[Prediction of activity and severity of endocrine ophthalmopathy by multidimensional linear regression modeling]

Endocrine ophthalmopathy (EOP) is a severe chronic autoimmune disease associated with autoimmune thyroid pathology that leads to loss of sight, cosmetic defects and quality of life decrease. EOP is difficult to give prognosis for due to various factors affecting its course and outcome.

To develop a reliable and precise prognosis method for EOP activity and severity based on personalized combination of risk factors in specific patients by multidimensional linear regression modeling.

A group of 139 patients (278 orbits) with newly diagnosed EOP associated with toxic diffuse goiter was observed during 1 year by an ophthalmologist and an endocrinologist; patient examination interval was 6 months. More than 250 indices were dynamically analyzed in the course of the study. Linear regression analysis was chosen as the research method; it allowed detection of linear dependencies between dependent and explanatory variables.

More than 600 various linear regression equations were derived that enabled prediction of EOP onset risk and development timeline, estimation of activity and/or severity of the disease, duration of active period in specific patients for the immediate and long-term outcome. From the derived models, the most reliable and safest for practical application were picked out. The present study introduces nine optimized models that can be used for patient follow-up since day one.

The onset risk, progression and outcome of the disease can be determined by a combination of factors revealed in the study.