Cerebral palsy (CP) is the most common cause of physical disability in childhood. Children with CP has a life-long needs for special services and CP is therefore a huge burden for the child and it?s family. This project will introduce a new ground-breaking artificial intelligence system, called the DeepInMotion system, for early detection of CP. DeepInMotion will identify characteristics in the infant spontaneous movements when their are between 12-18 weeks which are indicator for later development of CP. In Norwegian hospitals, CP are detected when the child is 18 months old even though the brain injury causing CP occurs at birth. Thus, early detection of CP before 5 months of age will provide opportunities for earlier onset of therapies and treatments in the period when plasticity of the infant brain is at its highest. Today, early detection of CP is performed in clinics by a subjective and a qualitative movement analysis. However, these early detection methods need highly qualified clinicians with long experience to be reliable and, thus, lack of widespread adoption among clinical teams. By developing a easy-to-use smartphone-based system, this project will provide a low-cost health care service reducing inequalities within and among countries and providing equal rights for health care services. DeepInMotion will improve help and support to clinicians and primary care givers in a safe and efficient manner and, thus, meeting the requirements of UN sustainable developmental goals. The DeepInMotion project will utilize one of the largest international database of videos of high-risk infants, administrated by St Olavs Hospital in Norway, to develop the next generation of artificial intelligent systems for infant care. The DeepInMotion project consist of a international unique interdisciplinary group of computer scientists, human movement scientists, and clinical specialists.
Cerebral palsy (CP) is the most common cause of physical disability in childhood which result in life-long needs for special services and a huge burden for the child and it’s family. The project aims to introduce a new ground-breaking concept of explainable AI techniques (XAI), the DeepInMotion system, for early detection of CP by identifying clinical explainable biomarkers in the infant motor repertoire. The DeepInMotion system will obtain these goals by solving the fundamental challenge in clinical movement analysis: Identification of diagnosis-specific motor phenotype of the infant spontaneous movements related to the CP outcome later in childhood. The DeepInMotion utilize new technology within explainable machine learning and deep learning identifying clinical explainable biomarkers without relying on preselected measures or subjective clinical evaluations (WP1). The DeepInMotion will have a user-centered design ensuring a prototype of clinical service implementation within the life-time of the project with technical and performance documentation ready for medical device approval (WP 2 and WP3). The project is also a start-up of the new interdisciplinary DeepMotion project group in collaboration with internationally well recognized research partners on pediatrics, clinical movement analysis, and artificial intelligence as well as medical device manufacturers in the project advisory board. The project utilizes large national and international register data providing the big data-enabled platform for the machine learning models and XAI techniques. The DeepInMotion project will include two 3-year PhD fellowship, and has a total budget of 15 MNOK where 12 MNOK are applied from the Norwegian Research Council.