Predictors of Depression Among Chinese Older Adults with Neurodegenerative Diseases: Evidence from Decision Tree Approach

Background: Neurodegenerative diseases (NDs), such as Parkinson’s disease (PD), Alzheimer’s disease and related dementias (AD/RD), have become a growing issue among older adults in China. By 2030, more than 20 million and 4.94 million Chinese older adults ages 65 and older will suffer from AD/RD and PD respectively, making China the largest and fastest-growing NDs patient population in the world. Depression is a major health issue among older adults and negatively impacts them physically and mentally. However, to date, little is known about the risk factors for depression among older Chinese adults who have NDs. Artificial Intelligence (AI) gains popularity exceedingly in the field of health care. AI approach, like decision tree which is a tree-shaped flowchart, is becoming one of the most well-known and efficient techniques used in prediction and analysis of physical and mental diseases. However, no prior studies have used AI for better understanding the association between NDs and depression among older adults. The purpose of this study is to develop a depression prediction model to analyze the relationship between depression and NDs, and from which to explore possible factors associated with depression among older Chinese adults with NDs.

Methods: Based on the stress-coping model, data from the latest wave of the Chinese Longitudinal Healthy Longevity Survey (CLHLS, 1998-2014) in 2014 were analyzed. Among the sample, 334 older adults aged 65 and older with NDs were included in the analysis. The data were randomly split into training dataset (70% of the sample, n= 234 cases) and test (30% of the sample, n= 100 cases) dataset. Three measures were used to evaluate the effectiveness of the decision tree model by comparing with the test dataset. They are accuracy, sensitivity, and specificity. Goodness index (Range: 0-) was assessed to measure the degree of performance of decision tree.

Results: Result showed that from the top node (root node) to the bottom node (leaf node), a total of 7 nodes built the decision tree which predicts whether Chinese older adults with NDs suffer from depression, including IADLs, MMSE, Health Status, ADLs, Gender, Self-rated Health Change, and Age. Through the comparison of accuracy (Training: 0.928 vs. Testing: 0.87), sensitivity (Training: 0.50 vs. Testing: 0.133), and specificity (Training: 0.99 vs. Testing: 0.976), and the Goodness Index of the decision tree is 0.38 (optimal if G ≤ 0. 25; good if 0. 25 < G < 0.70; random if G = 0. 70: bad if G > 0. 70), all proved that the decision tree model was good in prediction.

Implications: Depression can be a consequence of NDs, not just being a prodrome or risk factor. Subjective assessments like self-reported health status and self-reported health change under Chinese traditional culture, and the condition of physical function have significant impact on the occurrence of depression. Improving the quality of medical services, popularizing the knowledge of NDs and depression, and enhancing the education of caregiving to facilitate caregivers to provide care that fits older Chinese adults can greatly weaken the impact of depression.