What doctor? Why AI and robotics will define New Health - www.pwc.com
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Contents Introduction 1 Transforming 3 Keeping well 4 Early detection 4 Diagnosis 5 Decision-making 5 Treatment 6 End of life care 7 Research 7 Training 8 Investment 9 AI investment heats up 9 Rise of the robots 11 Findings 12 The public is ready 12 Will AI replace the consultation? 13 Service preferences 15 Will robots replace surgeons? 19 The new imperatives for healthcare 21 Conclusion 23 References 25 About 27 Contacts 28
Introduction Would you allow a medical robot with artificial intelligence (AI) to examine, diagnose and prescribe a treatment plan for you or a member of your family with cancer? Would you accept a robot as your primary surgeon? And if you were to do so, what would the human clinician do? AI and robots have long been a theme What if it could make healthcare services over the next ten in fiction. From Isaac Asimov’s ‘I, recommendations so that you could years will define our ability to deliver Robot’ in 1950 to Baymax, the care achieve and maintain your ‘optimal’ a more responsive health service with robot in ‘Big Hero 6’, we’ve long been physical or mental condition or detect improved health outcomes, while at fascinated with the possibilities for anomalies early? the same time enabling people to take machines that can understand, reason more control over their day-to-day and learn – and help us to make better Robots like Baymax and tricorder health needs. use of information. scanners are not science fiction. AI and robots that support, diagnose Just imagine that you had a hand-held and treat people are already in scanner the size of your smartphone homes, workplaces and clinical that could ‘sense, compute and record’ environments all over the world. And your daily health status just like the how we embrace AI and robotics to Star Trek medical ‘tricorder’. complement and enhance current New Health We are entering a new era What doctor? The answers to these two questions fundamentally impact clinicians and of health – New Health At the heart of this report is the caregivers throughout healthcare. The Health matters. It matters to each of us emergence and increasing use of answers will ultimately decide how as individuals and to society as a whole. artificial intelligence (AI) and robotics we implement the use of AI and robots It lies at the heart of our economic, within and throughout this New Health in developing our healthcare systems political, social and environmental ecosystem. We chose ‘What doctor’ as across the globe. We will specifically prosperity and is one of the largest our title for the report because during address these two questions in the next industries in the world. our research – and the response to that phase of our research initiative, but it research – it has become increasingly is important to keep them in mind as Modern health systems can treat and clear that policy makers, payers, we explore existing perceptions and cure more diseases than ever before. providers, educators and the public need attitudes towards AI and robotics in New technology is bringing innovation to consider:t this report. to old treatments. Yet significant • What is the role of the AI/ quality, access and cost issues remain robot doctor? and our health systems are becoming increasingly unsustainable. • What is the role of the human doctor? 1
Five distinct trends are converging, which mean AI and robotics will come to define New Health: Last decade 1. First and foremost is the value Past decades have focused on the Medical Products challenge that all countries across the innovation provided by medical products globe are facing – there is escalating delivering historic and evidence-based Equipment, Hardware, demand from long-term, chronic care. The present decade is one of medical Consumables disease, rising costs, often with an platforms focused on real-time, outcome- ageing population and limited resources based care. The next decade is moving (money, workforce specialists, etc). Yet towards medical solutions – using AI, we are continuing to invest in facilities robotics, and virtual and augmented and equipment that were built to solve reality – to deliver intelligent solutions a completely different set of healthcare for both evidence- and outcome-based Differentiation is solely through needs and are not designed to cope with health, and focusing on collaborative, product innovation. Focused on this demand. A hospital-centric system preventative care. This confluence of historic and evidence based-care. deals very well with serious health technology-based products, platforms episodes that require days or weeks and solutions is leading to a previously of acute care for very ill people. But it unimagined precision medicine, down to Current decade was never intended to deal with large the familial and individual level, which numbers of people whose conditions are one day may even be able to predict and Medical Platforms chronic, complex and require treatment thereby prevent disease. Wearable, Big Data, Health for the longer term. Analytics 4. Technology