Page 72 - 南京医科大学自然版
P. 72
第46卷第7期
·1018 · 南 京 医 科 大 学 学 报 2026年7月
集,结合迁移学习、领域自适应等技术,提升模型在 drug ⁃ naive depression[J]. Psychoneuroendocrinology,
不同中心和人群中的泛化能力 [38] 。其三,加强模型 2024,167:107086
可解释性和生物学合理性,结合可解释机器学习、 [5] BATAIL J M,COROUGE I,BLANCHARD T,et al. In⁃
网络神经科学和多组学分析,探索前额叶⁃边缘系 flammatory and MRI perfusion biomarkers in predicting
persistence of depression:a 6⁃month longitudinal study[J].
统、DMN 与炎症、代谢等通路之间的潜在联系,从相
Transl Psychiatry,2025,15(1):370
关性逐步走向机制假设的提出。其四,在临床转化层
[6] YIN B,CAI Y,TENG T,et al. Identifying plasma metabolic
面,开展前瞻性及嵌入式研究,评估多模态机器学习
characteristics of major depressive disorder,bipolar disor⁃
辅助系统对诊断一致性、治疗决策和复发预警的实际 der,and schizophrenia in adolescents[J]. Transl Psychia⁃
[38]
影响,同时关注患者体验、伦理规范和监管路径 。 try,2024,14(1):163
随着多模态数据资源的持续积累和建模方法 [7] ENKHBAYAR D,KO J,OH S,et al. Explainable artificial
的不断完善,有望逐步形成以神经影像为核心、整 intelligence models for predicting depression based on
合电生理、生物标志物、临床信息与数字行为的综 polysomnographic phenotypes[J]. Bioengineering,2025,
合评估框架,为抑郁症的早期识别、精准分层和个 12(2):186
体化干预提供更加坚实的证据支持 [39-40] 。 [8] HU Y X,SHI J Y,XIA G Y,et al. Analysis of functional
connectivity changes in attention networks and default
利益冲突声明:
mode networks in patients with depression and insomnia[J].
所有作者声明无利益冲突。
Sleep Breath,2024,28(4):1731-1742
Conflict of Interests:
[9] GRIMM B,YILMAM P,TALBOT B,et al. Multimodal ma⁃
All authors declare no conflict of interests.
chine learning for video based single question mental
作者贡献声明:
health assessment[J]. NPJ Digit Med,2025,9(1):65
高子祥、李翔、张欣负责论文撰写;钱昭均、嵇子慧负责
[10]CLAAS F,MICAH C,NILS O,et al. Systematic misesti⁃
材料整合;朱俐璇、徐忆初负责图表制作;陈亚丽、葛菲菲负
mation of machine learning performance in neuroimaging
责方向指导与修改。
studies of depression[J]. Neuropsychopharmacology,
Author’s Contributions:
2021,46(8):1510-1517
GAO Zixiang,LI Xiang,and ZHANG Xin drafted the manu⁃
[11]GUO Y,CHU T,LI Q,et al. Diagnosis of major depressive
script;QIAN Zhaojun and JI Zihui were responsible for material
disorder based on individualized brain functional and
integration;ZHU Lixuan and XU Yichu were responsible for
structural connectivity[J]. J Magn Reson Imaging,2025,
chart preparation;CHEN Yali and GE Feifei were responsible
61(4):1712-1725
for direction guidance and revisions.
[12]DAI P,HUANG K,HU T,et al. Altered effective connec⁃
[参考文献] tivity in patients with drug⁃naïve first⁃episode,recurrent,
[1] 王 颖,侯宇威,张 哲. 中国人群抑郁症疾病负担的 and medicated major depressive disorder:a multi⁃site fM⁃
年龄⁃时期⁃队列分析[J]. 四川精神卫生,2025,38(3): RI study[J]. Behav Brain Res,2025,495:115756
254-260 [13]GUO P,LIU X,SUN Y,et al. Effects of sMRI⁃guided
WANG Y,HOU Y W,ZHANG Z. Age⁃period⁃cohort anal⁃ rTMS on brain function and structure in major depressive
ysis of the disease burden of depression in the Chinese disorder[J]. J Affect Disord,2025,390:119772
population[J]. Sichuan Journal of Mental Health,2025, [14]VU T,DAWADI R,YAMAMOTO M,et al. Prediction of
38(3):254-260 depressive disorder using machine learning approaches:
[2] LIN S K,KUO P H,HSU C Y,et al. The effects of Lacto⁃ findings from the NHANES[J]. BMC medical informatics
bacillus plantarum PS128 in patients with major depres⁃ and decision making,2025,25(1):83
sive disorder:an eight⁃week double⁃blind,placebo⁃con⁃ [15]LUO Y,SHEN Y,FAN X. EEG microstates in adolescent
trolled study[J]. Asian J Psychiatr,2024,101:104210 depression:effects of depression severity and overall
[3] XU R F,LIU Z J,OUYANG S,et al. Machine learning⁃ symptoms[J]. J Affect Disord,2025,390:119819
driven development of a stratified CES⁃D screening sys⁃ [16]MULC D,VUKOJEVIC J,KALAFATIC E,et al. Opportu⁃
tem:optimizing depression assessment through adaptive nities and challenges for clinical practice in detecting
item selection[J]. BMC Psychiatry,2025,25(1):286 depression using EEG and machine learning[J]. Sensors,
[4] GAN X,LI X,CAI Y,et al. Metabolic features of adoles⁃ 2025,25(2):409-409
cent major depressive disorder:a comparative study be⁃ [17]YANG CY,CHEN Y Z. Support vector machine classifica⁃
tween treatment ⁃ resistant depression and first ⁃ episode tion of patients with depression based on resting ⁃ state

