Li Chong, Yang Yuqing, Wang Weijie, Li Huihuang, Mai Yiling, Zhao Jiubo
Department of Psychiatry, Zhujiang Hospital, Southern Medical University, No. 253, Industrial Avenue Zhong, Guangzhou, Guangdong, China.
Department of Psychology, School of Public Health, Southern Medical University, No. 1838 Guangzhou Dadao Road North, Guangzhou, Guangdong, China.
J Affect Disord. 2025 May 1;376:169-176. doi: 10.1016/j.jad.2025.02.006. Epub 2025 Feb 4.
Major depression disorder (MDD) is a common illness that severely limits psychosocial functioning and diminishes quality of life, particularly in young adults. Thus, it is imperial to identify MDD youth patients efficiently. This study aims to determine whether differential activation (DA) oriented recognizers can work efficiently.
This study collected heart rate variability (HRV) data and demographic information from 50 youth patients diagnosed with MDD and 53 healthy control participants. We developed six datasets, comprising baseline, stress, rest, differential activation period, Difference values between rest and stress period and combined dataset. From the provided data sets, we have developed machine learning models and also deep learning models. We then proceed to compare the performance metrics.
Models that utilized DA period and integration data sets exhibited superior performance compared to others. The deep learning model based on Long Short-Term Memory model we developed demonstrated the highest performance among all the models in each data set. Specifically, in the integration dataset, the model attained a mean cross-validation accuracy of 0.806 (95 % Confidential Interval (CI) 0.785-0.827), with a mean Area under Receiver Operating Characteristic Curve of 0.805 (95 % CI 0.784-0.826) and a mean Area under the Precision-Recall Curve of 0.863 (95 % CI 0.848-0.878).
The combination of DA theory and HRV record provides a new insight and also an efficient way for youth MDD identification.
重度抑郁症（MDD）是一种常见疾病，严重限制心理社会功能并降低生活质量，尤其是在年轻人中。因此，高效识别MDD青年患者至关重要。本研究旨在确定基于差异激活（DA）的识别器是否能有效工作。
本研究收集了50名被诊断为MDD的青年患者和53名健康对照参与者的心率变异性（HRV）数据及人口统计学信息。我们开发了六个数据集，包括基线、应激、静息、差异激活期、静息与应激期之间的差值以及组合数据集。从提供的数据集中，我们开发了机器学习模型和深度学习模型。然后我们继续比较性能指标。
利用差异激活期和整合数据集的模型表现优于其他模型。我们开发的基于长短期记忆模型的深度学习模型在每个数据集中的所有模型中表现最佳。具体而言，在整合数据集中，该模型的平均交叉验证准确率为0.806（95%置信区间（CI）0.785 - 0.827），平均受试者工作特征曲线下面积为0.805（95%CI 0.784 - 0.826），平均精确召回率曲线下面积为0.863（95%CI 0.848 - 0.878）。
差异激活理论与HRV记录的结合为青年MDD的识别提供了新的见解和有效方法。
