A residual attention-based privacy-preserving biometrics model of transcriptome prediction from genome DOI
Tian Cheng, Song Liu, Jinbao Li

et al.

Published: Nov. 1, 2023

Transcriptome prediction from genetic variation data is an important task in the privacy-preserving and biometrics field, which can better protect genomic achieve biometric recognition through transcriptome. Many transcriptome methods have achieved good accuracy data. However, these traditional problems of linear assumption, overfitting, expose personal privacy, extensive manual optimization. To solve shortcomings, we propose attention-based model named RATPM that improves protects participant In RATPM, introduce improve deep learning with multi-head self-attention into stage Predixcan, uncovers non-linear relationship between Moreover, a residual attention module to generate attention-aware features extract more accurate at different levels variation. Furthermore, BERT pre-training encode fully utilizing their contextual information. Our research enables scientific institutions publish only predicted transcriptomic for purposes, thus protecting information subjects. Finally, evaluated our on 1000 Genomes Geuvadis projects datasets compare other baseline models.

Language: Английский

Pre-trained multimodal large language model enhances dermatological diagnosis using SkinGPT-4 DOI Creative Commons

Juexiao Zhou,

Xiaonan He,

Liyuan Sun

et al.

Nature Communications, Journal Year: 2024, Volume and Issue: 15(1)

Published: July 5, 2024

Abstract Large language models (LLMs) are seen to have tremendous potential in advancing medical diagnosis recently, particularly dermatological diagnosis, which is a very important task as skin and subcutaneous diseases rank high among the leading contributors global burden of nonfatal diseases. Here we present SkinGPT-4, an interactive dermatology diagnostic system based on multimodal large models. We aligned pre-trained vision transformer with LLM named Llama-2-13b-chat by collecting extensive collection disease images (comprising 52,929 publicly available proprietary images) along clinical concepts doctors’ notes, designing two-step training strategy. quantitatively evaluated SkinGPT-4 150 real-life cases board-certified dermatologists. With users could upload their own photos for autonomously evaluate images, identify characteristics categories conditions, perform in-depth analysis, provide treatment recommendations.

Language: Английский

Citations

30
Privacy-preserving approach for IoT networks using statistical learning with optimization algorithm on high-dimensional big data environment DOI Creative Commons
Fatma S. Alrayes, Mohammed Maray, Asma Alshuhail

et al.

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: Jan. 27, 2025

Language: Английский

Citations

3
An AI Agent for Fully Automated Multi‐Omic Analyses DOI Creative Commons
Juexiao Zhou, Bin Zhang, Guowei Li

et al.

Advanced Science, Journal Year: 2024, Volume and Issue: 11(44)

Published: Oct. 3, 2024

Abstract With the fast‐growing and evolving omics data, demand for streamlined adaptable tools to handle bioinformatics analysis continues grow. In response this need, Automated Bioinformatics Analysis (AutoBA) is introduced, an autonomous AI agent designed explicitly fully automated multi‐omic analyses based on large language models (LLMs). AutoBA simplifies analytical process by requiring minimal user input while delivering detailed step‐by‐step plans various tasks. AutoBA's unique capacity self‐design processes data variations further underscores its versatility. Compared with online bioinformatic services, offers multiple LLM backends, options both local usage, prioritizing security privacy. comparison ChatGPT open‐source LLMs, code repair (ACR) mechanism in improve stability end‐to‐end Moreover, different from predefined pipeline, has adaptability sync emerging tools. Overall, represents advanced convenient tool, offering robustness conventional analyses.

Language: Английский

Citations

7
Interpretable vertical federated learning with privacy-preserving multi-source data integration for prognostic prediction DOI
Qingyong Wang

Engineering Applications of Artificial Intelligence, Journal Year: 2025, Volume and Issue: 148, P. 110408 - 110408

Published: March 9, 2025

Language: Английский

Citations

1
AI-powered precision medicine: utilizing genetic risk factor optimization to revolutionize healthcare DOI Creative Commons
Sakhaa B. Alsaedi, Michihiro Ogasawara,

Mohammed Alarawi

et al.

