Search

Purpose: To assess the psychometric properties of glaucoma-specific health-related quality of life (HRQoL) item banks (IBs), and explore their efficiency using computerized adaptive testing (CAT) simulations. Methods: In this cross-sectional, clinical study, 300 Asian glaucoma patients answered 221 items within seven IBs: Ocular Comfort Symptoms (OS); Activity Limitation (AL); Lighting (LT); Mobility (MB); Glaucoma Management (GM); Psychosocial (PSY); and Work (WK). Rasch analysis was conducted to assess each IB's psychometric properties (e.g., item "fit" to the construct; unidimensionality) and a set of analytic performance criteria guiding decision making relating to retaining or dropping domains and items was employed. CAT simulations determined the mean number of items for 'high' and 'moderate' measurement precision (stopping rule: SEM 0.3 and 0.387, respectively). Results: Participants' mean age was 67.2 ± 9.2 years (62% male; 87% Chinese). LT, MB, and GM displayed good psychometric properties overall. To optimize AL's psychometric properties, 16 items were deleted due to poor "fit", high missing data, item bias, low discrimination and/or a low clinical/patient importance rating. To resolve multidimensionality in PSY, we rehomed 16 items into a "Concern (CN)" domain. PSY and CN required further amendment, including collapsing of response categories, and removal of poorly functioning items (N = 7). Due to poor measurement precision, low applicability and high ceiling effect, low test information indices, and low item separation index the WK IB was not considered further. In CAT simulations on the final seven IBs (n = 182 items total), an average of 12.1 and 15.7 items per IB were required for moderate and high precision measurement, respectively. Conclusions: After reengineering our seven IBs, they displayed robust psychometric properties and good efficiency in CAT simulations. Once finalized, GlauCAT™-Asian may enable comprehensive assessment of the HRQoL impact of glaucoma and associated treatments. [ABSTRACT FROM AUTHOR]

Success in all sorts of situations is the most classical interpretation of general intelligence. Under limited resources, however, the capability of an agent must necessarily be limited too, and generality needs to be understood as comprehensive performance up to a level of difficulty. The degree of generality then refers to the way an agent's capability is distributed as a function of task difficulty. This dissects the notion of general intelligence into two non-populational measures, generality and capability, which we apply to individuals and groups of humans, other animals and AI systems, on several cognitive and perceptual tests. Our results indicate that generality and capability can decouple at the individual level: very specialised agents can show high capability and vice versa. The metrics also decouple at the population level, and we rarely see diminishing returns in generality for those groups of high capability. We relate the individual measure of generality to traditional notions of general intelligence and cognitive efficiency in humans, collectives, non-human animals and machines. The choice of the difficulty function now plays a prominent role in this new conception of generality, which brings a quantitative tool for shedding light on long-standing questions about the evolution of general intelligence and the evaluation of progress in Artificial General Intelligence. [ABSTRACT FROM AUTHOR]

Purpose: To compare the results from a simulated computerized adaptive test (CAT) for the 28-item Impact of Vision Impairment (IVI) questionnaire and the original paper-pencil version in terms of efficiency (main outcome), defined as percentage item reduction.Methods: Using paper-pencil IVI data from 832 participants across the spectrum of vision impairment, item calibrations of the 28-item IVI instrument and its associated 20-item vision-specific functioning (VSF) and 8-item emotional well-being (EWB) subscales were generated with Rasch analysis. Based on these calibrations, CAT simulations were conducted on 1000 cases, with 'high' and 'moderate' precision stopping rules (standard error of measurement [SEM] 0.387 and 0.521, respectively). We examined the average number of items needed to satisfy the stopping rules and the corresponding percentage item reduction, level of agreement between person measures estimated from the full IVI item bank and from the CAT simulations, and item exposure rates (IER).Results: For the overall IVI-CAT, 5 or 9.7 items were required, on average, to obtain moderate or high precision estimates of vision-related quality of life, corresponding to 82.1 and 65.4% item reductions compared to the paper-pencil IVI. Agreement was high between the person measures generated from the full IVI item bank and the IVI-CAT for both the high precision simulation (mean bias, - 0.004 logits; 95% LOA - 0.594 to 0.587) and moderate precision simulation (mean bias, 0.014 logits; 95% LOA - 0.828 to 0.855). The IER for the IVI-CAT in the moderate precision simulation was skewed, with six EWB items used > 40% of the time.Conclusion: Compared to the paper-pencil IVI instrument, the IVI-CATs required fewer items without loss of measurement precision, making them potentially attractive outcome instruments for implementation into clinical trials, healthcare, and research. Final versions of the IVI-CATs are available. [ABSTRACT FROM AUTHOR]