its Clinical Relevance: A Review

IEEE Access, Journal Year: 2023, Volume and Issue: 11, P. 61600 - 61620

Published: Jan. 1, 2023

Data generated from sources such as wearable sensors, medical imaging, personal health records, pathology and public organizations have resulted in a massive information increase the sciences over last decade. Advances computational hardware, cloud computing, Graphical Processing Units (GPUs), Tensor (TPUs), provide means to utilize these data. Consequently, many Artificial Intelligence (AI)-based methods been developed infer large healthcare Here, we present an overview of recent progress artificial intelligence biosensors life sciences. We discuss role machine learning precision medicine, for Internet Things (IoT). review most advancements biosensing technologies that use AI assist monitoring bodily electro-physiological electro-chemical signals disease diagnosis, demonstrating trend towards personalized medicine with highly effective, inexpensive, precise point-of-care treatment. Furthermore, advances computing technologies, accelerated intelligence, edge federated data, are also documented. Finally, investigate challenges data-driven approaches, potential issues IoT-based generate, distribution shifts occur among different data modalities, concluding future prospects.

Language: Английский

---

A scoping review of artificial intelligence-based methods for diabetes risk prediction

npj Digital Medicine, Journal Year: 2023, Volume and Issue: 6(1)

Published: Oct. 25, 2023

Abstract The increasing prevalence of type 2 diabetes mellitus (T2DM) and its associated health complications highlight the need to develop predictive models for early diagnosis intervention. While many artificial intelligence (AI) T2DM risk prediction have emerged, a comprehensive review their advancements challenges is currently lacking. This scoping maps out existing literature on AI-based prediction, adhering PRISMA extension Scoping Reviews guidelines. A systematic search longitudinal studies was conducted across four databases, including PubMed, Scopus, IEEE-Xplore, Google Scholar. Forty that met our inclusion criteria were reviewed. Classical machine learning (ML) dominated these studies, with electronic records (EHR) being predominant data modality, followed by multi-omics, while medical imaging least utilized. Most employed unimodal AI models, only ten adopting multimodal approaches. Both showed promising results, latter superior. Almost all performed internal validation, but five external validation. utilized area under curve (AUC) discrimination measures. Notably, provided insights into calibration models. Half used interpretability methods identify key predictors revealed Although minority highlighted novel predictors, majority reported commonly known ones. Our provides valuable current state limitations highlights development clinical integration.

Language: Английский

---

Decoding the exposome: Data science methodologies and implications in Exposome-Wide association studies (ExWASs)

Exposome, Journal Year: 2024, Volume and Issue: unknown

Published: Jan. 17, 2024

Abstract This paper explores the exposome concept and its role in elucidating interplay between environmental exposures human health. We introduce two key concepts critical for exposomics research. Firstly, we discuss joint impact of genetics environment on phenotypes, emphasizing variance attributable to shared non-shared factors, underscoring complexity quantifying exposome's influence health outcomes. Secondly, importance advanced data-driven methods large cohort studies exposomic measurements. Here, exposome-wide association study (ExWAS), an approach designed systematic discovery relationships phenotypes various exposures, identifying significant associations while controlling multiple comparisons. advocate standardized use term “exposome-wide study, ExWAS,” facilitate clear communication literature retrieval this field. The aims guide future researchers understanding evaluating studies. Our discussion extends emerging topics, such as FAIR Data Principles, biobanked healthcare datasets, functional exposome, outlining directions abstract provides a succinct overview our comprehensive complex dynamics implications

Language: Английский

---

Applications of AI in multi-modal imaging for cardiovascular disease

Frontiers in Radiology, Journal Year: 2024, Volume and Issue: 3

Published: Jan. 12, 2024

Data for healthcare is diverse and includes many different modalities. Traditional approaches to Artificial Intelligence cardiovascular disease were typically limited single With the proliferation of datasets new methods in AI, we are now able integrate modalities, such as magnetic resonance scans, computerized tomography echocardiography, x-rays, electronic health records. In this paper, review research from last 5 years applications AI multi-modal imaging. There have been promising results registration, segmentation, fusion imaging modalities with each other computer but there still challenges that need be addressed. Only a few papers addressed x-ray, or non-imaging As prediction classification tasks, only couple use multiple domain. Furthermore, no models implemented tested real world clinical settings.