has, of course, had an 2. The past decade has seen an explosion impact far beyond the developments in in the amount of health data that is healthcare. The explosion of technology now available to us. For example, for – digitally enabled, wireless connectivity a skin specialist there are 11,000 new across increasingly mobile devices – has Differentiation by providing dermatology articles published every created an increasing democratisation services to key stakeholders. year. In 2013, it was estimated that of access for healthcare. Some of the Focused on real time outcome the volume of health-related data had most powerful AI tools are already based-care. reached over four zettabytes – that’s embedded in Android or iOS. Harnessing four trillion gigabytes (1021) – and there this technology is providing consumers Next decade are those who project this exponential with the data and information they need growth rate to reach ten times that by to proactively manage their own health Medical Solutions 2020, and, even beyond, to yottabyte and wellness, and to make better, more Robotics, AI, Augmented Reality (1024) proportions1. Furthermore, fully informed decisions in partnership with 80% of this extraordinary amount of their healthcare providers. data is unstructured2, meaning that it’s not contained in a database or some 5. Finally, the willingness of the other type of data structure. Staying general public to be more active current with and being able to access this participants in their own health and data is simply beyond the scope of any wellness has now reached critical mass. human individual, no matter how capable As we reported in our 2016 survey Care Differentiation via intelligent or intelligent. Anywhere, the explosion of technology solutions for evidence/outcome and the increasing ubiquity of the based health. Focused on 3. Information technology development Internet of Things (IoT) is bringing preventive care. in healthcare has been rapidly moving about breakthroughs that are erasing Source: Frost & Sullivan, ‘Transforming from products to services to solutions healthcare boundaries and enabling care healthcare through artificial intelligence (Frost & Sullivan, 2016). anywhere and everywhere3. And this systems’, 2016 willingness is extending into the areas of AI and robotics, which we explore in this report. The Medical Futurist (2016). http://medicalfuturist.com/artificial-intelligence-will-redesign-healthcare/ 1 Healthcare Data Institute (2015). http://healthcaredatainstitute.com/2015/02/18/big-unstructured-datas-contribution-to-healthcare/ 2 PwC (2016). Care Anywhere: Moving health and wellness out of the hospital and into the hands of the consumer. https://www.pwc.com/m1/en/publications/ 3 documents/care-anywhere.pdf 2
Transforming No longer science fiction, AI and robotics are transforming healthcare Most of us are barely aware of it, but AI is getting increasingly sophisticated Throughout this section of the AI is already a part of our lives – it’s in at doing what humans do but more report, we highlight a range of our cars, telling us when it’s time for efficiently, more quickly and at a lower products and existing or emerging the engine to be serviced based on our cost. The potential for both AI and technology in side-bars or boxes driving patterns; it’s in our everyday robotics in healthcare is vast. Just like that we’ve found during our Google searches and the suggestions in our everyday lives, AI and robotics research for these eight areas. It is from Amazon that follow us around on are increasingly a part of our the web; it’s the chatbot on the end of healthcare ecosystem. extraordinary, and hard to keep the telephone in call service centres. AI up with as new uses of AI and is becoming ubiquitous and most of us robotics are launched every week. use it: Apple’s Siri processes two billion natural-language requests a week4, and 20% of requests on Android phones are made by voice alone5. Keeping Training Well Early Research Detection AI and Robotics End of Diagnosis Life Care Decision Treatment Making 3
Keeping well Early detection One of AI’s biggest potential benefits is Artificial Intelligence AI is already being used to detect to help people stay healthy so they don’t and Consumer Apps diseases, such as cancer, more need a doctor, or at least not as often. for keeping well accurately and in their early stages. The use of AI and the Internet of Medical For example, according to the Smart Performance Apparel Things (IoMT) in consumer health American Cancer Society, 12.1 million offers real-time biometrics, applications is already helping people to mammograms are performed annually personalised programming, and manage their own healthcare and to keep in the US, but a high proportion of customised reporting to help themselves well through healthier living. these mammograms yield false results, athletes reach their goals faster. For example, the Smart belt – welt has leading to 1 in 2 healthy women being a built-in mechanism that alerts the told they have cancer. The use of AI Samsung Electronics offers person when they overeat. It relies on a is enabling review and