Mobile App for Assessing Hemifacial Spasm Treatment Response Using Machine Learning

EasyChair Preprint 9577

Abstract

It is challenging to assess hemifacial spasm (HFS) patients as they exhibit high-frequency and heterogeneous anomalous eyelid movements. This study aimed to develop an application for a smartphone to objectively determine the eyelid movements frequency so that treatment responses in these pa- tients can be assessed accurately. The smartphone application was developed mainly using Python, a prominent and broadly used programming language focused on machine learning and data science tasks. The application can precisely predict the movement of the patient's eyes using an SVM regressor and classifier. The results are plotted for better visual inspection by using data visualization techniques. Thus, the application ena- bles a continuous study of each patient using an integrated da- tabase in Google spreadsheets, which could better track the re- sults of each treatment response. The application showed to be an efficient method to identify and represent eyelid movement occurrences in patients, objectively measuring the eyelid move- ment frequency and, thus, assessing the treatment response in patients with hemifacial spasm. This system could enable cus- tomized and fine adjustments to botulinum toxin doses based on each patient's needs.

Keyphrases: Application, Data Science, Eye Aspect Ratio, Hemifacial Spasm, machine learning

@booklet{EasyChair:9577,
  author    = {João Luiz Sotini da Silva and Carlos Marcelo Gurjão de Godoy and Tammy Hentona Osaki and Midori Hentona Osaki and Cristina Yabumoto and Regina Célia Coelho},
  title     = {Mobile App for Assessing Hemifacial Spasm Treatment Response Using Machine Learning},
  howpublished = {EasyChair Preprint 9577},
  year      = {EasyChair, 2023}}