Selecting Learning Algorithms for Simultaneous Identification of Depression and Comorbid Disorders

Abstract

Depression is a serious worldwide public health problem, and its diagnosis still remains a challenge in the medical community. The difficulties of detecting depression are largely due to its high comorbid factor. Given the reciprocal relationship between depression and physical illness, mental health professionals have called for a diagnostic approach that identifies and evaluates each disorder, concurrently. This paper reports the findings of a study based on data collected in Nigeria to investigate the simultaneous identification of depression and co-occuring physical illness using a multi-dimensional Bayesian network classification approach. The predictive model would be useful to clinicians of all categories in Nigeria in overcoming the challenges of depression diagnosis caused by its frequent co-occurence with physical illnesses. The benefits of this approach are demonstrated with anonymised multi-dimensional depression dataset comprising 1090 instances, 22 symptoms, and two class attributes. The results, are also described.

Ähnliche Arbeiten