Design of a mobile application based on artificial intelligence to identify pain in non-communicating individuals with cerebral palsy

Abstract

[eng] Introduction: Pain assessment in individuals with cerebral palsy (CP), particularly those unable to self-report, is a significant challenge. Pain is the most common comorbidity in CP, yet current evaluation methods are often subjective and unreliable. An AI-based facial recognition system integrated into a mobile application could provide an objective, reliable tool for pain assessment in this population. Objectives: Evaluate the feasibility of a mobile app using Artificial Intelligence for pain identification in non-communicative individuals with CP, developed by the research team.Identify key application requirements, incorporating health app design recommendations, accessibility criteria, and universal design principles.Define users’ needs and functional demands to ensure effectiveness and usability. Methods: Three approaches were employed:Literature review on automated facial recognition systems in CP.Software development planning by a multidisciplinary team of physiotherapists and computer scientists.Consultation with families, caregivers, legal guardians, healthcare professionals, and rehabilitation center managers to define functional demands, followed by collaboration with a software development company. Results: A systematic review identified seven studies on automated facial recognition systems for pain detection. However, only one of these systems—ePAT/PainCheck—has been developed into a functional mobile application for clinical use, though not specific to individuals with cerebral palsy. This underscores the novelty of the current initiative. The feasibility of our proposed app was confirmed, and key technical and functional requirements were outlined, including intuitive design, dual local/cloud processing, and mechanisms for system improvement. Stakeholders emphasized ease of use, and suggested incorporating features such as accuracy estimation, offline functionality, multi-language support, and open communication fields. Conclusions: This novel and feasible app represents a significant advance in pain assessment for CP, with potential applications in other neurological conditions with communication impairments and unique facial expressions.