ICF:'Transforming diagnosis of Developmental Dysplasia of the Hip:Validation and Implementation modules of an AI supported handheld ultrasound scanner

In Developmental Dysplasia of the Hip (DDH), the ball-and-socket hip joint fails to develop normally in babies and young children. DDH is the most common Paediatric Orthopaedic condition in newborns, affecting 2% of all babies born. If DDH is detected early, while hips are still soft cartilage, it can be cured in most infants by wearing a simple harness for 6-12 weeks. The best way to detect DDH in babies is by ultrasound.

Unfortunately, due to lack of access to well-trained staff and financial constraints, the UK only screens babies with risk factors or whose hips seem abnormal on physical examination. This selective screening strategy misses an unknown, but likely large, number of DDH cases.

The Newborn hip examination at birth and the 6-8 week baby health checks include the Barlow and Ortolani manoeuvres which are known to be flawed and not sensitive enough to pick up the subtleties of hip dysplasia. Reidy et al. showed in 2019 that only 20% of family doctors were able to identify DDH from a physical examination, while Harper et al (2020) showed that "even experts can be fooled" [1,2].

These limitations of the hip physical examinations can result in misdiagnosis and late discovery of DDH cases that are, by that time, complex and difficult to treat. There is also a well described correlation between hip dysplasia and hip arthritis.

Across the UK, despite national screening of at-risk babies, current rates of DDH diagnosis and late detection remain similar to 35-years ago. Introduction of a universal ultrasound screening programme that tests all babies has been viewed as unaffordable and impractical. This highlights the need for innovative solutions.

In this study, we aim to develop validation and implementation modules for an artificial intelligence (AI)-based ultrasound screening tool (Hip AI) within an NHS ecosystem.

The Hip AI hardware is a handheld ultrasound probe connected to innovative AI software which automatically suggests whether DDH is present on images taken by the user. In this validation study, we will compare Hip AI directly to gold standard ultrasound imaging to evaluate whether Hip AI can be used at birth. We will also collect neonatal data to develop and refine the AI tool.

Develop implementation protocols, and identify benefits and risks in real-world clinical integration within the UK, NHS system.

Using Hip AI to detect DDH in babies has the potential to significantly improve the speed and accuracy of diagnosis and improve the management of this condition by identifying cases early. It may reduce missed and late presentation cases and as a result, reduce the need for costly surgeries. Overall improving outcomes for infants and their families.

If proved successful this innovation could lead to a much larger study that would propose an innovative service redesign using the already well established baby hip and wellness checks at birth and 6 weeks, using non expert staff and existing resources for DDH surveillance.

Future studies could also gather further evidence of risk and benefit of our implementation protocols within NHS and support the ongoing development of Hip AI for global good particularly in developing nations with a high disease burden.