Development and validation of a novel model based on hand knob score and white matter injury on MRI to predict hand function in children with cerebral palsy

Jing-Jing Zhang; Yan-Li Yang; Jie Hu; Chun-Feng Zhao; Xing-Hong He; Qian-Yu Yang; Xiao-Shan Qi; Hong Lu; Cheng He; Heng Liu

doi:10.21037/atm-22-4112

Development and validation of a novel model based on hand knob score and white matter injury on MRI to predict hand function in children with cerebral palsy

Ann Transl Med. 2022 Oct;10(20):1102. doi: 10.21037/atm-22-4112.

Authors

Jing-Jing Zhang^#^{1

2}, Yan-Li Yang^#¹, Jie Hu¹, Chun-Feng Zhao¹, Xing-Hong He¹, Qian-Yu Yang¹, Xiao-Shan Qi¹, Hong Lu³, Cheng He⁴, Heng Liu¹

Affiliations

¹ Department of Radiology, The Affiliated Hospital of Zunyi Medical University, Zunyi, China.
² Department of Radiology, Mianyang Hospital of T.C.M, Mianyang, China.
³ Department of Radiology, The Seventh People's Hospital of Chongqing, The Central Hospital Affiliated to Chongqing University of Technology, Chongqing, China.
⁴ Medical Imaging Department, Chongqing University Central Hospital, Chongqing, China.

^# Contributed equally.

Abstract

Background: Childhood hand function is considered to be one of the strongest predictors of the ability to participate in daily activities as children with cerebral palsy (CP) reach adulthood. The manual ability classification system (MACS) is currently the most widely used for grading hand function in children with CP. However, the MACS method is subjective and may be affected by the raters' experience. Hand knob is an important control center for hand movement. Therefor this study aimed to develop and validate an objective model for hand function estimation in children with CP and visualize it as a nomogram.

Methods: A total of 70 Children (2-12 years old) with CP underwent magnetic resonance imaging (MRI) scanning, MACS assessment. According to MACS, children with CP were divided into mild impairment group (grade I-III) and severe impairment group (grade IV-V). Hand function prediction models based on (I) hand knob score, (II) clinical features, and (III) the combination of clinical features and hand knob score were developed and validated separately. The models were subjected to stepwise regression according to the maximum likelihood method, and the Akaike information criterion was used to select the best model. Model discrimination was assessed using receiver operating characteristic (ROC) and calibration curves. The nomogram was finally built according to the best model.

Results: The area under the curve (AUC) of the hand knob score model in the training set was 0.752, the clinical features model was 0.819, and the hand knob score and clinical features combined model was 0.880. The AUC of the hand knob score model in the validation set was 0.765, the clinical features model was 0.782, and the combined model was 0.894. The best model was the hand knob score-clinical features combined model, and the nomogram finally incorporated two assessment items: the hand knob score and white matter injury. The estimated probability of hand function injury degree of the combined model displayed good agreement with the actual occurrence probability.

Conclusions: The hand knob score-clinical features combined model can be used to preliminarily assess the degree of hand impairment in children with CP, with good calibration.

Keywords: Cerebral palsy (CP); hand function; hand knob; magnetic resonance imaging (MRI); predictive model.