What can you tell us about the research so far?
This exciting programme of work is a collaboration between ThinkSono – the company which has developed the AI algorithm, called AutoDVT, and Oxford University/OUH Hospitals. The premise is that normally a DVT is diagnosed by a highly trained, ultrasonographer – either a radiographer or a doctor. Training to diagnose DVT takes a long time. AutoDVT however can direct a non-specialist how to use an ultrasound device to detect the presence, or absence of a clot.
AutoDVT has been tested in many groups of individuals – both healthy volunteers and also within the DVT clinic in Oxford. These studies have shown that AutoDVT is good at picking up vein clots. So far, we haven’t tested AutoDVT against the standard ultrasound method of picking up DVT and we will be starting a multi-centre study in several UK DVT hospital clinics to see if it is as reliable as the normal ultrasound.
How will the process of diagnosing DVT with the help of AI work in practice?
If AutoDVT is a reliable tool, it will potentially revolutionise the normal diagnostic pathway for patients who seek medical review for a possible DVT. Currently, patients see their GP and then are required to go to their local hospital or local radiology services for an ultrasound scan, after initial risk stratification. Ultrasound appointments can be limited and patients may need to wait for an appointment. This means that the patient is likely to need to take an anticoagulant treatment until the scan has been performed. AutoDVT will enable patients to have an immediate scan at their GP practice, improving patient access to rapid diagnosis which is close to home. NICE guidance recommends that DVT diagnosis is made within 24 hours of a patient presenting to their GP/doctor and this will enable this to reliably happen.
How could this improve outcomes?
As above – and also, it will reduce the need for patients to take interim anticoagulation, whilst awaiting a scan, which will reduce side effects from these drugs. Early diagnosis will also reduce the chances of a DVT moving into the lung, causing a pulmonary embolus.
When might the AI algorithm become available in the UK?
There is a little bit more work to do, however, we would hope within the next few years.
What might be the main challenges to its uptake?
Making people comfortable with a new clinical pathway usually takes time, especially if they are used to doing things a certain way. However, our multi centred clinical study will hopefully provide evidence and reassurance that this new technology is safe and accurate. Additionally there will also be work required to enable seamless integration with hospital IT systems such as PACS (used commonly used for imaging). But we don’t expect these challenges to be too big of a barrier for adoption.
What’s your Holy Grail – the one thing you’d like to achieve in your research career?
To develop a more reliable means of risk stratifying patients who are at risk of bleeding or thrombosis.
Who has inspired you in work or life?
In work – the first haematologist I worked for, Dr Charlie Gutteridge in London. A totally awesome doctor – caring, knowledgeable and with a great sense of humour.
How do you achieve work-life balance?
I switch my computer off on a Friday (when not on call) and don’t switch it back on until Monday.
There’s an app for that. What’s new on your phone?
The ‘Seek’ app – rather ridiculously, I’ve only just got it. [The seek app enables identification of any living thing – plants, flowers, insects, fungi, birds and animals]