应用于真实医疗环境和生命科学领域的人工智能：利用联邦学习解决数据隐私、安全和政策挑战.docx

下载文档

0
0
约2.68万字
约 8页
2024-04-23 发布于境外
举报
版权申诉
保障服务

应用于真实医疗环境和生命科学领域的人工智能：利用联邦学习解决数据隐私、安全和政策挑战.docx

1、本文档共8页，可阅读全部内容。
2、原创力文档（book118）网站文档一经付费（服务费），不意味着购买了该文档的版权，仅供个人/单位学习、研究之用，不得用于商业用途，未经授权，严禁复制、发行、汇编、翻译或者网络传播等，侵权必究。
3、本站所有内容均由合作方或网友上传，本站不对文档的完整性、权威性及其观点立场正确性做任何保证或承诺！文档内容仅供研究参考，付费前请自行鉴别。如您付费，意味着您自己接受本站规则且自行承担风险，本站不退款、不进行额外附加服务；查看《如何避免下载的几个坑》。如果您已付费下载过本站文档，您可以点击这里二次下载。
4、如文档侵犯商业秘密、侵犯著作权、侵犯人身权等，请点击“版权申诉”（推荐），也可以打举报电话：400-050-0827(电话支持时间：9:00-18:30)。

TACKLING

TACKLINGDATAPRIVACY,SECURITYANDPOLICYCHALLENGESWITHFEDERATEDLEARNING|PAGE171

PAGE

PAGE172|TACKLINGDATAPRIVACY,SECURITYANDPOLICYCHALLENGESWITHFEDERATEDLEARNING

ARTIFICIAL

ARTIFICIALINTELLIGENCEINSCIENCE?OECD2023

ARTIFICIAL

ARTIFICIALINTELLIGENCEINSCIENCE?OECD2023

ApplyingAItoreal-worldhealth-caresettingsandthelifesciences:Tacklingdataprivacy,securityandpolicychallengeswithfederatedlearning

M.Galtier,Owkin,France

D.Meadon,Owkin,UnitedKingdom

Introduction

Everyday,millionsofpatientdatapointsarecollectedfromawiderangeofsources.Harnessingtheinsightsheldwithinthisever-increasingvolumeofdatainasafe,secureandethicalmanneristhekeytounlockingbetterhealthoutcomesforeveryone.Throughrecognisingpatternsatascaleunachievablebyhumans,machine-learningmodelscanpowerthediscoveryoftheseinsights.However,whenitcomestohealthcare,thejourneyfrombigdatatoactionableinsightsisfraughtwithchallenges.Healthdataaresensitiveandrequirehandlingwithcarethroughtightregulation.Thisessayexploreshowthischallengecanbeovercomethroughfederatedlearning(FL),anovelmachine-learningtechnology.FLdrasticallyreducesprivacyconcernsbykeepingpatientdatastoredsecurelyonsiteduringmodeltraining.TheessayalsodiscussesthesalientpolicyimplicationstorealisethefullpotentialofFLtechnologyforpatients,health-caresystemsandlifesciencescompanies.

Thedatachallenges

Technologybasedonartificialintelligence(AI),especiallymachine-learningapproaches,canpowerscientificbreakthroughsatanunprecedentedspeedandscale.Thistechnologycanhelpusanswersalientquestionsinmedicalresearch,suchas:whichpatientsshouldbeincludedinclinicaltrials?Whichmoleculesarethemostpromisingtargetsfordrugdevelopment?Whatisthebestwaytoextractinformationfromawidevarietyofmedicalimages,suchas