translation of consumers devices such as magnetic sensor to track waste size and mammograms 30 times faster with smartwatches and activity tension to determine when the users may 99% accuracy, reducing the need trackers, as well as their Tizen have over eaten and alerts the wearer. for unnecessary biopsies as well as open-source operating system, Further, IBM has announced three new reducing the uncertainty and stress which gives users quick access consumer-focused partnerships, one of a misdiagnosis6. to their mobile devices and help of which is with Under Armour who them track their fitness. will use Watson to power a ‘cognitive The proliferation of consumer coaching system’. wearables and other medical devices Lumo Lift is a posture monitoring combined with AI is also being applied device that alerts the person These applications and others (see Box- to oversee early-stage heart disease, when their posture changes, thus Out) all encourage healthier behaviour in enabling doctors and other caregivers providing the feedback necessary individuals and help with the proactive to better monitor and detect potentially to ensure correct posture whether management of a healthy lifestyle. It life-threatening episodes at earlier, sitting or standing. puts consumers in control of health and more treatable stages. well-being. Additionally, AI increases Under Armour will use IBM’s the ability for healthcare professionals On the horizon, Microsoft is developing Watson to power a ‘cognitive to better understand the day-to-day computers programmed for use at coaching system’ in an application patterns and needs of the people they a molecular level to start fighting that provides customised advice care for, and with that understanding cancerous cells as soon as they are for sleep, fitness, activity and they are able to provide better feedback, detected. They are also doing research nutrition. The insights originate guidance and support for staying healthy for using AI to interpret online search from three sources: crunching – although, as we have noted, the sheer engine behaviour, for example, at the data from Under Armour’s 200 volume of data is increasingly difficult for point where someone might research million-member Connected humans to deal with, or make sense of. symptoms online long before they Fitness community, external approach their physician. academic research studies and institutions, and IBM Watson similarity analytics. Under AI and Wearables for Armour will be adding new capabilities including behavioural early detection and performance management, Cyrcadia’s iTBraTM is a wearable food intake tracking and overall vest being used to screen for the nutrition management as well detection of breast cancer at as the effects of weather and earlier stages. environment on training. CardioDiagnostics has Pathway Genomics is developed a device that is able to developing an application that remotely monitor its wearer for will provide customised health heart irregularities and is used to advice based on a user’s specific improve cardiac monitoring and genetic makeup. rhythm management. The Ringer (2016). How Siri Got Left Behind. 4 https://theringer.com/amp/p/2ffc86fe5dfa 5 The Economist (2017). How voice technology is transforming computing. http://www.economist.com/news/leaders/21713836-casting-magic-spell-it-lets-people- control-world-through-words-alone-how-voice 6 Wired (2016). http://www.wired.co.uk/article/cancer-risk-ai-mammograms 4
Diagnosis Decision-making It’s estimated that 80% of health data Google’s Deepmind Improving care requires the alignment is invisible to current systems because In 2016, Google’s DeepMind of broad base data analysis with it’s unstructured7. IBM’s Watson for stunned the Go-playing world by appropriate and timely decisions, Health uses cognitive technology to beating 18-time world champion and predictive analytics can support help healthcare organisations unlock Lee Se-dol, one of the highest clinical decision-making and actions as vast amounts of health data and power ranked players in the world. It well as prioritise tasks. diagnosis. Watson can review and store was the first time a computer far more medical information – every programme took on a professional Using the system dynamics driven medical journal, symptom, and case player of this ancient Chinese pattern recognition to identify study of treatment and response around board game, which has long patients at risk of developing a the world – exponentially faster than been considered impossible condition – or seeing it deteriorate any human. And it doesn’t just store for computers to play at the due to lifestyle, environmental, data, it’s capable of finding meaning in international professional level. genomic, or other factors – is another it. Unlike humans, its decisions are all It’s much more difficult than other area where AI is beginning to take evidence-based and free of cognitive games like chess, even though hold in healthcare. For example, biases or overconfidence, enabling rapid the rules are relatively simple, in an extension and application of analysis and vastly reducing – even because of