• AI
  • England’s NHS tale of two technologies

    While some of England’s NHS is leading the way with AI, as previously reported in eHNA, other parts seem heavily reliant on FAX machines. This tale of two technologies, with apologies to Charles Dickens for modifying his book’s title, was revealed in a survey by the Royal College of Surgeons (RCS).

    The Independent, a newspaper, says the RCS finding’s that the NHS remains "stubbornly attached" to fax machines. It identified almost 9,000 FAX machines in about 75% of NHS trusts. Using this archaic technology for a significant proportion of their communications looks even more odd when at the other extreme, aspirations for AI are underway and being fulfilled.

    The BBC has highlighted that one NHS trust has 603 machines, over 6.5% of the total identified. Nest in line are 400 and 369 at England's biggest trust. Taken together, the three organisations have about 15% of the total. It’s a heavily skewed distribution.

    In its blog last year, Deep Mind had identified that the NHS was the world’s biggest buyer of FAX machines. The RCS findings are not a surprise. 

    Another dimension of the ICT legacy is that National Health Executive, a blog, highlighted that most NHS trusts had about 160 different computer systems. It seem that one end of the NHS’s two technology continuum’s a long way from the other, AI end. It reveals a legacy and investment challenge that all healthcare organisations face.

  • AI in England’s NHS improves precision and saves time

    The UK’s NHS is 70 years old this year. It was born on 5 July 1948, so happy birthday Since then, it’s seen a continuous surge in new technologies and techniques leading to transformation. eHealth’s been an increasing component over the last 30 years or so. Now, AI’s coming into the investment frame.

    The Guardian has an article on AI at  Addenbooke’s Hospital, Cambridge. It’s used to delineate tumours. AI completes the work in minutes. Doctors use InnerEye from Microsof to mark-up scans prostate cancer patients. Images from completed scans are anonymised, encrypted and sent to InnerEye to create 3D models. It’s learnt to do by training with previous patients

    Brain tumours are next on the list. For some cancer patients, may have to review over 100 images doctors to plan their radiotherapy treatment. It’s obviously time-consuming, both for doctors, their colleagues and can defer the start of treatment for patients.

    The other significant AI benefit’s greater precision, so improved effectiveness. By focusing more precisely on cancer cells, it helps doctors providing radiotherapy treatments to avoid healthy tissues. 

    Both benefits, precision and time-saving, are leading to more streamlined, more effective and more efficient healthcare. These are some of eHealth’s main goals.

    The NHS has also Heart Flow, developed by Stanford University. It’s AI uses routine CT scans from patients with suspected heart disease AI to create personalised 3D models of their hearts and blood flow. It reveals how specific blockages disrupt blood flow in individual blood vessels, leading to better treatment decision, or none where appropriate. Over half the patients with HeartFlow data avoided angiograms. 

    London’s Royal Free Hospital has an AI development underway. It analyses and refines blood test results and to predict which patients are most likely to die, or have serious problems such as kidney failure. It’s trained from almost 1 billion blood test results from 20 hospitals, and identifies subtle changes in red and white blood cells and electrolytes such as sodium and potassium. It reveals which patients’ health may be deteriorating. 

    Other AI services include:

    Skin cancer diagnosesEye disorders from retinal scansHeart disease from echocardiogramsStrokes.

    The results are encouraging. Africa’s eHealth strategies and plans need a place for AI the data it needs.

  • What does eHealth have to do for radiology services?

    Radiologists are in short supply.  Radiology workloads and demand are rising. A report from Digital Health explores the opportunities to use AI and Radiology Information Systems (RIS) in the UK’s NHS to fill the gap. It identifies essential requirements for national eHealth too.

    Two solutions are proposed, both needing RIS: 

    Sharing reporting workloads across healthcare organisations

    Using AI to automate some of the clinical workload.

    Current images and workflow sharing relies on  an Image Exchange Portal run by Sectra. It’s fast, but seems it needs replacing to meet radiologists’ needs of: 

    Knowing when an image is there for reviewA single system that displays their own images and other clinicians’ images for individual patientsAccess to each patient’s reporting history and images needed for full and useful reports.

    This needs a specific organisational structure, a lesson for Africa’s health systems. In the days before England’s National Programme for IT (NPfIT) was abandoned, radiology information could be shared across each of England’s five NPfIT regions.

    Since then, smaller geographic consortia have emerged to procure Picture Archiving and Communications Systems (PACS) and RIS from single vendors. It achieves lower costs, smoother, more efficient workflows and makes their sharing easier. Patients, radiologists and organisations outside these consortia don’t benefit.

    Vendor-neutral standards are the solution. Two, Soliton and Wellbeing Software, provide solutions share radiology reporting across several sites with different RIS vendors. Their impacts are constrained because there isn’t a single or unified procurement organisation.

    Is RIS becoming obsolete? EPRs and PACS may be able to deal with scheduling and remote reporting. Some radiologists see it differently. They may be increasingly dependent on RIS.

    AI may be a solution too. It’s already dealing with some basic reporting. Wellbeing has a platform for  an AI algorithm to report directly into its RIS. 

    Agfa uses the term Augmented Imaging (AI). It’s exploring the potential for its AI to automate some administrative tasks. Algorithms are already available to detect TB on chest X-rays. Partnering’s already in place with hospitals and research institutes that need Agfa’s workflow engine to develop their own algorithms. 

    Lessons for Africa’s eHealth are clear. Radiology needs its own eHealth engagement, strategy, plans and procurement.

  • AI and machine learning need data storage resources

    Many things come in bundles. Amit Ray, author of Mindfulness Meditation for Corporate Leadership and Management says “As more and more artificial intelligence is entering into the world, more and more emotional intelligence must enter into leadership.” It’s not enough. A report by Source Media, sponsored by Pure Storage says powerful, advanced computing and storage capacity and capabilities are needed too.

    It recognises AI’s “vast” potential. Currently, some radiology departments use it effectively to improve workloads. Progress across other clinical activities depends on extra computing and storage power for two activities, training and clinical use.

    When researchers deliver AI and machine learning techniques to clinical practice and healthcare, solutions need huge amounts of data for training models, including labelling data. It’s especially important for neural networks. These are hardware and software patterned on the way neurons work in human brains. They’re deep learning technologies often focusing on solving complex signal processing or pattern recognition problems.

    If storage’s inadequate, it can’t keep up with the workload. The result’s diminished AI. Healthcare’s typical eHealth investment model’s to buy enough computing storage infrastructure as a minimum requirement, then expand it a few years after it’s clogged up. Eventually, it’s replaced with modern solutions after a period of obsolescence.

    This doesn’t fit AI and machine learning. It has to match the computer power and storage capacity needed as AI and machine learning expands. Developers and healthcare organisations can then move beyond exploring AI’s potential and bring into full use. The, patients benefit.

    While assembling the resources needed for AI and machine learning’s challenging for Africa’ health systems, the infrastructure requirements add to the constraints. Before venturing into the AI space, it’s essential to contemplate and deal with the whole resource requirements and their affordability. 
  • Experts offer their different views on London hospital’s AI

    The AI project announcement by a major London hospital’s attracted a wide span of opinions and ideas. Building on the plan reported in eHNA, the Times has several letters on the initiative. 

    Prof Sir Robert Lechler, President of the Academy of Medical Science emphasised the requirement to have the basics in place to realise the benefits and more healthcare to lead toe fourth industrial revolution. It includes sufficient resources for collaboration with industry, academia and regulators.  He sets the goal of people’s good mental and physical health.

    Hilary Evans, CEO Alzheimer’s Research UK, sees AI as an opportunity to revolutionise dementia research. Her goal is to improve early detection and diagnosis of the progressive disease. 

    Another prof, Harold Thimbleby from Swansea University has a different view. He says more AI’s a simple theory and that much of it data is flawed. Instead, fixing bad ICT would be more cost-effective, offer increased medical value and extend across more health conditions. The effect can be dramatic. 

    Nicola Perrin, Head of Understanding Patient Data, says AI success relies in patients having confidence in how their data’s used. A constructive dialogue’s needed with the public, She evokes the alarms raised from the Facebook and Cambridge Analytic controversy.

    J Merrion Thomas, a surgeon, says the money for AI would be better spent  on earlier benefits, such as highlighting known risk factors and early diagnosis. These will save lives immediately, rather than wait for AI’s benefits.

    These wide-ranging comments are just as relevant for all types of eHealth. They illustrate the engagement and commitment challenges of eHealth’s numerous stakeholders and provide valuable lessons for Africa’s health systems. eHealth never goes ahead in a straight line.

     

  • Is AI set to take off in a London hospital?

    As marches go, AI in healthcare’s still in its early stages. It may be that it’s about to make a big leap forward. The UK’s Guardian newspaper  reports that University College London Hospitals (UCLH) and the Alan Turing Institute has agreed a three-year partnership to realise AI’s benefits to healthcare on an “unprecedented scale.” Planned projects include using AI to: 

    Improve UCLH’s A&E department’s performance, currently below 77% of patients needed urgent care treated within four hours, well below the standard set for England and stuck at 2010 levelsAnalyse CT scans of 25,000 former smokers recruited as part of a research projectAutomate cervical smear tests assessments. 

    A challenge is avoiding learned helplessness. It’s where health professionals become too reliant on automated instructions and abandon common sense. AI’s algorithms might be correct 99.999% times, but are rarely 100% reliable.

    Another’s sustaining rigorous data governance standards, especially privacy and confidentiality. The plan’s to apply algorithms to UCLH’s servers to avoid breaches. Private companies won’t have access. 

    A previous AI project in Engalnd’s NHS was a collaboration between London’s Royal Free Hospital and Google’s DeepMind. The Royal Free accidently gave Google access to 1.6 million records of identifiable patients.

    Alan Turing was an English computer scientist, mathematician, logician, cryptanalyst. He was highly influential in developing theoretical computer science, formalising concepts of algorithm and computation using the Turing machine. In the 1940s at Bletchley Park, he was instrumental in developing the Bombe machine to crack enemy’s complex and rapidly changing Enigma code.

     

  • AI is also attractive for cyber-criminals

    As healthcare increases investment on eHealth projects and services, there should be synchronous investment in security measures.  In 2017, 25% of all data breaches were related to the healthcare industry.  This is because cyber-criminals have been working to make their attacks more advanced to easily target connected devices, cloud, and multi-cloud environments.  These advanced cyber-attacks are even able to evade detection by most legacy security solutions in place. 

    Advancements are aided by adopting AI and machine learning to carry out complex attacks at a rapid pace. Botnets such as Reaper have been made more sophisticated, enabling them to target multiple vulnerabilities at once.  Others, such as polymorphic malware allows for hundreds of variations of a threat to be created for different purposes in a matter of hours. 

    To address these challenges, Fortinet has recently released a few product enhancements that will tip the scales back in the favour of the healthcare industry;

    Fort iOS 6.0 – provides an integrated security architecture that spans the distributed networkFortiGuard AI – is an AI solution that is able to address automated attacksThreat Intelligence Services (TIS) - provides visibility into network activity and metrics to give healthcare security teams an understanding of their threat landscape 

    It has become inexpensive for criminals to mount attacks on healthcare data, but increasingly expensive for their targets. One key to the healthcare security transformation is flipping this paradigm.

  • Philips introduces AI tools for healthcare efficiency

    HIMSS Conference and Exhibition is synonymous for sharing new innovations and showcasing next world technology.  At HIMMS18 in Las Vegas, Phillips announced the launch of a new set of tools that supports the progressive adoption of analytics and artificial intelligence (AI) in key healthcare domains.

    Their HealthSuite Insights gives data scientists, software developers, clinicians and healthcare providers access to advanced analytic resources to compile and analyse healthcare data.  It also offers tools and technologies to build, maintain, deploy and scale AI-based solutions. 

    AI-based solutions have great potential to improve patient outcomes and healthcare efficiency. However, developing and deploying AI solutions for healthcare use cases can be time consuming, resource intensive and expensive.  Philips’ Insights Marketplace can help with this.  The Insights Marketplace will provide the healthcare industry’s first ecosystem where curated AI assets from Philips and others are readily available for license. 

    Philip’s HealthSuite Insights and Insights Marketplace may help accelerate Africa’s eHealth development. African countries are increasingly aware of the necessity of technology in improving the performance of healthcare.  Some parts of Africa have already started integrating artificial intelligence in their healthcare systems.

  • Facebook’s using AI to prevent suicides

    According to the World Health Organisation (WHO), a suicide occurs every 40 seconds globally.  Social, psychological, cultural and other factors can interact to lead a person to suicidal behavior.  Facebook believes that they are uniquely positioned to help combat suicides amongst adolescents and its users.

    They’re using AI and smart algorithms to detect suicidal tendencies and patterns.  The AI software scans users’ messages and posts for signs of suicide, such as asking someone if they are troubled.  Facebook already has tools in place for people to report concerns about friend's who may be considering self-harm, but the new AI software can speed the process and even detect signs people may overlook. 

    Posts that are flagged as worrisome are communicated to first-responders.  It’s also dedicating more human moderators to suicide prevention, training them to deal with the cases 24/7. They have partnered with organisations like Save.org, National Suicide Prevention Lifeline and Forefront from to provide resources to at-risk users and their networks. 

    Ubiquitous technologies often come with unrealised responsibilities.  Facebook’s demonstrating they're willing to take on these responsibilities and use their platform for greater social and health benefits.

  • Will robots be cooking on gas in hospital kitchens?

    Inpatients need nutritious meals as part of their care plans. This puts hospital catering services as an important part of healthcare teams. While robots in clinical activity have received considerable attention, their opportunities in hospital catering hasn’t. Flippy might change that.

    A report  in Tech Crunch says Miso Robotics is rolling out a robotic kitchen assistant. It’s called Flippy. It’s first job’s flipping burgers. Already, it’s a bit of a celebrity, with a YouTube and Vimeo performances. 

    While burgers may not be the ideal meal for inpatients, Cali Burger makes and sells burgers in twelve countries and found Flippy its first job. It doesn’t look like a chef.

    It’s a small, wheeled cart with a six-axis robotic arm and  a sensor bar. It takes data from thermal sensors, 3D sensors and several cameras to assess its environment. Digital systems send tickets from the counter to the kitchen as Flippy’s orders.

    Then, it picks up unwrapped burgers, puts onto a hot grill, tracks their cooking time and temperature, then alerts chefs when to apply cheese or other toppings. When that’s done, Flippy plates the burgers.

     but doesn’t wrap them or add finishing touches like lettuce, tomatoes, avocado or a restaurant’s signature sauce.

    Momentum Machines makes kitchen robots too. Flippy’s different. It relies on  AI software and machine learning, so it learns to make new foods, adapting to a restaurant’s seasonal menu changes. This might be the potential for Flippy’s descendants to take on more sophisticated jobs in hospital kitchens. Let’s hope they’re not wayward offspring called Floppy.