Page 73 - 南京医科大学自然版

第46卷第7期 高子祥，李 翔，张 欣，等. 从数据整合到临床转化：机器学习在抑郁症诊断中的应用［J］.
2026年7月 南京医科大学学报（自然科学版），2026，46（7）：1011-1019 ·1019 ·

electroencephalography［J］. Asian Biomed（Res Rev signatures in major depression［J］. Mol Psychiatry，2025，
News），2024，18（5）：212-223 30（9）：3963-3974
［18］DENIER N，WALTHER S，BREIT S，et al. Electroconvul⁃ ［30］HAN W，WU X，WANG L，et al. Altered brain function in
sive therapy induces remodeling of hippocampal co⁃activa⁃ treatment ⁃ resistant depression patients：a resting ⁃ state
tion with the default mode network in patients with depres⁃ functional magnetic resonance imaging study［J］. Neuro⁃
sion［J］. Neuroimage Clin，2023，38：103404 sci Lett，2024，842：138004
［19］ALOTAIBI N M，BAKHEET D M. Predicting depression ［31］LEE D Y，KIM N，PARK C，et al. Explainable multimodal
severity using effective and functional brain connectivity prediction of treatment⁃resistance in patients with depres⁃
of the electroencephalography signals［J］.Comput Biol sion leveraging brain morphometry and natural language
Med，2025，190：110045 processing［J］. Psychiatry Res，2024，334：115817
［20］RAVAN M，NOROOZI A，GEDIYA H，et al. Using deep ［32］QAYYUM A，RAZZAK I，TANVEER M，et al. High⁃den⁃
learning and pretreatment EEG to predict response to ser⁃ sity electroencephalography and speech signal based
traline，bupropion，and placebo［J］. Clin Neurophysiol， deep framework for clinical depression diagnosis［J］.
2024，167：198⁃208 IEEE/ACM Trans Comput Biol Bioinform，2023，20（4）：
［21］BAUER J F，GERCZUK M，SCHINDLER⁃GMELCH L，et 2587-2597
al. Validation of machine learning ⁃ based assessment of ［33］WEBER J，WEBER M，LOPEZ A. Depression diagnosis
major depressive disorder from paralinguistic speech char⁃ from patient interviews using multimodal machine learn⁃
acteristics in routine care［J］. Depress Anxiety，2024， ing［J］. Front Psychiatry，2025，16：1694762
2024（1）：9667377 ［34］KIM J，MA S P，CHEN M L，et al. Optimizing large lan⁃
［22］IKÄHEIMONEN A，LUONG N，BARYSHNIKOV I，et al. guage models for detecting symptoms of depression/anxi⁃
Predicting and monitoring symptoms in patients diag⁃ ety in chronic diseases patient communications［J］. NPJ
nosed with depression using smartphone data：observation⁃ Digit Med，2025，8（1）：580
al study［J］. J Med Internet Res，2024，26：e56874 ［35］YANG W，WANG X，KANG C，et al. Establishment of a
［23］EDER J，PFEIFFER L，WICHERT P S，et al. Deconstruct⁃ risk prediction model for suicide attempts in first⁃episode
ing depression by machine learning：the POKAL ⁃ PSY and drug naïve patients with major depressive disorder［J］.
study［J］. Eur Arch Psychiatry Clin Neurosci，2023，274 Asian J Psychiatr，2023，88：103732
（5）：1153-1165 ［36］SCHNELLBÄCHER G J，RAJKUMAR R，VESELINOVI
［24］ALOTAIBI N M，ALHOTHALI A M，ALI M S. Multi⁃atlas C T，et al. Structural alterations as a predictor of depres⁃
ensemble graph neural network model for major depres⁃ sion⁃a7⁃Tesla MRI⁃based multidimensional approach［J］.
sive disorder detection using functional MRI data［J］. Mol Psychiatry，2025，30（6）：2517-2524
Front Comput Neurosci，2025，19：1537284 ［37］COLOMBO F，CALESELLA F，BRAVI B，et al. Multimod⁃
［25］SAMPRIT B，YIYUAN W，S K B，et al. Trajectories of re⁃ al brain⁃derived subtypes of major depressive disorder dif⁃
mitted psychotic depression：identification of predictors of ferentiate patients for anergic symptoms，immune⁃inflam⁃
worsening by machine learning［J］. Psychol med，2024，54 matory markers，history of childhood trauma and treatment⁃
（6）：1142-1151 resistance［J］. Eur Neuropsychopharmacol，2024，85：45-57
［26］MARDINI M T，KHALIL G E，BAI C，et al. Identifying ［38］LIU R，HOU X，LIU S，et al. Predicting antidepressant re⁃
adolescent depression and anxiety through real⁃world data sponse via local⁃global graph neural network and neuroim⁃
and social determinants of health：machine learning model aging biomarkers［J］. NPJ Digit Med，2025，8（1）：515
development and validation［J］. JMIR Ment Health，2025， ［39］MAZURKA R，CUNNINGHAM S，HASSEL S，et al. Rela⁃
12：e66665 tion of hippocampal volume and SGK1 gene expression to
［27］MA S，NIE Z，ZHANG M，et al. Towards precision psychi⁃ treatment remission in major depression is moderated by
atry：metabolomics identifies three biological subtypes of childhood maltreatment：a CAN⁃BIND⁃1 report［J］. Eur⁃
depression［J］. PLoS Digit Health，2025，4（12）： Neuropsychopharmacol，2024，78：71-80
e0001125 ［40］WANG M，WANG H，FENG Z，et al. Predicting depres⁃
［28］TOZZI L，ZHANG X，PINES A，et al. Personalized brain sion among chinese patients with narcolepsy type 1：a ma⁃
circuit scores identify clinically distinct biotypes in de⁃ chine ⁃ learning approach［J］. Nat Sci Sleep，2024，16：
pression and anxiety［J］. Nat Med，2024，30（7）：2076-2087 1419-1429
［29］JIAO Y，ZHAO K，WEI X，et al. Deep graph learning of （收稿：2026-03-19；修回：2026-04-28；录用：2026-04-30）
multimodal brain networks defines treatment ⁃ predictive （本文编辑：唐 震）