1. An Item Response Theory Model for Functional Assessment of Spinal Muscular Atrophy Based on the Hammersmith Functional Motor Scale Expanded
- Author
-
FAN Yabei, CHEN Jinjin, GE Panli, CHENG Xi, ZHANG Yichen, and LIN Feng
- Subjects
spinal muscular atrophy ,Hammersmith functional motor scale expanded ,item response theory ,functional analysis ,Medicine - Abstract
ObjectiveTo establish an item response theory (IRT) model for functional assessment of spinal muscular atrophy based on the Hammersmith functional motor scale expanded.MethodsA total of 23 patients with SMA were treated with corrective treatment in the Department of Geriatric Neurology and the Department of Pediatrics of the First Affiliated Hospital with Nanjing Medical University from February 2021 to June 2023. IRT was constructed using these repeated measurement data based on stack design. The comparisons between well-fitted models were based on likelihood ratio. And item guessing, carelessness, difficultyand personal abilities (θ) of selected optimal model were calculated. Differential test functioning (DTF) test was conducted for gender and rehabilitation therapy to test the bias. Monte Carlo simulation was used to prove that the item difficulty distribution of the model was not affected by the sampling process. And person-item plot was drawn according to the median expected difficulty. and the differences in the difficulty of each item in HFMSE scale and personal abilities among each evaluations were discussed.ResultsA total of 32items were included in this study and 4 alternative models were constructed. Three-parametric logistic model was selected as the optimal model with the best fitness (P>0.05, M2 statistics = 479.75). In 3PLM, the caution of all items was 1, and the discrimination, guessing and difficulty of the items were calculated. There were no DTFs for gender and rehabilitation therapy. The results showed high reliability (Cronbach's α = 0.955, LCRC = 0.981) and validity [θ value is positively correlated with the original total score (P
- Published
- 2024
- Full Text
- View/download PDF