Development of Parkinson's Dementia Prediction Model Using Regression with Optimal Scale

Parkinson's disease dementia (PDD) is frequently found in patients with Parkinson’s disease (PD). Identifying early stage of Parkinson's disease dementia is an important primary goal of dementia treatment because it is possible to delay the progression of dementia when dementia is detected and treated in an early stage. However, the sociodemographic and neuropsychological characteristics of early stage PDD are less known than those of Alzheimer's dementia or vascular dementia. The objectives of this study were to develop a PDD prediction model considering environmental factors, health behaviors, physical functions, depression, medical history, and cognitive functions (eg. neuropsychological profile) by using the “Parkinson’s Dementia registry Data​​​​​​​,” a nationwide survey. This study analyzed 289 Parkinson patients (110 PDD patients and 179 PD-MCI), young than 65 years. The PDD prediction model was developed using a optimal scaling regression (It is also called categorical regression with optimal scaling). The results presented the optimization scale as an ordinal spline (graph). In addition, the quantification index (HAYASI I score) of each neuropsychological test for optimal prediction of PDD was presented. Based on this study, it will be needed to develop a customized screening test that can early detect PDD using medical big data.