the almost infinite AI, PwC’s Bodylogical™ captures eliminating – misdiagnosis. number of possibilities and the learnings in mechanistic modeling to high levels of intuition required of digitally represent the physiology of Google’s DeepMind Health is a world-class player. Deepmind’s the human body. This enables true- working in partnership with clinicians, programme displayed game- life simulations to predict the likely researchers and patients to solve winning creativity, in some cases progression of chronic diseases in the real-world healthcare problems. The finding moves that challenged future based on today’s actions and technology combines machine learning millennia of Go wisdom. interventions. These simulations help and systems neuroscience to build pharmaceutical companies, providers, powerful general-purpose learning DeepMind Health began working payers, employers, researchers and algorithms into neural networks that with hospitals in the UK in consumers better understand how daily mimic the human brain. 2016, searching for early signs life choices and therapeutics impact of disease that lead to blindness individual patients or population health and cancer, as well as developing outcomes and associated costs. new clinical mobile apps linked to electronic patient records. AI and analytics platforms for decision- making Quest Diagnostics’ Quanum platform is an integrated suite of healthcare information technology and predictive analytics tools that analyses patient test data together with their medical data to help primary care physicians identify patients with early onset memory loss and dementia. VitreosHealth has developed a predictive analytics platform that identifies people whose health is most likely to deteriorate and implements change protocols to prevent catastrophic health events. Healthcare Data Institute (2015). http://healthcaredatainstitute.com/2015/02/18/big-unstructured-datas-contribution-to-healthcare/ 7 5
Treatment Beyond scanning health records to AI Robotics help providers identify chronically ill Innovaccer’s AI Assisted Care KASPAR is a child-sized humanoid individuals who may be at risk of an Coordination Platform addresses robot designed to help teachers and adverse episode, AI can help clinicians persistent issues in care and parents support children with autism. take a more comprehensive approach connectivity, and assures adherence for disease management, better to care plans by tracking deviations Giraff is a mobile communication coordinate care plans and help patients and staying updated on post-acute robot that facilitates a chronically to better manage and comply with their patient needs. ill patients’ contact with the outside long-term treatment programmes. world. It is remote-controlled, AiCure has built an application to Sentrian is a remote patient on wheels, and has a camera monitor patients with long-term intelligence service provider that and monitor. conditions and help them adhere to delivers integrated remote patient medication intake. The application uses monitoring and analytics to health Bestic is a robotic-assisted dining a visual recognition system to identify plan members with complex appliance for people who are unable the patient’s face, the medication pulmonary diseases. to move their arms or hands. they’re taking, and confirm ingestion. The data is then sent back to the Royal Philips delivers remote care Toyota has created four robots that care provider or to a pharmaceutical options to proactively help patients enable immobilized patients to walk company conducting a clinical trial. at home. The services are powered or balance. by data analytics and connected Robots have been used in medicine sensors enabling 24/7 monitoring Xenex robots disinfect hospital for more than 30 years. From the first and delivery of an ongoing stream of facilities using UV light, destroying programmable universal machine for monitoring data to a remote home- microorganisms that can be the cause assembly (PUMA), used in urology care team. of hospital acquired infections (HAIs). surgery in the 1980s, to the da Vinci robot, the most widely used robotic Google DeepMind is partnering Aethon’s TUG robots automate system in clinical use today, robots with the University College London the delivery and transportation of have developed to perform a wide Hospitals’ radiotherapy department. the immense amount of materials range of tasks and functions. They DeepMind will test the use of AI and – including food, laundry and range from simple laboratory robots machine learning to reduce the time it prescriptions – that move through a to highly complex surgical robots takes to plan radiotherapy treatment hospital every day, freeing staff to that can either aid a human surgeon for hard-to-treat cancers of the head focus on patient care. or execute operations by themselves. and neck. In addition to surgery, they’re used Veebot is a robot that can draw blood in hospitals and labs for repetitive IBM’s Watson for Oncology gives faster and more safely than a human. tasks, in rehabilitation, physical treatment recommendations based therapy and in support of those with on patients’ medical records and is long-term conditions. RoBear is a now being used around the world, nursing-care robot that is able to lift including 21 hospitals across China and move patients in and out of bed and in Thailand and India. into a wheelchair, help those who need assistance to stand, and even turn patients in bed to prevent bedsores. 6
