SCIENTISTS are building an artificial intelligence platform that could revolutionise how Parkinson’s disease is identified and tracked.

The initiative, coordinated by the University of Bradford alongside Leeds Teaching Hospitals NHS Trust and Hospital de Clínicas in Paraná, Brazil, centres on analysing smartphone video footage to identify minor motor irregularities that may precede official diagnosis.

Dr Ramzi Jaber, a data science researcher specialising in applied AI at the University of Bradford, explained that one of the project’s primary goals was facilitating remote identification of movement disorders through examination of video recordings where patients perform various movement exercises.

Individuals can capture footage on their mobile devices of themselves completing these exercises and transmit the files to cloud-based systems for processing.

By deploying artificial intelligence to examine these recordings, subjective clinical assessments of movement irregularities can be converted into measurable information, enabling precise tracking of how Parkinson’s symptoms evolve.

A person with Parkinson’s movement patterns being documented by University of Bradford researchers, with a camera capturing a series of structured movement assessments

Parkinson’s disease constitutes a progressive neurological condition resulting in trembling, muscle stiffness, diminished equilibrium, and reduced mobility speed.

Current assessment relies heavily on a 50-question form largely dependent on doctors’ visual evaluation of motor symptoms.

The artificial intelligence framework under development incorporates tools such as Google’s MediaPipe alongside bespoke models refined using patient information.

The system can evaluate movement severity according to the same five-level scale employed by neurologists.

The research group, which additionally comprises Dr John Buckley, a reader in movement biomechanics within the School of Computing and Engineering, has evaluated the approach on elderly individuals and those living with Parkinson’s.

Subjects are filmed while completing standardised movement exercises including finger tapping, wrist turning, and foot tapping.

The artificial intelligence subsequently employs computer vision to produce exact movement measurements for interpretation.

Given that Parkinson’s also correlates with increased fall susceptibility, the Bradford researchers are creating a supplementary artificial intelligence instrument to evaluate lower limb capability, particularly focusing on ankle function.

Their 30-second heel-rise assessment can be conducted securely without professional oversight, presenting a convenient substitute for conventional walking analysis.

The developed technology applies computer vision to examine how someone moves and maintains consistency while repeatedly rising onto their toes for 30 seconds.

Clinical testing, which commenced in 2019 with 120 volunteers, has indicated that the artificial intelligence tools demonstrate promise not only for recognising Parkinson’s but also for detecting fall vulnerability among elderly populations.

The technology could provide quicker identification, more uniform observation, and timely support for susceptible individuals.