Objective
Stroke is a significant cause of disability, rendering patients with inability to perform activities of daily living due to lack of functional recovery. Precise prognosis in the early stage after stroke could enable realistic goal-setting and efficient resource allocation. Prediction algorithms have been tested and validated in the past, but they were using neurological biomarkers; thus, they were time-consuming, difficult to apply, expensive, and potentially harmful. The aim of this study was to create a new prediction algorithm that would not utilize any biomarkers.
Methods
A total of 127 stroke patients prospectively enrolled at day 3 after their stroke (mean age: 71, males n: 84, females n: 43). First, a sum of shoulder abduction and finger extension (SAFE) Medical Research Council (MRC) score was graded at day 3. Secondly, a binarized response was marked by the Mobilization and Simulation of Neuromuscular Tissue (MaSoNT) concept’s basic application on the upper limb. Third, the National Institutes of Health Stroke Scale (NIHSS) score was assessed. All data from the patients were included in a Classification and Regression Tree analysis to predict upper limb function 3 months post-stroke according to the Action Research Arm Test score at week 12.
Results
The Classification And Regression Tree (CART) analysis was performed that combines three different scores in order to predict upper-limb recovery: the SAFE score, MaSoNT’s application response, and the NIHSS. The overall correct prediction of the new algorithm is 69% which is lower than previous algorithms, though not significantly.
Conclusion
This study offers basic data to support the validity of the APRAHL algorithm. The new algorithm is faster and easier, but less accurate. Future studies are needed to create new algorithms that do not involve neurological biomarkers so that they will cost less and be easily applicable by health professionals.