End of life care Developed to assist the elderly in their own homes, they are able to monitor We are living much longer than previous for falls and other health parameters, generations, and as we approach the provide alerts, and connect via video- end of life, we are dying in a different conference with healthcare providers and slower way, from conditions as well as with friends and family. like dementia, heart failure and osteoporosis. It is also a phase of life that is often plagued by loneliness. End of life care time data which is then sent to the cloud wirelessly where algorithms Paro Therapeutic Robots are extrapolate such advanced Robots have the potential to advanced interactive robots that information as whether someone is revolutionise end of life care, helping enable the documented benefits of showing signs of dementia. They can people to remain independent animal therapy to be administered also remind the elderly about daily for longer, reducing the need for to patients in environments such tasks and important information, hospitalisation, caregivers and care as hospitals and extended care and track conditions over time. homes by performing routine tasks such facilities where live animals present as taking vital signs and prompting treatment or logistical difficulties. Zora Robotics is building AI for medication. AI combined with the They’ve been found to reduce stress, software into Zora Bots that have advancements in humanoid design are stimulate interaction and improve patient-facing roles in hospitals and enabling robots to go even further and socialisation. are training humanoid robots to hold have ‘conversations’ and other social interactions with the people that keep conversations with the elderly. Robot Era is developing robots ageing minds sharp and solves problems that move around on wheels but of loneliness and isolation. Kompai have a friendly humanoid face. The robots talk, understand speech, remind robot’s sensors and cameras are people of meetings, keep track of used to gather and analyse real- shopping lists and play music. Research processes there is the potential to Each and every one of us has most likely significantly cut both the time to market Research for new drugs and their costs, not only taken a medication prescribed by our for the labs who develop the drugs, The Japanese Government has doctors for a symptom or illness at some but for those people whose health entered into a public-private alliance point in our lives. Those with chronic depends upon them. Pharma.AI is the to develop a self-learning AI at the diseases often depend upon medication Pharmaceutical Artificial Intelligence government-affiliated National to manage what might otherwise be division of Insilico Medicine, a Institutes of Biomedical Innovation, debilitating diseases. But the path from bioinformatics company located at the Health and Nutrition. The research lab to patient is a long and Emerging Technology Centers at the partnership will support researchers costly one. Johns Hopkins University, launched in streamlining the development in March 2016. They focus on drug of new drugs in order to boost the According to the California Biomedical discovery programmes for cancer, nation’s competitive advantage. Research Association, it takes an Parkinson’s, Alzheimer’s, and other average of 12 years for a drug to travel ageing and age‑related health issues. BERG Health, a startup that uses from the research lab to the patient. an AI platform for drug discovery, Only five in 5,000, or .1%, of the drugs However, medical research is not just initiated a phase II clinical trial in that begin pre-clinical testing ever make about finding new drugs to combat 2016 for a drug compound that could it to human testing and just one of these disease. It also includes research into potentially treat pancreatic cancer. five is ever approved for human usage. disease itself, with the ultimate goal Furthermore, on average, it will cost a being to inoculate against or completely Atomwise found two existing drugs company US $359 million to develop eliminate disease. Meta is a Canadian – reportedly in one day using its AI a new drug from the research lab to start-up that uses AI to quickly read technology – that may also work the patient8. and comprehend scientific papers and against the Ebola virus. then provide insights to researchers; Drug research and discovery is one it was bought by the Chan Zuckerberg of the more recent applications for Initiative in January 2017 as part of AI in healthcare. By directing the the charitable foundation’s mission to latest advances in AI to streamline the eradicate disease. drug discovery and drug repurposing California Biomedical Research Association. New Drug Development Process. http://www.ca-biomed.org/pdf/media-kit/fact-sheets/CBRADrugDevelop.pdf 8 7
