Abstract：
OBJECTIVE: A plethora of studies have unequivocally established the profound significance of harmonious familial relationships on the psychological well-being of the elderly. In this study, we elucidate the intergenerational relationships, probing the association between frequent interactions or encounters with their children and the incidence of depression in old age.
METHODS: We employed a retrospective cross-sectional study design, sourcing our data from the 2018 wave of the China Health and Retirement Longitudinal Study (CHARLS). To identify cases of depression, we utilized the 10-item Center for Epidemiologic Studies Depression Scale (CESD). Employing a five-fold cross-validation methodology, we endeavored to fashion five distinct machine learning models. Subsequently, we crafted learning curves to facilitate the refinement of hyperparameters, assessing model classification performance through metrics such as accuracy and the Area Under the Receiver Operating Characteristic (AUROC) curve. To further elucidate the relationship between variables and geriatric depression, logistic regression was subsequently applied.
RESULTS: Our findings accentuated that sleep patterns emerged as the paramount determinants influencing the onset of depression in the elderly. Relationships with offspring ranked as the second most significant determinant, only surpassed by sleep habits. A negative correlation was observed between sleep patterns (Odds Ratio [OR]: 0.78, 95 % Confidence Interval [CI]: 0.75-0.81, P < 0.01), communication with offspring (OR: 0.86, 95 % CI: 0.82-0.90, P < 0.01), and the prevalence of depressive symptoms. Among the evaluated models, the k-Near Neighbor algorithm demonstrated commendable discriminative power. However, it was the Random Forest algorithm that manifested unparalleled discriminative prowess and precision, establishing itself as the most efficacious classifier.
CONCLUSIONS: Prolonging the duration of nocturnal sleep, and elevating the frequency of communication with offspring have been identified as measures conducive to mitigating the onset of geriatric depression.
METHODS: We employed a retrospective cross-sectional study design, sourcing our data from the 2018 wave of the China Health and Retirement Longitudinal Study (CHARLS). To identify cases of depression, we utilized the 10-item Center for Epidemiologic Studies Depression Scale (CESD). Employing a five-fold cross-validation methodology, we endeavored to fashion five distinct machine learning models. Subsequently, we crafted learning curves to facilitate the refinement of hyperparameters, assessing model classification performance through metrics such as accuracy and the Area Under the Receiver Operating Characteristic (AUROC) curve. To further elucidate the relationship between variables and geriatric depression, logistic regression was subsequently applied.
RESULTS: Our findings accentuated that sleep patterns emerged as the paramount determinants influencing the onset of depression in the elderly. Relationships with offspring ranked as the second most significant determinant, only surpassed by sleep habits. A negative correlation was observed between sleep patterns (Odds Ratio [OR]: 0.78, 95 % Confidence Interval [CI]: 0.75-0.81, P < 0.01), communication with offspring (OR: 0.86, 95 % CI: 0.82-0.90, P < 0.01), and the prevalence of depressive symptoms. Among the evaluated models, the k-Near Neighbor algorithm demonstrated commendable discriminative power. However, it was the Random Forest algorithm that manifested unparalleled discriminative prowess and precision, establishing itself as the most efficacious classifier.
CONCLUSIONS: Prolonging the duration of nocturnal sleep, and elevating the frequency of communication with offspring have been identified as measures conducive to mitigating the onset of geriatric depression.
摘要：
目的：大量研究已经明确确立了和谐家庭关系对老年人心理健康的深远意义。在这项研究中,我们阐明了代际关系，探讨与孩子频繁互动或相遇与老年抑郁症发病率之间的关系。
方法：我们采用回顾性横断面研究设计，我们的数据来自2018年中国健康与退休纵向研究浪潮(CHARLS)。为了识别抑郁症病例，我们使用了10项流行病学研究中心抑郁量表(CESD).采用五重交叉验证方法，我们努力打造五种不同的机器学习模型。随后，我们制作了学习曲线来促进超参数的细化，通过准确度和接收器工作特性下面积(AUROC)曲线等指标评估模型分类性能。为了进一步阐明变量与老年抑郁症之间的关系，随后应用逻辑回归。
结果：我们的研究结果强调，睡眠模式成为影响老年人抑郁症发作的最重要决定因素。与后代的关系被列为第二重要的决定因素，只有睡眠习惯才能超越。睡眠模式之间呈负相关（赔率比[OR]：0.78，95％置信区间[CI]：0.75-0.81，P<0.01），与后代的沟通（OR：0.86，95%CI：0.82-0.90，P<0.01），和抑郁症状的患病率。在评估的模型中，k-NearNeighbor算法表现出了值得称赞的判别能力。然而,它是随机森林算法，表现出无与伦比的辨别能力和精度，将自己确立为最有效的分类器。
结论：延长夜间睡眠时间，提高与后代的交流频率已被确定为有助于减轻老年抑郁症发作的措施。
方法：我们采用回顾性横断面研究设计，我们的数据来自2018年中国健康与退休纵向研究浪潮(CHARLS)。为了识别抑郁症病例，我们使用了10项流行病学研究中心抑郁量表(CESD).采用五重交叉验证方法，我们努力打造五种不同的机器学习模型。随后，我们制作了学习曲线来促进超参数的细化，通过准确度和接收器工作特性下面积(AUROC)曲线等指标评估模型分类性能。为了进一步阐明变量与老年抑郁症之间的关系，随后应用逻辑回归。
结果：我们的研究结果强调，睡眠模式成为影响老年人抑郁症发作的最重要决定因素。与后代的关系被列为第二重要的决定因素，只有睡眠习惯才能超越。睡眠模式之间呈负相关（赔率比[OR]：0.78，95％置信区间[CI]：0.75-0.81，P<0.01），与后代的沟通（OR：0.86，95%CI：0.82-0.90，P<0.01），和抑郁症状的患病率。在评估的模型中，k-NearNeighbor算法表现出了值得称赞的判别能力。然而,它是随机森林算法，表现出无与伦比的辨别能力和精度，将自己确立为最有效的分类器。
结论：延长夜间睡眠时间，提高与后代的交流频率已被确定为有助于减轻老年抑郁症发作的措施。
