基于人工智能特征融合的多模态在抑郁症识别中的应用.pdfVIP

58ModernInstrumentsMedicalTre024Vo1.30N0.

综述

基于人工智能特征融合的多模态在抑郁症识别中的应用

谭翻弟马晓兰王元元陶田田王国燕黄生辉*

（1.甘肃中医药大学中西医结合学院，兰州730000；2.甘肃省中医院神志病科，兰州730050）

【摘要】抑郁症是一种常见的精神障碍疾病，早期的检测和诊断对抑郁症预防和治疗至关重要。人工智能已经参与抑

郁症早期识别与干预，本文从CNKI和GoogleScholar中分别以关键词“人工智能”“文本特征”“面部特征”“语音特征”“脑核

磁特征”和“多模态”进行文献检索，再结合主题筛选精读并使用追溯法获得相关有代表性文献，综述“文本特征”“语音特

征”“面部特征”“脑核磁特征”和“多模态”等技术在抑郁症识别与诊断中的应用，并讨论其优势、不足与展望。

【关键词】人工智能；多模态；抑郁症；识别；机器学习；研究进展

TH77;R31810.11876/mimt202401011

【中图分类号】T【文献标志码】A【DOI]

Applicationofmulti-modalitybasedonartificialintelligencefeature

fusionindepressionrecognition

TanFandi,MaXiaolan,WangYuanyuan,TaoTiantian,WangGuoyan,HuangShenghui?

(1.SchoolofIntegratedTraditionalChineseandWesternMedicine,GansuUniversityofChineseMedicine,Lanzhou730030,China;

2.DepartmentofPsychiatry,GansuProvincialHospitalofTraditionalChineseMedicine,Lanzhou730050,China)

Correspondingauthor:HuangShenghui,E-mail:534044812@qq.com

[Abstract]

Depressionisacommonmentaldisorder,andearlydetectionanddiagnosisarecrucialforthepreventionand

treatmentofdepression.Artificialintelligencehasbeeninvolvedintheearlyidentificationandinterventionofdepression.Inthispaper,

thekeywords“artificialintelligence,“textfeatures,“speechfeatures,“facialfeatures,“cerebralnuclearmagneticfeatures

and“multimodalwereusedforliteraturesearchfromCNKIandGoogleScholarrespectively.Then,relevantrepresentativeliterature

wasobtainedbyintensivereadingcombinedwithsubjectscreeningandretrospectivemethod,andtheapplicationof“textfeature,

speechfeature,“facialfeature,“brainnuclearmagneticfeaturean