# A novel CAT method for QoL screening: proof-of-principle study with comparisons to standard methods

## TL;DR

This study introduces a new computer adaptive testing method called LSCAT that improves accuracy in screening for depression symptoms compared to existing methods.

## Contribution

The novel LSCAT method combines latent-class and sum score approaches for more accurate HR-QoL screening.

## Key findings

- LSCAT outperformed SC and DTCAT in predictive accuracy with the lowest Type I error rates.
- LSCAT had Type II error rates as low as SC and significantly lower than DTCAT across all scenarios.
- LSCAT shows promise for developing efficient and valid HR-QoL screening tools.

## Abstract

This proof-of-principle study investigated a novel Computer Adaptive Testing (CAT) method termed Latent-class and Sum score based Computerized Adaptive Testing (LSCAT), developed for screening purposes. LSCAT was assessed for its ability to accurately predict depression symptoms during health-related quality of life (HR-QoL) screenings.

LSCAT’s performance was compared with two benchmark CAT methods, Stochastic Curtailment (SC) and Decision Tree based Computer Adaptive Testing (DTCAT), using data from the Patient Health Questionnaire-9 (PHQ-9).

LSCAT consistently outperformed both SC and DTCAT in terms of predictive accuracy, achieving the lowest rates of Type I error. Furthermore, LSCAT’s Type II error rates were at least as low as those of SC and significantly lower than those of DTCAT across all simulation scenarios.

These results suggest that LSCAT is a promising method for developing valid and efficient screening tools in HR-QoL research and practice.