The “Clot” Thickens with Artificial Intelligence

<blockquote> <h3>Fast Facts:</h3> <ul> <li>Every year in the United State, more than 400,000 people develop deep vein thrombosis (DVT), usually in the lower extremities (e.g., legs and thighs). And 30%-50% of those who develop DVT can have long-term symptoms and disabilities.</li> <li>According to the Centers for Disease Control and Prevention (CDC), around 60,000 to 100,000 Americans die of deep vein thrombosis each year. 10%-30% of those people die within one month of their diagnosis.</li> <li>In the past, to diagnose DVT, doctors would perform simple tests – such as pulling a person’s toes toward their nose (Homan’s sign) and squeezing their calf muscles (Pratt’s sign) to see if they would experience any pain; however, these tests were not very effective or reliable.</li> <li>Up to 90% of patients are either undiagnosed or misdiagnosed using diagnostic tools for DVT, such as a D-dimer blood test or an ultrasound scan. This could be problematic because some patients may be prescribed drugs for DVT when they don’t have it, or worse, other patients are not being treated for the DVT that they do have.</li> </ul> </blockquote> <p>Millions of Americans’ derrières have been spending more and more hours of the day in chairs since they transitioned to working from home during, and after, the pandemic lockdown.</p> <p>If that’s you, you might have wondered at some point if you are at risk of developing deep vein thrombosis (DVT), a blood clot in a vein deep beneath the surface of your skin. After all, isn’t sitting for long stretches of time said to be a major risk factor for DVT?</p> <p>DVT may seem innocuous, especially since symptoms only occur in about half of the people who have it, and when diagnosed early and treated immediately, DVT isn’t life-threatening. However, the challenge is detecting and treating DVT before it becomes dangerous.</p> <p>Well, it may not be so much of a problem any longer because a team of scientists at the University of Oxford have developed an artificial intelligence (AI) algorithm to diagnose DVT as efficiently as a doctor, but more quickly and potentially more accurately.</p> <h2>What is deep vein thrombosis (DVT)?</h2> <p>The body naturally forms blood clots to stop you from bleeding when you get injured. Sometimes a clot can form inside a vein deep below the surface of your skin, called a deep vein thrombosis (DVT). This can happen because your veins are injured or the blood flowing through them is sluggish. DVT usually forms in the legs – notably the thigh and calf. Though rare, DVT may also develop in an arm.</p> <p>And while a blood clot forming in these parts of the body may seem like no big deal, it’s what happens next that could be deadly.</p> <p> </p> <figure><img alt="" src="https://cdn.storymd.com/optimized/1dmYxYsLdp/thumbnail.jpg" /> <figcaption>Deep Vein Thrombosis <em>Source: sportEX journals</em></figcaption> </figure> <h2>DVT: the silent killer</h2> <p>DVT can become a silent killer, meaning it develops without any obvious symptoms or warning signs. When symptoms do occur, they’re often in the form of swelling, redness, and pain along the affected vein.</p> <p>When DVT goes undetected and untreated, it can lead to other problems, such as chronic vein inflammation, varicose veins, and persistent skin ulcers near the site of the blocked vein.</p> <p>Even more worrisome, DVT has the potential to be life-threatening. That’s because these blood clots in your legs can break loose and travel through your bloodstream to an artery in your lungs, causing a dangerous – sometimes deadly – blockage called a pulmonary embolism (PE) (also referred to as a lung clot).</p> <h2>Are you a sitting duck for DVT?</h2> <p>The problem is, sitting for extended periods of time is not the only risk factor for developing DVT, but any condition that alters how your blood moves through your veins can raise your risk. These include:</p> <ul> <li>Being overweight or obese, which puts more pressure on the veins in your legs and pelvis</li> <li>Taking birth control or any hormone therapy</li> <li>Being over 40 years of age</li> <li>Having a family history of DVT or any other blood clotting disorder</li> <li>Having cancer, inflammatory bowel disease, or heart disease</li> <li>Being a smoker</li> <li>Having an injury that damages your veins, like a bone fracture</li> </ul> <ul> <li>Having two or more of these risk factors</li> </ul> <h2>COVID pandemic increased the risk of DVT</h2> <p>Yes, scientists have found that COVID-19 increased the risk of DVT, but it was a problem even before the pandemic. More than 400,000 Americans develop DVT every year, usually in the lower extremities (e.g., thighs and calves). And 30-50% of those who develop it have long-term symptoms and disabilities. Even more worrisome, according to the Centers for Disease Control and Prevention (CDC), around 60,000 to 100,000 Americans die of DVT each year and between 10-30% of those people died within one month of their diagnosis.</p> <p>While DVT is usually treatable if discovered early, many people are not even aware they have DVT until they experience more severe symptoms. Sometimes people are even misdiagnosed with other conditions because symptoms of DVT can mimic an infection or cellulitis.</p> <h2>Diagnosing DVT</h2> <p>In the past, to diagnose DVT doctors would perform simple tests – such as pulling a person’s toes toward their nose (Homan’s sign) or squeezing their calf muscle (Pratt’s sign) to see if they would experience any pain. However, these tests were not very effective nor reliable.</p> <p>Now, in order to more reliably diagnose DVT, a blood test (called D-dimer) to detect a type of protein produced by blood clots and/or an ultrasound scan interpreted by a radiologist are used.</p> <p>And while these diagnostic tools are way more reliable than the simple Homan’s and Pratt’s tests, up to 90% of patients are still either undiagnosed or misdiagnosed. You can see how this could be problematic because in such cases some patients may be prescribed drugs for DVT when they don’t have it or worse, some are not being treated for the DVT that they do have.</p> <h2>Teaching AI news tricks</h2> <p>Well, the good news is a team of scientists at Oxford University have developed an artificial intelligence (AI) algorithm that can diagnose DVT more quickly and as effectively (or more) as traditional radiologist-interpreted diagnostic scans.</p> <p>The scientists collaborated with a tech company, ThinkSono, to train a machine learning AI algorithm, called AutoDVT, to distinguish patients with DVT from those without DVT. The researchers found that the AI algorithm accurately diagnosed DVT when compared to the gold standard ultrasound scan.</p> <p>Not only can AutoDVT be trained to analyze ultrasound images to determine the presence versus the absence of a blood clot, it can also direct the technician using the ultrasound wand to the right locations along the vein so that even a non-radiologist can acquire the right images and detect the blood clot.</p> <p>On top of that, after crunching the numbers, it turns out that using AutoDVT could potentially save patients about $150 in health care costs.</p> <p>This was the first study to show that machine learning AI algorithms can potentially diagnose DVT. And while this is great news, the use of AutoDVT may still be a few years away because the researchers still need to conduct a double-blinded clinical study to compare the accuracy of AutoDVT with the current standard of care to determine the sensitivity of the AutoDVT for picking up DVT cases.</p> <h2>More on Deep Vein Thrombosis</h2><ul><li><a href="https://soulivity.storymd.com/journal/6wxgrdrhzm-deep-vein-thrombosis" target="_blank">Deep Vein Thrombosis (DVT): Risk Factors, Symptoms, Treatment</a></li><li><a href="https://soulivity.storymd.com/journal/4m8add60pw-artificial-intelligence" target="_blank">Artificial (AI) for Health: How Computers Are Helping Medicine</a></li><li><a href="https://soulivity.storymd.com/journal/zjb7nrdi2m-blood-clots-and-travel" target="_blank">Blood Clots and Travel: What You Need to Know</a></li></ul>