Six Ways Artificial Intelligence Is Impacting Patients
AI is no longer the stuff of science fiction – it’s impacting patients today
It’s difficult to open a newspaper nowadays without seeing an article about artificial intelligence. These column-inches spark our imaginations with heady visions of possible futures and crease our brows with concern in equal measure. But one thing you cannot escape is that AI is here now and it’s only going to become more pervasive.
While fear of an unknown technology is understandable, in many ways it does a disservice to the incredible impact that AI is already having on the world around us. In the healthcare space alone, it is offering ways to fundamentally rethink clinical practice, speeding up diagnosis, driving patient support programs and aiding drug discovery.
In only five years, more than 200 venture capital and private equity deals to fund research into the use of AI in healthcare have been signed. Governments are equally interested; the UK Prime Minister Theresa May recently promised to revolutionize the NHS by deploying AI and smart technology to diagnose cancer earlier, aiming to save 20,000 lives a year.
“2018 seems to be the year that AI is really accelerating on the high curve,” says AJ Triano, SVP of Engagement Strategy at GSW, a Syneos Health company. “But it’s also accelerating on the practical applicability.”
Triano describes AI as a “broad term for a collection of technologies that consume or utilize data”. This includes natural language processing (NLP) – where AI can hear and make sense of conversation not just pre-scripted words – programmed bots that listen via text, search or voice and use NLP to interact in a conversation, and machine learning, or “the ability for a machine to learn based on a series of engagements and arrive at new meaning or insight on its own.”
While there will be countless dazzling applications of AI in the future, what technologies are already impacting patients today?
1. Clinical trial recruitment
Apps such as Deep 6 AI and Antidote are rethinking clinical trial recruitment to overcome the shortage of suitable participants, which delays research and consequent access to valuable treatments.
Deep 6 applies NLP and machine learning to medical data to match patients to appropriate trials in “minutes rather than months.” It analyzes both structured data (such as recognized diagnosis codes) and unstructured data, from doctor’s notes to pathology reports, extracting tens of thousands of data points like symptoms, diagnoses and treatments. The company also claims its software can identify patients with conditions not explicitly mentioned in medical records.
Researchers, analysts and doctors can use the “multidimensional profiles” to compare patients, cohorts or populations at “machine speed”, matching them to protocols for available trials.
Antidote – a digital health company that aims to “accelerate breakthroughs in potentially life-saving treatments by bridging the gap between medical research and the people who need it” – has the same goal of structuring clinical data.
It’s “clinical trial matching platform” has a range of tools for patients and researchers, including Match, which has been embedded into the websites of hundreds of health publishers and nonprofits, “making trial search accessible to millions”.
2. Chatbots for triage
Healthcare bots are part of a growing wave of virtual assistant-based AI that harness existing speech recognition platforms, such as Apple’s Siri and Amazon’s Alexa.
Last year, Alexa joined the medical ranks through a collaboration with health-advice website WebMD, a deal that allows Alexa-enabled devices like Amazon Echo or Fire TV to answer basic health-related questions.
Asking, “Alexa, ask WebMD how to treat a cough,” searches the platform’s considerable archives and yields either direct advice – in the case of a basic question – or directs you to the relevant section of WebMD’s website.
Many companies are following suit. Health insurance company Vitality recently launched its own Alexa ‘skill’ focused on health and wellbeing that allows users to verbally request recipes, tips on healthy living or workout advice.
Facebook Messenger is also a popular platform for developers; “global health community”, HealthTap, offers medical diagnoses through AI chat conversations, with an option to submit questions to over 100,000 US doctors.
“The largest single application of AI at the population level is in the UK, where Babylon Health is working with the NHS system,” says Triano. “That is proving out the idea that AI can be a tool to help alleviate the everyday doctor drowning in data and compressed by time.”
A recent trial saw Babylon’s AI-powered chatbot ‘triage’ service as an alternative to the NHS’ standard telephone helpline service (which is used by patients for medical advice and for directions to local and out-of-hours services).
NLP is at the base of Babylon’s AI, while machine learning taps into huge volumes of medical data, enabling billions of combinations of symptoms, diseases and risk factors to be evaluated per second, and possible diagnoses suggested.
Babylon’s goal? “We will make our doctors more human, and let computers do the computing,” according to CEO, Ali Parsa.
3. Virtual patient support
Bots are also being harnessed in the form of ‘virtual nurses’ in messaging apps. Virtual assistants are often presented with real names and avatars to make the AI experience feel more human and less like talking to a machine – a crucial factor in the uptake of such digital health technologies. Future applications could include remote health monitoring through webcams, prompts to take medication or live feedback of vital signs.
A number of solutions are drawing on AI to provide patient support between healthcare visits. Sensely offers UK and US patients an AI-powered virtual nursing assistant, announcing a pilot within the UK NHS. Ask NHS is an “all-in-one integrated platform” for patients to access NHS services; the app includes a symptom checker, health advice and resources, appointment booking and information on local services.
However, the Sensely app has wider applications beyond symptom triage and self-care education. Remote monitoring is delivered through an interactive nurse avatar, which patients can converse with via voice and/or text. Additionally, the app can link to connected medical devices to monitor weight and blood pressure.
Web and smartphone app TAVIE also offers remote patient support through a virtual nurse, with a focus on coaching patients to improve adherence to treatment regimens in conditions ranging from type 2 diabetes to HIV. Highly-rated chatbot Ada, available across Europe, is a “personal health companion”, with AI capabilities that allow symptom checking and advice, and, in the UK, the app can then switch the conversation to a real physician, based on what answers are given.
4. Accelerating diagnosis
The use of AI in radiology prompted headlines across the world when human-trained machines outperformed humans at reading scans. Based on computer vision – where a machine can make sense of an image – and machine learning, algorithms can match patterns and flag potential problems far faster than a professional could hope to.
Google’s AI for object recognition technology, when adjudicated by physicians to become more specific, had an 89% success rate for malignant breast cancer detection compared with 73% for humans.
“In the oncology space around radiology, GE Healthcare have been doing some fabulous work,” says says Nick Bartlett, Managing Director of patient-experience benchmarking company, Patient-ly. “They are looking to link up their machines to create a database with an artificial mind, which can scan and learn from the hundreds of thousands of images it takes over the year.” GE has been a proponent of widening access to AI development, building open platforms that can be leveraged around the world to hasten progress.
Others in this space are Chinese tech company, Infervision – “an extra pair of eyes for radiologists” – which uses AI to read CT scans and X-rays to better detect the early signs of lung cancer, and Viz.ai, which recently gained FDA approval for its AI-backed tech that analyzes CT images of the brain to detect potential strokes.
The Viz.ai algorithm automatically alerts a neurovascular specialist if it spots a suspected large vessel blockage at the same time as the first review of the images takes place, saving time usually spent waiting for a radiologist to review the images. Researchers in Italy have also reported positive early results for algorithms that identify the development of Alzheimer’s before symptoms are shown, while an algorithm developed by researchers from Stanford University has diagnosed 14 medical conditions from chest X-rays, including the better diagnosis of pneumonia than an expert radiologist working alone.
Several firms, including Skin Analytics, are using computer vision to detect melanoma via smartphone images, and IBM’s supercomputer Watson, which combines AI and sophisticated analytical software to allow it to answer questions and remains at the forefront of the AI revolution, is working on a number of projects, including an ambitious mission to use machines to beat cancer.
In India, Remidio has developed the world’s first smartphone-based retinal imaging system, where AI object recognition software can identify diabetic retinopathy at a low cost. “It’s a really great application where we go from the big capability and the future potential to actual application,” says Triano.
5. Digital therapeutics
A chatbot may seem an unlikely confidant, but a growing body of research suggests great promise for AI to deliver talking therapies for psychological support.
As well as offering support at the click of a button, digital therapy has the advantage of being available 24/7. “This is a big up-and-coming thing that goes far beyond chatbots,” says Duncan Arbour, SVP Strategy, Innovation, Europe, at Syneos Health Communications.
He points to “poster boy” Woebot, which draws on cognitive behavioral therapy (CBT) to ask people how they are feeling during brief daily conversations.
Teaming NLP with psychological expertise, Woebot sends videos and tools based on how it judges your needs. A randomized study at Stanford University showed using Woebot led to significant reductions in anxiety and depression compared to a control group reading self-help material.
Fellow mental-health chatbot, Tess, built by clinical psychologists, uses text message conversations for coaching.
Designers drew on a range of therapies, including CBT, solutionfocused (brief) therapy (SFBT) and mindfulness, when putting the app together. Combination platform Ieso uses human therapists, but is powered by AI algorithms to support the best allocation of therapists, and to enable real-time monitoring of therapy sessions. The company says this helps assess risks and guides therapists in relation to their clinical decisions.
6. Drug discovery
As health professionals in West Africa battled the deadly Ebola virus, pharmaceutical companies scrambled to develop vital medications and vaccinations for treatment and prevention. One company, Atomwise, that put AI on the case found two drugs that may one day reduce the infectivity of Ebola, an analysis that can, without the latest technology, take years.
Atomwise’s engines are based on convolutional neural networks – “the same AI technology that recognizes faces in a crowd, enables self-driving cars and allows you to talk to your phone.” Insights are extracted from millions of measurements and protein structures; as it scans billions of compounds, the AI spots patterns that humans never would, identifying small subsets for testing.
By allowing chemists to pursue meaningful molecules, their time is optimized, and the lengthy and costly process of drug development is given a welcome boost. As such AI could be the future of meeting unmet medical needs, and saving hundreds of thousands of lives.
Since you're here...
... and value our content, you should sign-up to our newsletter. Sign up here