Artificial intelligence based discovery of the association between depression and chronic fatigue syndrome.

BACKGROUND: Both of the modern medicine and the traditional Chinese medicine classify depressive disorder (DD) and chronic fatigue syndrome (CFS) to one type of disease. Unveiling the association between depressive and the fatigue diseases provides a great opportunity to bridge the modern medicine with the traditional Chinese medicine.

METHODS: In this work, 295 general participants were recruited to complete Zung Self-Rating Depression Scales and Chalder Fatigue Scales, and meanwhile, to donate plasma and urine samples for 1 H NMR-metabolic profiling. Artificial intelligence methods was used to analysis the underlying association between DD and CFS. Principal components analysis (PCA) and partial least squares discriminant analysis (PLS-DA) were used to analyze the metabolic profiles with respect to gender and age. Variable importance in projection and t-test were employed in conjunction with the PLS-DA models to identify the metabolite biomarkers. Considering the asymmetry and complexity of the data, convolutional neural networks (CNN) model, an artificial intelligence method, was built to analyze the data characteristics between each groups.

RESULTS: The results showed the gender- and age-related differences for the candidate biomarkers of the DD and the CFS diseases, and indicated the same and different biomarkers of the two diseases. PCA analysis for the data characteristics reflected that DD and CFS was separated completely in plasma metabolite. However, DD and CFS was merged into one group.

LIMITATION: Lack of transcriptomic analysis limits the understanding of the association of the DD and the CFS diseases on gene level.

CONCLUSION: The unmasked candidate biomarkers provide reliable evidence to explore the commonality and differences of the depressive and the fatigue diseases, and thereby, bridge over the traditional Chinese medicine with the modern medicine.