Predicting human movements

Smart multiscale models of human joints and muscles are being developed by bioengineers in the Insigneo Institute for in silico Medicine, which will give clinicians the power to assess the risk of bone fractures in patients with osteoporosis and help predict the onset and development of Parkinson’s disease.

Claudia Mazza

Dr Claudia Mazzà, who recently moved to Sheffield from the University of Rome “Foro Italico” , says the project is developing patient-specific tools using a range of data obtained from laboratory gait analysis (study of human motion), MRI and CT scanning, fluoroscopy and wearable sensors.

“We know the traditional approach to gait analysis is not sufficient. Often it is based on common assumptions about bone structures and muscle movements which don’t capture the patient-specific information we need to build more predictive models,” said Dr Mazzà, whose research lab is in Insigneo.

“We know that four patients with four different pathologies will need very different algorithms and methodologies. But being able to develop the right algorithms for the right population and the right pathology is not a trivial challenge,” she said.

“For instance, if we want to be able to predict the probability of a patient having a fall we can’t really achieve this in laboratory. We really need to see what happens where the patient lives. And, in the case of patients with Parkinson’s disease and other degenerative conditions, we need to measure them more often.”

One avenue being explored by clinicians is a range of commercially available sensors. Dr Mazzà urges caution, however. Following a detailed investigation into their reliability, she says: “The results – especially when used on patients rather than healthy subjects – were disappointing. In some cases you would have had better predictive odds tossing a coin.”

The tools we are developing are much more sophisticated. They will be able to tell the clinician how, why and when a patient’s gait is getting worse. They should even be able to pick up changes that are so subtle the clinician cannot see them. These methods are bespoke to the patient. Our goal is to be able to provide the clinician with a simple number that tells them what they need to know about the patient’s condition and how it is likely to develop.

Dr CLAUDIA MAZZZà, Insigneo, the university of sheffielD

Discussing her sensor findings at a recent conference organised by the Brain and Movement Research Group in Newcastle, she told her audience there was some good news. “When we did the same analysis a week later the errors were consistent,” she said, provoking laughter from the delegates. “This is good because when you know the errors, you have a better chance of overcoming them.”

And that is her new challenge. “My job is trying to find useful information in the signals. We are taking every single one of the variables suggested in the literature and comparing them using all the different techniques for processing the signals in order to find a common language, so we find a solution that is sufficiently reliable for use in clinical diagnosis and treatment,” she said.

Her modelling and analytical work is already helping her colleagues at the Sheffield Institute for Translational Neuroscience (SITraN) and at Devices for Dignity (D4D) Healthcare Technology Co-operative, who have designed a neck support collar for patients suffering from Motor Neuron Disease. “The collar has been a huge success with patients, but SITraN and D4D would like to take it to market and that requires scientific validation. Our models are helping to provide that rigorous evaluation of the collar’s effectiveness.”

She is also working with fellow Italian, Professor Fabio Ciravegna, who is a specialist in large-scale information management in the Department of Computer Science. The aim is to combine his skills in big data analysis with her bioengineering expertise to analyse biomedical information from the wearable sensing technologies that clinicians might think they can use.

“Sometimes the most difficult part is helping a clinician to understand that there is no magic solution. You must not think that if you can measure it, the data is informative. Some clinicians seem to think you just put the sensor on and you get good data. It’s not that easy.


The Insigneo Institute for in silico Medicine is a collaborative initiative between the University of Sheffield and Sheffield Teaching Hospitals NHS Foundation Trust. Multi-disciplinary in its structure, the Institute involves 139 academics and clinicians who collaborate to develop computer simulations of the human body and its disease processes that can be used directly in clinical practice to improve diagnosis and treatment.

In silico medicine (also known as “computational medicine”) is the application of in silico research to problems involving health and medicine. It is the direct use of computer simulation in the diagnosis, treatment, or prevention of a disease. More specifically, in silico medicine is characterised by modelling, simulation, and visualisation of biological and medical processes in computers with the goal of simulating real biological processes in a virtual environment. The Institute’s work will bring about a transformational change in healthcare through multidisciplinary collaborations across many strategic areas, which will include personalised diagnosis and treatment and improvements in independent, active and healthy ageing.

Find out more: