• Cardiology
  • AI predicts heart failure

    Heart failure’s not always easy to predict. CADence, described in eHNA post, shows how difficult it can be and how eHealth can help. Now, Artificial Intelligence (AI) can help too. It can save lives by identifying patients who need more aggressive treatment, says the UK's Medical Research Council (MRC) team in an article in BBC News. Technology’s advanced so much that AI can now calculate accurately when patients with heart disorders will die. The software can to do this by analysing blood tests and scans of beating hearts to spot signs that show they’re about to fail.  

     Researchers, at the MRC London Institute of Medical Sciences, were investigating patients with pulmonary hypertension, raised blood pressure in arteries supplying the lungs. Results indicate that high blood pressure in the lungs damages part of the heart, and about a third of patients die within five years of being diagnosed.  

    The AI software helps doctors to predict how long the patients will live. The data improves informed clinical decisions, leading to more precisely prescribed and individually tailored intensive treatments, such as drugs, injections into blood vessels or in extreme cases, lung transplants if necessary.  

    The AI was given MRI scans of 256 patients' hearts, and blood test results. The software measured movement of 30,000 different points in the hearts’ structures during each heartbeat. Data was coupled with eight years of patient’s health records. AI learned which abnormalities predicted when patients would die.

    It estimated up to approximately five years into the future, with an 80% potential to predict whether people would live beyond a year.  This’s a third better than doctors performance of 60% accuracy.  Researches want to use AI for other forms of heart failure such as cardiomyopathy, diseased of heart muscles, to see who needs pacemakers or supplementary treatments. They also hope to test AI on other patients in several hospitals to assess whether it should be widely available to doctors.

    eHNA’s previously posted about AI revolutionising healthcare. Pulmonary hypertension might be a good place for Africa’s health systems to start.

  • Med-e-Tel’s conference’s 5 to 7 April

    Telecardiology, social media and beyond are Med-e-Tel’s annual conference themes for 5 to 7 April 2017. It’ll include sessions on:

    1. Pharmacy and m and eHealth, presented with the Pharmaceutical Group of the European Union ( PGEU)
    2. Primary care
    3. Mental health
    4. Telenursing
    5. Women's and maternal health
    6. Disease management and remote monitoring
    7. Education and eLearning

    Agence eSanté Luxembourg will present on:

    1. Artificial intelligence for decision support
    2. Multidisciplinary consultation meetings and coordination in oncology
    3. Quality of mHealth apps, with Med-e-Tel.

    The preliminary conference programme’s now available. Join and be a part of this exciting program and networking event. Hear from and meet with colleagues from around the world, and develop new partnerships and collaboration. You can register now.

  • Algorithms identify heart failure risk

    Lateral thinking was devised by Edward de Bono. Born in Malta, one of his opinions is “Dealing with complexity is an inefficient and unnecessary waste of time, attention and mental energy.” Cardiologists don’t see it quite that way. Africa’s health systems should follow the cardiologists lead.

    Algorithms using clinical data from EHRs offer opportunities for better healthcare. Cardiology services are taking advantage of these new analytic techniques, A study in the Journal of the American Medical Association (JAMA) Cardiology describes how cardiologists are develop algorithms that use readily available clinical data to identify hospital patients with heart failure. They’re diagnosing heart failure diagnosis based on discharge diagnosis and their review of sampled EHRs.

    It’s leading them to better, real-time case identification so they can target interventions to improve quality and outcomes for hospital patients with heart failure. Problem lists aren’t good enough for the task. They’re useful for case identification, but often inaccurate or incomplete. Machine-learning’s seen as a way to improve accuracy, but have drawbacks too, such as implementation complexities.

    The team completed a retrospective study of random 75% of hospital admissions of patients over 18 months at New York University Langone Medical Center. Data included demographics, laboratory results, vital signs, problem list diagnoses and medications to treat heart failure. Five algorithms for identifying heart failure were developed using data from EHRs.

    1. Heart failure on problem lists
    2. Presence of at least one of three characteristics: heart failure on problem list, inpatient loop diuretic or brain natriuretic peptide level of 500 pg/mL or higher
    3. Logistic regression of 30 clinically relevant structured data elements
    4. Machine-learning, using unstructured notes with over 1,118 data items in the model
    5. Machine-learning using structured and unstructured data, with 947 data items.

    The problem list algorithm identified about half the patients with heart failure. It’s insufficient for real-time identification. The next two had better results, but the machine-learning ones had the best predictive accuracy because they relied on free text notes and reports.

    However, it’s not a simple decision for Africa’s health systems to opt for machine learning algorithms. The research team says they’re difficult to implement because they rely on unstructured data and may need special expertise and resources. Instead, the researchers suggest that investment choices may depend on cardiologists’ clinical and operational needs. For Africa’s cardiology services, it may depend on the availability of data from EHRs too.

    De Bono went on to say “There is never any justification for things being complex when they could be simple.” Selecting an algorithm to identify heart failure may not be simple. The team says there may be a trade-off between costs benefits. Complex and simple aren’t binary choices. For Africa, the five methods may offer an investment ladder to eventually reach complexity and maximise benefits.

  • mHealth for cardiology may not be mature yet

    As mHealth marches on, the front line may not be as straight as a military model, and cardiology may be lagging behind, or perhaps slightly out of step. An article by a team from Stanford University School of Medicine and Center for Digital Health in the American Medical Association’s (AMA) JAMA Cardiology, authors from Stanford University School of Medicine and Center for Digital Health sets out the challenges.

    While cardiovascular care has been at the centre of healthcare transformation, including using EHRs, there’s much more to do. Displaying data in EHRs is seen as effective, but associated with information overload. Expanding the data by transferring more from wearables and remote monitoring will exacerbate the problem. Investment in sifting data and using it effectively’s needed too. It’s not just clinicians who face this challenge, but patients too, as they have more access to their own data.

    As mHealth apps keep cropping up in various locations, they’ve not been accompanied with good evidence about their therapeutic value. Oversight and management by healthcare teams are needed to ensure patients remain engaged, so they have to be integrated into overall healthcare models.

    Gathering and assembling rigorous evidence is becoming increasingly important. Rigorous studies of validation, effectiveness and implementation in actual healthcare settings for new digital technologies is vital. Studies randomised trials show effectiveness need supplementing with patient-level data and results. Sharing results between hospitals’s needed too, alongside implementation case studies and increased investment for training clinicians in eHealth.

    The authors say healthcare organisations and vendors are collaborating more to work out these problems. While this is welcome, there’s a big workload looming. Africa’s health systems should take heed of the USA’s experiences before expanding mHealth in their cardiology services.

  • Apple Watch can identify your stroke risk

    Atrial Fibrillation (AFib) can cause the heart to beat irregularly. While people with the disorder don't necessarily feel any symptoms, they're at risk of a stroke or heart failure. If a device close to a patient, such as an Apple Watch, could identify people at risk, that would be a great advantage.

    UC San Francisco and Cardiogram experts are collaborating in the Health eHeart study to investigate whether sensors in the Apple Watch can be used to identify people at risk of strokes. mRhythm, the observational study aims to "Contribute your data and save lives." Brandon Ballinger, a cofounder of Cardiogram, tells Fast Company that researchers have two major goals for the study:

    • Quantify the accuracy of the Apple Watch sensor for clinical research
    • Collect evidence that the Cardiogram app and algorithm can detect AFib.

    Experts are not convinced that a smart-watches have yet proven their ability to monitor and track heart rates safely enough to support making medical decisions. A chest strap that closely emulates an electrocardiogram (ECG) is more accurate than wrist-worn devices. Perspiration and rapid movement, for instance, can affect a smartwatch's ability to measure heart rate accurately.

    Ballinger is more upbeat. "We are trying to understand the distribution of errors when you're sitting or walking, or male versus female’"  He’s hoping to recruit 10,000 patients for the study. His team will test other devices that track heart rate, such as Fitbit.

    Medical device startup Alivecor has created a medical-grade ECG band for Apple Watch. It measures electrical activity in hearts and analyses it for problems. The company claims it’s the first wearable to provide instant ECG analysis approved by the US Food and Drug Administration (FDA). Kardia band has an app that displays the EKG in real time. When users touch their Kardia Band's sensor, it sends information to the app. The algorithm then looks for signs of AFib. The system also includes a Normal Detector that indicates if heart rates and rhythms are in the normal range. Another feature notifies users when they need to retake their ECGs. Users can share their data with a health professional at any time.

    AliveCor spokeswoman Rebecca Phillips says the Kardia Band only needs good contact between the electrode and skin, "anywhere on the forearm to the fingertip." Neither sweat nor tattos diminish the ECG results. Kardia’s intended for patients diagnosed with AFib. It could make a big difference for these patients in Africa too.

  • Heart failure apps need improving

    Few apps for Heart Failure (HF) conditions meet the required criteria for quality, content, or functionality. This’s the finding of a study by a team from Columbia University, its College of Physicians and Surgeons and New York-Presbyterian/Columbia University Medical Center, and reported in the Journal of Medical Internet Research (JMIR). The team proposed that the apps need refining and mapping to evidence-based guidelines to meet the need for overall quality improvement in monitoring HF symptoms and self-care.

    HF’s a common, complex, and costly cardiovascular condition. A study by Banerjee and Shanthi Mendis in Current Cardiology Reviews identified a need for a global health perspective because “HF is likely to grow as a major clinical and public health challenge due to demographic changes as well as rising prevalence of causative risk factors in aging patients, including hypertension, coronary artery disease, degenerative valve disease and obesity.” Africa’s stretched health systems will have to deal with this challenge, and effective mHealth should offer a viable solution. It seems that it’s some way off.

    The Columbia study’s the first to undertake a comprehensive, independent review and evaluation of commercially available apps for HF symptom monitoring and self-care. Disappointingly, and despite the HF patients’ needs, most mHealth apps support healthy living rather than chronic disease management. However, many apps are used with minimal knowledge of their functionality and ability to integrate data into health care systems, and with limited efficacy for improving patient or clinical outcomes.

    Consequently, many HF apps have not yet been readily adopted into routine clinical management. They need further development to support comprehensive symptom management for HF patient. These findings are important for cardiologists and their patients in Africa. mHealth for HF needs rigorous assessment and business case before a decision to proceed.

  • A web-based tool estimates these heart patients’ risk

    In 2013, the USA’s American College of Cardiology (ACC) and the American Heart Association Task Force on Practice Guidelines (AHA) published:

    • 2013 ACC/AHA Guideline on the Assessment of Cardiovascular Risk
    • 2013 ACC/AHA Guideline on the Treatment of Blood Cholesterol to Reduce Atherosclerotic Cardiovascular Risk in Adults.

    There’s now an ASCVD Risk Estimator as a companion tool to the guidelines. It’s for cardiologists and other health workers caring for patinets with atherosclerotic Cardiovascular Disease (ASCVD). ASCVD includes coronary death, nonfatal myocardial infarction, or fatal or nonfatal stroke. It relies on the Pooled Cohort Equations and Lifetime Risk Prediction tools.

    Healthcare providers can use the Risk Estimator for patients with a Low-Density Lipoprotein (LDL) cholesterol level of <190 Milligrams per Decilitre (mg/dl) to assess the ten-year and lifetime risks for ASCVD. It’s not designed for patients with LDL levels of >190 mg.dl.

    ASCVD risk factors include age, sex, race, total cholesterol, HDL cholesterol, systolic blood pressure, blood pressure lowering medication use, diabetes status, and smoking status. Measures of the ten-year ASCVD risk are derived from data from multiple community-based populations.

    Estimates of lifetime risk for ASCVD are provided for adults aged between 19 to 60.They’re shown as a lifetime ASCVD risk for a 50-year old without ASCVD who has the risk factor values entered in the Estimator.

    As Non-Communicable Disease (NSD) rates increase in Africa, the Risk Estimator can have an expanding value. It, and equivalent tools should be part of countries’ latest eHealth strategies.


    Image from naplesperiodontist.com

  • Whatever you say, Cordio can tell if you'll have a heart attack in a week

    Voices are revealing. George Santayana, a Spanish philosopher, said “The Soul is the voice of the body's interests.” Now, an app can listen to your voice and tell if your heart’s about to take a dive over the next week. It works by tracking changes in people’s tone of voice.

    At the Times Cheltenham Science Festival, Prof Tony Young, NHS England Clinical Director for aInnovation, said Cordio, the voice-monitoring app’s “one of the most brilliant things” he had seen.

    He’s also consultant neurologist at Southend Hospital. His view of mHealth’s that enabling patients and their families to monitor and treat conditions is crucial to caring for an ageing population with an increasing number of chronic diseases. Cordio’s clinical trials showed it accurately predicted admission to hospital for people with congestive cardiac failure, one week before they were taken gravely ill. There’s a piece on Cardio in Networked India.

    As Africa steadily amasses apps, their possibilities keep expanding toward the horizon. Each country needs a clear strategy that achieves two main goals. One’s to roll out proven mHealth across all appropriate communities. The other’s to take on new technologies and ideas so mHealth initiatives can expand their impact.

    Cordio’s a good example. It’s not likely to stray into monitoring thinking soon. Plato said “Thinking: the talking of the soul with itself.” It has to talk to the Cordio app to head off a cardio crash.


    Image from Dr Bill Sukala

  • eHealth algorithm and regulation let down cardiovascular patients

    When eHealth contributes to healthcare mission-critical activities, it’s vital it’s accurate, reliable, consistent and available. It seems that the UK’s NHS may have experienced a bit of a problem. 

    A report in The Times says “hundreds of thousands of patients could have been put at risk of heart attack and stroke or wrongly prescribed statins because of a software glitch.” It seems that an algorithm, the QRISK2 Calculaotr, wasn’t right. As a result, GPs had to contact all their patients assessed cardiovascular conditions using the SystemOne clinical programme since 2009, so over the last seven years. Some 2,700 GP surgeries may be affected. The Medicines and Healthcare Products Regulatory Agency (MRHA) has started an investigation.

    TPP, SystemOne’s owners, says that its solution contain over 40m patient records from more than 5,000 NHS organisations. These include more than 2,700 GP practices and 142 prisons. The system is verified by the NHS Health and Social Care Information Centre

    TPP and MHRA are collaborating to resolve the matter. It’s envisaged that it’ll take some time. The events reveal the need for effective eHealth regulation and compliance, a lesson for African countries. As eHealth moves further into using algorithms to support clinical activities, the more regulation and compliance reviews are needed. Effective eHealth regulation is costly. So’s the lack of regulation.

  • Apps can track your heart rate

    Numerous apps are available to measure and track your heart rate. They use a combination of the phone’s camera and flash. When covered by your finger, the flash lights up the flesh and the app measures subtle changes in light reflection as blood pulses through. Basic versions are generally free to download. 

    Standard features with most of the apps include options to add various types of information, to help track what you were doing when you measured your heart rates. This information can be processed through an analytical engine to provide insights into your state of health, and feed it back to you in graphical views that help you to make sense of the numbers. Cloud storage and cloud processing are common, as are ways to export or share the information with others, either directly via the app or through email or social media. 

    One of my favourite heart rate apps is Cardiio (rated 4* for iOS) that I’ve written about before. Cardiio adds a novel second way of capturing the heart rate by measuring subtle changes in your face when you look into the camera. 

    Cardiograph (rated 4* for Android and iOS) adds a geotag to each heart rate saved, helping to create a record of where you were for different heart rates. It also allows synchronization of the readings captured across multiple devices.

    Instant Heart Rate by Softonic (rated 4* for Android) is another. It’s available on iOS and Android. The basic version is Free. The preium version, needed to unlock some of the more sophisticated analytics, will cost you R199,99 per month or R579,99 per year.

    Among other features, the premium license adds automatic acquisition of sleep length from it’s sleep app, including this data in the analytics that estimate your state of health. Instant Heart Rate also integrates with the Argus calorie counter and activity app to help you plan to improve your fitness levels.

    Upgrading also gives you access to its StandUpTM test. It collects two readings, one seated and one standing, and provides additional insights from analytics run on this data.

    There are many more heart rate apps available. With so many choices, you may want to examine the field carefully before committing to one. You’ll need to consider which features you need, such as:

    • Other information you’d like the app to collect
    • Insights you’d like analytics to provide
    • Ways to share information form the app with other people

    That should narrow the choices. Next, consider:

    • How easy do you find the interface to use?
    • Is there support for when you get stuck?
    • What’s the cost of the feature package you need?
    • Have there been any formal reviews of the app, in reputable publications, and what do they say?
    • Does you doctor, nurse or other health worker support integration between their systems and the apps, possibly even accepting automatic alerts if your readings raise any red flags?

    Working through these questions is likely to be tedious, so I expect many of you will simply download the apps that look interesting and use the first that seems to fit your requirements. That’s what I did. But as the role of apps and their health-enhancing potential grows, more rigorous evaluation and reporting, in more accessible ways, is likely to become valuable.

    In the mean time, email eHNA if you discover something you’d like to share with other eHealth enthusiasts in Africa. We’ll publish your views.