Language: Английский

---

An overview of methods and techniques in multimodal data fusion with application to healthcare

International Journal of Data Science and Analytics, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 10, 2025

Language: Английский

---

Harnessing Big Data Analytics for Healthcare: A Comprehensive Review of Frameworks, Implications, Applications, and Impacts

IEEE Access, Journal Year: 2023, Volume and Issue: 11, P. 112891 - 112928

Published: Jan. 1, 2023

Big Data Analytics (BDA) has garnered significant attention in both academia and industries, particularly sectors such as healthcare, owing to the exponential growth of data advancements technology. The integration from diverse sources utilization advanced analytical techniques potential revolutionize healthcare by improving diagnostic accuracy, enabling personalized medicine, enhancing patient outcomes. In this paper, we aim provide a comprehensive literature review on application big analytics focusing its ecosystem, applications, sources. To achieve this, an extensive analysis scientific studies published between 2013 2023 was conducted overall 180 were thoroughly evaluated, establishing strong foundation for future research identifying collaboration opportunities domain. study delves into various areas BDA highlights successful implementations, explores their enhance outcomes while reducing costs. Additionally, it outlines challenges limitations associated with discusses modelling tools techniques, showcases deployed solutions, presents advantages through real-world use cases. Furthermore, identifies key open field aiming push boundaries contribute enhanced decision-making processes.

Language: Английский

---

Review of multimodal machine learning approaches in healthcare

Information Fusion, Journal Year: 2024, Volume and Issue: unknown, P. 102690 - 102690

Published: Sept. 1, 2024

Language: Английский

---

The performance of large language models on fictional consult queries indicates favorable potential for AI-assisted vascular surgery consult handling

JVS-Vascular Insights, Journal Year: 2024, Volume and Issue: 2, P. 100052 - 100052

Published: Jan. 1, 2024

Type of Research: Cross-sectional study Key Findings: Readily available large language models can identify vascular surgery emergencies with an accuracy rate from 76% to 100%. Models select the correct next most important management steps between 36% and 68% cases. 89.5% generative free-responses adhere scientific consensus, while 17.5% missed information. Take home Message: Existing reliably based on clinical vignettes. However, ability recommend treatment requires further fine-tuning. IntroductionRecently, use in medicine has become a prominent topic discussion due rapid improvement these tools understanding responding natural language. Several are widely public, both proprietary open-sourced. We aim evaluate possible such LLMs by their abilities process common consult requests.MethodsThe senior author created twenty-five fictional consultation queries requests. Five attending surgeons four (GPT 3.5, GPT 4, Bard, Falcon 40B) were asked answer whether each was emergency that needed immediate attention within hour. Responders also best step examination, additional imaging, or urgent operation. 3.5 4 provided free-response answers step, graded accuracy, harm, content completeness.ResultsThe rates accurate identification 88%, 100%, 76%, 88% for 40B, respectively. While they have similar overall high sensitivity at Bard specificity 90%. 4.0 had 100% specificity. agreed majority surgeon opinion 64% 3.5), 32% 4), (Falcon 40B), (Bard) collective ratio adhering consensus. Only 5% responses highly likely cause clinically significant harm. only 4% included incorrect contact, content. There no difference regarding grade.ConclusionExisting, exhibited solid emergencies, agreeing attendings continue identifiable deficiencies recommendations, higher-level task. Future might help triage incoming consults provide preliminary suggestions. The utility practice remains be explored.

Language: Английский

---

A Graph Neural Network-based Traffic Flow Prediction System with Enhanced Accuracy and Urban Efficiency

Deleted Journal, Journal Year: 2024, Volume and Issue: 19(4), P. 336 - 349

Published: Jan. 25, 2024

This research culminates in a robust Traffic Flow Prediction System poised to redefine the landscape of Intelligent Transportation Systems (ITS). Our findings highlight substantial promise this system through meticulously structured methodology spanning data generation, dynamic network construction, multi-modal integration, and employment state-of-the-art Graph Neural Networks (GNNs). Notably, "Current Framework" stands out, demonstrating superior performance over alternative regression models, substantiated by remarkable 35% reduction Mean Squared Error (MSE) commendable 7% increase R-squared (R²). Nevertheless, is not without its caveats. Ongoing model refinement, adaptability ever-evolving traffic landscape, scalability considerations are essential for future exploration. These achievements usher new era management, with potential curtail congestion up 20%, bolster safety measures, an enhanced urban transportation efficiency.

Language: Английский

---

Multi-modal data clustering using deep learning: A systematic review

Neurocomputing, Journal Year: 2024, Volume and Issue: 607, P. 128348 - 128348

Published: Aug. 14, 2024

Language: Английский

---