A new artificial intelligence tool developed by researchers at the University of Sheffield could help improve care for people with Motor Neurone Disease (MND), also known as Amyotrophic Lateral Sclerosis (ALS). The AI model is designed to predict when patients will need a feeding tube, which is often necessary as the disease progresses and swallowing becomes difficult.
MND is a progressive and fatal condition that attacks nerve cells controlling muscles. As it advances, many patients experience significant weight loss and malnutrition due to problems with swallowing. A gastrostomy, which involves placing a feeding tube directly into the stomach, can be vital for maintaining nutrition and quality of life.
Timing for this procedure is important. If performed too early, it may negatively affect quality of life; if done too late, it can carry greater risks or even become impossible because of weakened breathing muscles.
The research team, led by Professor Johnathan Cooper-Knock at the University of Sheffield's Institute for Translational Neuroscience (SITraN), created an AI model using routine diagnostic measurements to estimate how quickly MND will progress in each patient. This allows clinicians to identify the best time for a gastrostomy.
"One of the hardest aspects of living with MND is the uncertainty, it is a cruel and devastating disease," said Professor Johnathan Cooper-Knock from the University of Sheffield.
"Until now it has been impossible for clinicians to predict when someone living with MND may need a feeding tube - it could be anything from eight months after diagnosis to 20 years.
"By pinpointing the optimal window for a gastrostomy to within three months, doctors and patients can better plan for the surgery and we can help ensure the best possible quality of life and potentially extend survival."
The AI model was trained on data from over 20,000 MND patients. It predicts when significant weight loss—a key indicator that a feeding tube is needed—will occur. At diagnosis, its predictions had a median error of 3.7 months; accuracy improved further six months after diagnosis, reducing median error to 2.6 months.
Professor Johnathan Cooper-Knock added: "This is not just about a surgical procedure; it's about preserving a patient's dignity and ability to maintain nutrition safely. For a clinician, knowing this critical window allows us to move from reacting to the disease's progression to proactively managing it, providing optimal care and avoiding the distressing complications of rushing a patient to surgery when they are already too frail.
"Ultimately, this tool ensures patients get the right care at the right time, maximizing the quality of every single day."
Results from this study have been published in eBioMedicine. Researchers now plan clinical trials before making this tool part of standard MND care.