NAR Genomics and Bioinformatics, Journal Year: 2025, Volume and Issue: 7(2)

Published: March 29, 2025

Abstract The convergence of artificial intelligence (AI) and biomedical data is transforming precision medicine by enabling the use genetic risk factors (GRFs) for customized healthcare services based on individual needs. Although GRFs play an essential role in disease susceptibility, progression, therapeutic outcomes, a gap exists exploring their contribution to AI-powered medicine. This paper addresses this need investigating significance potential utilizing with AI medical field. We examine applications, particularly emphasizing impact prediction, treatment personalization, overall improvement. review explores application algorithms optimize GRFs, aiming advance screening, patient stratification, drug discovery, understanding mechanisms. Through variety case studies examples, we demonstrate incorporating facilitated into practice, resulting more precise diagnoses, targeted therapies, improved outcomes. underscores empowered AI, enhance improving diagnostic accuracy, precision, individualized solutions.

Language: Английский

Citations

1
Comprehensive review on single-cell RNA sequencing: A new frontier in Alzheimer's disease research DOI
Wengang Jin, Jinjin Pei, Jeane Rebecca Roy

et al.

Ageing Research Reviews, Journal Year: 2024, Volume and Issue: 100, P. 102454 - 102454

Published: Aug. 12, 2024

Language: Английский

Citations

4
From challenges and pitfalls to recommendations and opportunities: Implementing federated learning in healthcare DOI Creative Commons
Ming Li, Pengcheng Xu, Junjie Hu

et al.

Medical Image Analysis, Journal Year: 2025, Volume and Issue: 101, P. 103497 - 103497

Published: Feb. 14, 2025

Federated learning holds great potential for enabling large-scale healthcare research and collaboration across multiple centers while ensuring data privacy security are not compromised. Although numerous recent studies suggest or utilize federated based methods in healthcare, it remains unclear which ones have clinical utility. This review paper considers analyzes the most up to May 2024 that describe healthcare. After a thorough review, we find vast majority appropriate use due their methodological flaws and/or underlying biases include but limited concerns, generalization issues, communication costs. As result, effectiveness of is significantly To overcome these challenges, provide recommendations promising opportunities might be implemented resolve problems improve quality model development with

Language: Английский

Citations

0
ADPHE-FL: Federated learning method based on adaptive differential privacy and homomorphic encryption DOI
Tao Wu, Yulin Deng, Qizhao Zhou

et al.

Peer-to-Peer Networking and Applications, Journal Year: 2025, Volume and Issue: 18(3)

Published: April 5, 2025

Language: Английский

Citations

0
Federated Machine Learning Enables Risk Management and Privacy Protection in Water Quality DOI
Yuqi Wang, Hongcheng Wang, Wenzhe Wang

et al.

Environmental Science & Technology, Journal Year: 2025, Volume and Issue: unknown

Published: May 16, 2025

Real-time water quality risk management in wastewater treatment plants (WWTPs) requires extensive data, and data sharing is still just a slogan due to privacy issues. Here we show an adaptive system federated averaging (AWSFA) framework based on learning (FL), where the model does not access but uses parameters trained by raw data. The study collected from six WWTPs between 2018 2024, developed 10 machine models for each effluent indicator, with best performance bidirectional long-term memory network (BM) as Baseline. Compared direct training classical (FedAvg), AWSFA reduces mean absolute percentage error (MAPE) of BM significantly. Analysis input dimensions, set size, interpretability reveals that improvement driven complexity algorithm design via parameter sharing. By simulation possible disturbances quality, remained robust when 50% key features were missing. provides way forward preservation systems offers theoretical support digital transformation era big model.

Language: Английский

Citations

0
A unified method to revoke the private data of patients in intelligent healthcare with audit to forget DOI Creative Commons
Juexiao Zhou, Haoyang Li, Xingyu Liao

et al.

Nature Communications, Journal Year: 2023, Volume and Issue: 14(1)

Published: Oct. 6, 2023

Abstract Revoking personal private data is one of the basic human rights. However, such right often overlooked or infringed upon due to increasing collection and use patient for model training. In order secure patients’ be forgotten, we proposed a solution by using auditing guide forgetting process, where means determining whether dataset has been used train requires information query forgotten from target model. We unified these two tasks introducing an approach called knowledge purification. To implement our solution, developed audit forget software (AFS), which able evaluate revoke pre-trained deep learning models. Here, show usability AFS its application potential in real-world intelligent healthcare enhance privacy protection revocation

Language: Английский

Citations

7