ImportanceBiological psychiatry aims to understand mental disorders in terms of altered neurobiological pathways. However, for one of the most prevalent and disabling mental disorders, major depressive disorder (MDD), no informative biomarkers have been identified.ObjectiveTo evaluate whether machine learning (ML) can identify a multivariate biomarker for MDD.Design, Setting, and ParticipantsThis study used data from the Marburg-Münster Affective Disorders Cohort Study, a case-control clinical neuroimaging study. Patients with acute or lifetime MDD and healthy controls aged 18 to 65 years were recruited from primary care and the general population in Münster and Marburg, Germany, from September 11, 2014, to September 26, 2018. The Münster Neuroimaging Cohort (MNC) was used as an independent partial replication sample. Data were analyzed from April 2022 to June 2023.ExposurePatients with MDD and healthy controls.Main Outcome and MeasureDiagnostic classification accuracy was quantified on an individual level using an extensive ML-based multivariate approach across a comprehensive range of neuroimaging modalities, including structural and functional magnetic resonance imaging and diffusion tensor imaging as well as a polygenic risk score for depression.ResultsOf 1801 included participants, 1162 (64.5%) were female, and the mean (SD) age was 36.1 (13.1) years. There were a total of 856 patients with MDD (47.5%) and 945 healthy controls (52.5%). The MNC replication sample included 1198 individuals (362 with MDD [30.1%] and 836 healthy controls [69.9%]). Training and testing a total of 4 million ML models, mean (SD) accuracies for diagnostic classification ranged between 48.1% (3.6%) and 62.0% (4.8%). Integrating neuroimaging modalities and stratifying individuals based on age, sex, treatment, or remission status does not enhance model performance. Findings were replicated within study sites and also observed in structural magnetic resonance imaging within MNC. Under simulated conditions of perfect reliability, performance did not significantly improve. Analyzing model errors suggests that symptom severity could be a potential focus for identifying MDD subgroups.Conclusion and RelevanceDespite the improved predictive capability of multivariate compared with univariate neuroimaging markers, no informative individual-level MDD biomarker—even under extensive ML optimization in a large sample of diagnosed patients—could be identified.

中文翻译：

---

基于机器学习的重度抑郁症生物标志物的系统评估

重要性生物精神病学旨在从神经生物学途径改变的角度来理解精神障碍。然而，对于最普遍和最致残的精神障碍之一——重度抑郁症 (MDD)，尚未发现有信息的生物标志物。目的评估机器学习 (ML) 是否可以识别 MDD 的多变量生物标志物。设计、设置和参与者本研究使用来自马尔堡-明斯特情感障碍队列研究的数据，这是一项病例对照临床神经影像研究。2014 年 9 月 11 日至 2018 年 9 月 26 日，从德国明斯特和马尔堡的初级保健和普通人群中招募患有急性或终生 MDD 的患者以及年龄 18 至 65 岁的健康对照。明斯特神经影像队列 (MNC)用作独立的部分复制样本。数据分析时间为 2022 年 4 月至 2023 年 6 月。暴露患有 MDD 的患者和健康对照。主要结果和测量使用广泛的基于 ML 的多变量方法，跨全面的神经影像模式（包括结构和功能磁）在个体水平上量化诊断分类准确性。共振成像和弥散张量成像以及抑郁症的多基因风险评分。 结果 1801 名参与者中，1162 名 (64.5%) 为女性，平均 (SD) 年龄为 36.1 (13.1) 岁。共有 856 名 MDD 患者（47.5%）和 945 名健康对照者（52.5%）。MNC 复制样本包括 1198 名个体（362 名患有 MDD [30.1%] 和 836 名健康对照者 [69.9%]）。总共训练和测试了 400 万个 ML 模型，诊断分类的平均 (SD) 准确率在 48.1% (3.6%) 和 62.0% (4.8%) 之间。整合神经影像学模式并根据年龄、性别、治疗或缓解状态对个体进行分层并不能提高模型性能。研究结果在研究地点得到了重复，并且还在跨国公司的结构磁共振成像中观察到。在完美可靠性的模拟条件下，性能并没有显着提高。分析模型错误表明，症状严重程度可能是识别 MDD 亚组的潜在焦点。 结论和相关性 尽管与单变量神经影像标记物相比，多变量的预测能力有所提高，但没有提供信息的个体水平 MDD 生物标记物——即使在大样本的广泛 ML 优化下也是如此。确诊患者——可以被识别。