Item reduction of the “Support Intensity Scale” for people with intellectual disabilities, using machine learning

Intensity
DOI: 10.1111/bld.12616 Publication Date: 2024-08-19T10:19:05Z
ABSTRACT
Abstract Background The study focuses on the need to optimise assessment scales for support needs in individuals with intellectual and developmental disabilities. Current are often lengthy redundant, leading exhaustion response burden. goal is use machine learning techniques, specifically item‐reduction methods selection algorithms, develop shorter more efficient scales. Methods A data set of 93 participants was analysed using Supports Needs Scale. Five feature‐selection algorithms were evaluated create a shortened questionnaire. For each algorithm, Random Forest model trained, performance assessed metrics like accuracy, precision, recall F1‐score measure how well predicted needs. Findings "Select from Model" algorithm successfully identified key items that could predict level Support model. Only 51 variables, out original 147, needed maintain predictive accuracy. reduced questionnaire maintained good reliability internal consistency compared instrument, strong F1 score indicating excellent performance. Conclusions demonstrates techniques effective reducing length questionnaires while preserving their psychometric properties. These can help institutions provide access information about without compromising validity or reliability, potentially better resource allocation improved care
SUPPLEMENTAL MATERIAL
Coming soon ....
REFERENCES (44)
CITATIONS (1)