TITLE:
Cubic Spline Regression: An Application to Early Bipolar Disorder Dynamics
AUTHORS:
Petronilla Uchenna Ogoke, Chinaka Ethelbert Nduka, Ajibola Taiwo Soyinka
KEYWORDS:
Bipolar, Gold Standard, Multinomial Model, Response Bias, Risk Tendency
JOURNAL NAME:
Open Journal of Statistics,
Vol.6 No.6,
November
14,
2016
ABSTRACT: Owing to the fact that the major challenge
of predicting the risk of having bipolar is the absence of a gold standard to
distinguish between true cases and false positive; this study employed the
extension of cubic spline function to the multinomial model to explore the risk
tendency of unnoticed early bipolar across three different groups of mood
disorder. The intermediate group was used to accommodate for false negative and
false positive while mapping the true value of bipolar risk tendency across the
three groups to a scale. Hence for all distributions of “yes” ticked in a mood
disorder questionnaire, the study predicts the bipolar risk tendency while
simultaneously accommodating for the patients response bias. The coefficients
of the polynomial are obtained using the maximum likelihood method. The spline
graph reveals how bipolar disorder build up slowly and lingers in the body for
long without been noticed due to fluctuations in risk tendency of the mood
scores.