Federated Learning for Diabetes Prediction: Investigating Client-Level Statistical Heterogeneity
编号:217 访问权限:仅限参会人 更新:2026-11-16 09:59:13 浏览:7次 In-person

报告开始:2026年11月20日 15:15(Asia/Kolkata)

报告时间:15min

所在会场:[S4] Artificial Intelligence and Intelligent Computing [S4-1] Artificial Intelligence and Intelligent Computing

暂无文件

摘要
Diabetes prediction using machine learning can support  early  identification  of  individuals  at  risk,  however, existing centralized methods rely on patient data being collected and  aggregated  into  a  central  repository.  This  raises  data sharing and data governance issues for healthcare organizations that  must  cooperate  for  this  purpose.  Another  approach  is federated   learning   in   which   a   joint   model   is   trained cooperatively by healthcare organizations and raw data does not leave the respective organization. Although the performance of federated learning depends on the degree of class imbalance, differences in data distributions on the side of the clients, as well as the number of clients, its exact effect is yet unknown. This research explores the above factors via a comparative study of centralized  and  federated  learning  for  predicting  diabetes. Logistic regression is the base prediction model used, while class imbalance  is  solved  by  SMOTENC  algorithm.  In  federated learning, local SMOTENC technique is performed separately for each client, local training is performed using FedProx, and server aggregation is performed using sample size weighted average  aggregation  technique.  Evaluation  is  performed  for both IID and non-IID client data distributions, where non-IID client   data   distributions   are   performed   using   Dirichlet distribution (α = 0.30). The study also examines the impact of the  change  of  the  number  of  clients  from  3  to  5  and  10. Evaluation of the resulting models is done on the same unseen testing set, using the accuracy, precision, recall, F1 score, ROC- AUC, and PR-AUC metrics. The outcomes allow comparing the performance   of   centralized   and   federated   approaches   to diabetes prediction.
关键词
Federated Learning,Diabetes Prediction,logistic regression,class imbalance,SMOTENC,FedProx,Client heterogeneity
报告人
Sanjana R
Student Rajalakshmi Engineering College

稿件作者
Aswana Lal Rajalakshmi Engineering College
Prathyush R Rajalakshmi Engineering College
Sanjana R Rajalakshmi Engineering College
Srevarsha N Rajalakshmi Engneering College
Naveed Ahmed Basha Rajalakshmi Engineering College
发表评论
验证码 看不清楚,更换一张
全部评论
重要日期
  • 会议日期

    11月20日

    2026

    至

    11月22日

    2026

  • 10月31日 2026

    初稿截稿日期

  • 11月07日 2026

    提前注册日期

  • 11月22日 2026

    注册截止日期

主办单位
United Societies of Science
承办单位
Pondicherry University
协办单位
IEEE Section
IEEE Madras Section
移动端
在手机上打开
小程序
打开微信小程序
客服
扫码或点此咨询