Providing independent clinical excellence since 2005
Posted on Monday January 15, 2024 in News
Written by Dr Edward Leatham, Consultant Cardiologist
Healthcare today faces numerous challenges, including rising costs and limited access for those most in need. The UK’s healthcare system, particularly the NHS, is often hitting national news, which makes negative reading. However, at SCVC, we’re embracing innovative technologies and remain very positive.
As we begin 2024, I want to highlight three groundbreaking developments from 2023 that promise a brighter future for healthcare, two of which leverage the power of artificial intelligence (AI).
Cardiac CT (CCT) scans have been used more and more over the past decade to investigate patients with coronary artery disease (CAD). Although a CCT is over 98% sensitive for the detection of a narrowed coronary artery (stenosis), it can over-sense severity and thus create false positive results, which leads to many unnecessary hospital admissions for invasive coronary angiography. Less than 10% of the 1GB of digital data produced in a single CT can be seen and interpreted by the human eye, so there are many new data processing companies to assist in medical diagnosis. The first of these became available in 2016, when cardiologists all over the world started using ‘CT FFR’ data processing from HeartFlow Inc in the US, to visualise the degree of flow limitation present in each coronary artery.
8 years on, HeartFlow FFR CT processing is still only available in high volume centres (and thus not available at most private hospitals). At SCVC/VCL we have sufficient volume of specialists using our cardiac CT service to be able to send selected scans to HeartFlow in USA to measure the CT FFR. This means far fewer of our patients need invasive angiography or other tests. Indeed our consultation rooms are next to our own in-house cardiac CT scanner (operated by VCL Surrey), which facilitates one stop clinics where the CT results can be reviewed during the consultation. There are not many places in Europe that patients can schedule a consultation, testing and a coronary angiogram and walk out with their results and plan a few hours later.
In the example shown below, HeartFlow data processing (left panel below) confirmed a severe narrowing of the left circumflex coronary artery. The patient was subsequently admitted to hospital for invasive angiography where the non invasive findings were confirmed (right panel below) and a severely narrowed (yellow arrow) vessel stented.
Below: the left panel below shows three coronary arteries with the HeartFlow FFR colour flow map superimposed: where red signifies flow limiting stenosis. The right panel shows a single frame from the hospital cine angiogram taken a few weeks later in the catheter lab, where a stenosis was confirmed (arrow) and then stented.
Predicting future coronary events: CT images are assessed using AI.
Not only can we send our patient’s CCT for functional assessment using FFR CT, but from 2023, we may also send the same CT data for a different form of data processing which uses AI and machine learnt systems, based in Oxford UK. By detecting and quantifying ‘coronary inflammation’, Caristo’s AI technique now also predicts an individual’s survival. The fat attenuation index (FAI) fed into Caristo algorithms predict the likelihood of future fatal heart attacks. The implementation of this technology at SCVC gives our cardiologists the option to diagnose patients with active coronary inflammation, then offer more effective preventative measures, transforming patient outcomes.
Examples of two patients CaRi heart data processing below.
For more information on CaRi heart score and FAI, see related blog can we use AI to predict cardiovascular death?
While high LDL cholesterol, Apo B, and LP (a) have long been recognised as important risk factor for coronary inflammation and coronary heart disease, about half of a patient’s future risk remains a mystery, so scientists keep searching for other risk factors. New insights have recently emerged that indicate that glucose spiking and variability may be an important one – since even in non diabetics, wide excursions in blood and tissue glucose promote an inflammatory response . At SCVC, we’ve shifted our focus to metabolic syndromes and glucose dysregulation. Widespread use of continuous glucose monitors (CGM) shows that non-diabetic individuals can experience glucose spikes and glucose variability (GV) at high levels typically only seen in poorly treated diabetics. By incorporating CGM testing into our metabolic health screen and incorporating dietary advice, we’ve seen significant improvements in glucose variability in our patients with early signs of CHD.
The graphs below show the results of two weeks of glucose monitoring in this 31 year old, apparently healthy non diabetic female with normal HbA1c, fasting blood glucose and normal response to 75G glucose load at 2 hours – oral glucose tolerance test (GTT).
