By Ryan Morrison For Mailonline
Published: 20:00 BST, 21 September 2020 | Updated: 20:00 BST, 21 September 2020
AI can be used to to determine whether someone will develop Osteoarthritis by analysing cartilage texture three years before it starts wearing away, study found.
Researchers from John Hopkins Hospital and others ran an artificial intelligence model over scans of 86 people with no discernible symptoms of osteoarthritis.
The machine learning model was about to detect the beginning stages of the condition with 78 per cent accuracy up to three years before symptom onset.
In the UK about 8.5 million people have Osteoarthritis, a condition which causes joints to become painful and stiff - particularly in people over the age of 65.
If the condition can be detected early a combination of weight loss and exercise could make Osteoarthritis less severe when it happens - or even delay onset.
Researchers from John Hopkins Hospital and others ran an artificial intelligence model over scans of 86 people with no discernible symptoms of Osteoarthritis. Stock image
A number of diseases - including Osteoarthritis - lack any visual cues in the early stages which elude image-based detection such as x-ray or MRI.
When it comes to Osteoarthritis, by the time it is diagnosed it is already at an irreversible stage - well after bone damage has already been caused.
Current treatment focuses on palliative care or invasive and costly surgery and combined with other musculoskeletal conditions cost the NHS £10.2 billion per year.
As osteoarthritis primarily affects older people and the population is getting older, it is thought that figure could rise to over £118 billion in the next decade, the NHS said.
Even after symptoms have started and irreversible bone damage has set in, it is still difficult to use medical imaging to correlate pain symptoms to disease progression.
There is some evidence that the condition could be reversible in future - with the right therapies and treatments, the researchers claim.
Early biochemical changes that occur in cartilage often seem to precede pain and bone damage symptoms by several years. This model looks for those changes.
'In the future, our approach may enable more accurate image-based assessments of early disease development,' the team said, adding it could also improve the ability to track the condition as it develops.
The new technique developed by the US researchers will allow doctors to spot signs of the condition developing before it becomes irreversible.
They use artificial intelligence to hunt for patterns in MRI scan images and could even allow for the automated discovery of future damage in cartilage