Composite scores aim at quantifying the often multi-dimensional nature of health-care related outcomes in a single numeric value. While being used in almost all therapeutic areas, they are of particular importance for progressive neurologic diseases and mental disorders without reliable biomarkers, such as Alzheimer's disease, Parkinson's disease, multiple sclerosis and schizophrenia. In addition to the quantification of disease status, composite scores also arise through patient-reported outcomes (PROs) that aim at measuring outcomes important to patients including symptom status, physical function, mental health, social function, and wellbeing. Total score-based approaches dominate the analyses of composite scores. These methods essentially ignore the individual characteristics of an assessment and treat the assessment process as a black box. Item response (IR) models is an alternative analysis tool that considers the structure of each particular assessment and takes its properties into account. This project aims to develop methodology that allow expanded use of IR models for analysis of patient data in general and when composite scores are used repeatedly over time as is common in clinical trials and patient care, in particular. The plan is also to develop models for total score analysis that better agree with the underlying nature of the composite score data and that allow bridging between different types of analyses and data bases. Additionally, methodology will be developed for the robust and efficient building of these models and assessment of how they can improve drug development and patient care.