The upper panel shows an episode calculation profile (ECP) where the baseline BG runs at a (healthy) 5.5 mmol/L (100 mg/dl). The blood glucose exceeded 7.8 mmol/L (140 mg/Dl) an average of once/ day. The lower figure ‘spikogram’ plot shows coloured points where blood sugar changes upwards or downwards very rapidly (hence spikes). Of all the movements during the week, there were 8 +ve and 5 -ve ‘spikes’, which are considered to be unhealthy. For more information see Unravelling the Conundrum of Glucose Spikes
Like many, we found Chat GPT (used without any possible patient identifiers, and disabling the vendors ) marks a significant improvement in our data management processes. While initially, the AI’s overly creative interpretations posed challenges, its real strength lies in correcting the spelling, grammar, and punctuation of our speech recognition platform’s output. This integration has not only expedited the generation of accurate medical letters but also streamlined the data entry process, enhancing our auditing capabilities and reducing medical mishaps associated with delayed communication and prescription errors.
The example below shows how the doctors voice recognised transcription that appears on the screen in real time, can be converted into nearly perfect script in seconds using ChatGPT linguistic processing. This development has lifted the quality and turn around time of medical communications.
Each of these technologies represents a significant step forward in our quest for a more efficient, effective, and patient-centered healthcare system. They exemplify our commitment to innovation and our belief in the power of technology to transform healthcare for the better.
It is my belief that such developments will inch us towards a better, as well as more cost effective healthcare system, but only if combined with a greater focus on individual responsibility to prevent or delay oneself from developing common diseases of the 21st century. In my head, at least, doctors need to partially change their role to become the educators, tutors and influencers that guide people, who are confused and awash with fake news and poorly evidenced health fads. Our patients need more accurate, early diagnosis and to be encouraged towards evidence-based lifestyle and preventative strategies that they can do for themselves.
The idea of predicting a heart attack may have once seemed like science fiction, but it is rapidly becoming a reality. With the advent of advanced imaging techniques and artificial intelligence, we are now able to detect the early signs of coronary heart disease and intervene before a heart attack occurs. This represents a major shift in the way we approach cardiovascular care, moving from a reactive model to a proactive one.
Is is increasingly obvious that not everyone with raised coronary inflammation and elevated risk of heart attack simply has a raised LDL Cholesterol- in fact there appears to be a mixed bag of risk factors associated with high coronary artery inflammation, including raised LDL, raised LP (a), raised Homocysteine, hypertension and dysregulated glucose. Our metabolic health assessment is therefore designed to offer a comprehensive dive into known contributory causes, the latest and most controversial of which is 'glucose dysregulation', the principle topic of this article.
For patients who have been diagnosed with an acute coronary syndrome, requiring bypass surgery or stent, or have CT Angiography and or raised FAI, the latest European and UK guidelines reflect the finding that there is no lower limit for LDL yet discovered, basically meaning the lower the better. These recommendations come after new trials with lower targets showed better outcomes. As a result, millions of people around the world need to be looking more carefully at whether they are taking sufficient medication to lower their LDL to these new targets, for the very best outcomes to be achieved.
Ultimately, the approach to cardiac CT screening is highly individualized, reflecting the diverse presentations and risk factors among patients. By employing a sophisticated pre-consultation process, cardiologists can tailor their assessment and management plans to best suit each patient's unique needs, enhancing the precision and effectiveness of cardiac care.
There are two main reasons left atrial thrombus forms then detaches to cause embolic stroke. The first is that during atrial fibrillation or flutter, a thrombus may develop in the left atrial appendage, which does not move as much as it does when in sinus rhythm. The likelihood of this occurring in any given year is low, correlated to the CHA2DS2-VASc score. The second reason is not related to CHA2DS2-VASc score
While predicting a heart attack with absolute certainty may still be beyond our reach, the ability to identify patients at high risk is a significant step forward. The latest techniques used to find people at risk are summarised in a […]
Being born with almost no circulating cholesterol (or reduction of LDL to zero by drugs) results in nothing except a zero-lifetime risk of atherosclerosis (or cessation and reduction in atherosclerosis).
Recognition that significant coronary artery disease exists without calcification led to the wider user of contrast-enhanced CT scans to detect both calcified and noncalcified plaques. Even more advanced scans identify not only the non calcified plaques, but also areas of inflammation (using data processing to measure FAI), that a non-contrast CT (such as a CAC scan) would otherwise miss. Sequential CAC to follow up disease progression quickly established that an increase in CAC with age was inevitable in most patients and not linked to clinical course, so is not widely practiced.