ARTIFICIAL INTELLIGENCE–BASED AUTONOMOUS SOCKET PROPOSAL PROGRAM: A PRELIMINARY STUDY FOR CLINICAL DECISION SUPPORT SYSTEM YAPAY ZEKA TABANLI OTONOM SOKET ÖNERME PROGRAMI: KLİNİK KARAR DESTEK SİSTEMİ İÇİN ÖN ÇALIŞMA


Creative Commons License

Çinar M. A., Haznedar B., Bayramlar K.

Turkish Journal of Physiotherapy and Rehabilitation, cilt.35, sa.2, ss.206-213, 2024 (ESCI) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 35 Sayı: 2
  • Basım Tarihi: 2024
  • Doi Numarası: 10.21653/tjpr.1421321
  • Dergi Adı: Turkish Journal of Physiotherapy and Rehabilitation
  • Derginin Tarandığı İndeksler: Emerging Sources Citation Index (ESCI), Scopus, TR DİZİN (ULAKBİM)
  • Sayfa Sayıları: ss.206-213
  • Anahtar Kelimeler: Amputation, Artificial Intelligence, Decision Support System
  • Erciyes Üniversitesi Adresli: Hayır

Özet

Purpose: The aim of this study is to develop artificial intelligence-based interfaces that can be used by professionals (clinicians and/or academics) working with disabled individuals who need prosthetics and to create a sample data set for professionals working in this field. Methods: 101 patients who had undergone amputation were enrolled. The residual limbs of all patients were scanned using a three-dimensional (3D) scanner and saved on the computer. The prosthetic sockets, fabricated using traditional methods, were also scanned with the same scanner and saved as a 3D model. Residual limb–prosthetic socket matches were obtained using data points and a deep neural network (DNN)-based decision support system was developed. Results: Simulation studies conducted with the point cloud data sets of 101 patients yielded a training success rate of 86%. The DNN model exhibited a generalization success rate of 78%. Conclusion: The artificial intelligence–based software interface has potential and could assist professionals by suggesting a suitable 3D socket model for patients in need of a prosthesis. Further studies will benefit from additional sample data to enhance the accuracy of the model.