Training This shift will inevitably alter many of the roles of healthcare professionals. AI allows those in training to go As these new technologies and through naturalistic simulations in perspectives become more integrated a way that simple computer-driven within and across our healthcare algorithms cannot. The advent use systems and more ubiquitous among of natural speech in technology the population, the skills that are and the ability of an AI computer to required by our new health landscape draw instantly on a large database of may well be markedly different than scenarios means AI can respond to those that are needed today. We questions, decisions or advice from currently train our doctors and nurses a trainee and can challenge more in the context of health systems that effectively than a human can. And the may no longer exist once they graduate training programme can learn from medical school. An understanding previous responses from the trainee, of technology will be imperative. meaning that the challenges can be Programming, data analytics and continually adjusted to meet their human behaviour may well be as much learning needs. a part of the medical curricula as anatomy and neurology. And training can be done anywhere with the power of AI embedded on AI and robotics technology will free up a smartphone, for example, as quick clinicians for other types of work that catch up sessions, after a tricky case in enable them to spend more meaningful a clinic or while travelling. time with their patients. Rather than a profession of ‘healthcare providers’, AI To date, the main way new and robotics will open opportunities for technologies have been used to more holistic patient care, with a focus augment training is through virtual on keeping patients healthier longer, reality (VR). Combining VR with AI instead of primarily treating illness. will offer boundless opportunities for extending the skills of trainees in a AI will likely challenge the traditional targeted fashion. role of the doctor. But, rather than worrying if these technologies are AI and robotics are redrawing the going to replace doctors and other healthcare landscape. The wave of healthcare professionals, we should be innovation being driven by these considering more deeply their wider technologies is not only transforming role in the entire healthcare continuum clinical decision-making, patient with a clear eye towards training our monitoring and care, and surgical healthcare workforce for the future. support, but fundamentally changing how we approach healthcare for our populations. We are already experiencing this shift as we focus on integrating prevention and wellness into our health systems and we are heading towards a time when people work more proactively with their healthcare professionals across illness and wellness. 8
Investment AI investment heats up From predictive insights to Figure 1: Artificial Intelligence for Healthcare Applications, World, 2014, 2021 (in Figure 1: Artificial Intelligence Market for Healthcare Applications, Millions) robotic surgery, investment World, 2014, 2021 (in Millions) is booming Investing in health is always a long- term prospect, whether it is a matter 2014 663.8 of training a doctor, developing a new drug, building a new facility or developing AI and robotics. The question is, given that long-term view, 2021 6,662 and the potential of AI and robotics, where should that investment go? 0 1000 2000 3000 4000 5000 6000 7000 8000 Already AI is coming of age, and the AI healthcare market is poised for dramatic Source: Frost & Sullivan 2016 Transforming healthcare through artificial intelligence systems growth. Frost & Sullivan predicts that the AI market for healthcare will increase by 40% between 2014 and In 2011 there were fewer than ten deals According to CB Insights, healthcare 2021. And they estimate growth from related to healthcare AI; this leapt to is the hottest area of investment US$633.8m to US$6.662bn9. almost 70 deals in 2016. Some of the within AI. They identified over 100 2016 deals included a US$25m Series companies that have raised an equity According to Frost & Sullivan analyst A round raised by London-based health funding round since January 201312. Harpreet Singh Buttar, ‘By 2025, AI services startup, Babylon Health, and From insights and analytics, imaging systems could be involved in everything a US$154m Series A round raised by and diagnostics, drug discovery to from population health management, China-based iCarbonX11. remote patient monitoring and virtual to digital avatars capable of answering assistants, AI is poised to impact every specific patient queries.’10 aspect of healthcare. 9 Frost & Sullivan (2016). Transforming healthcare through artificial intelligence systems. AI Health and Life Sciences. 10 As quoted in Huffpost Tech UK, “Artificial Intelligence For Healthcare Is Booming”, by Reuven Cohen (31 Jan 2017). http://www. huffingtonpost.co.uk/reuven-cohen/artificial-intelligence-f_b_14437718.html 11 CB Insights (Feb. 3, 2017). https://www.cbinsights.com/blog/artificial-intelligence-startups-healthcare/ 12 Ibid. 13 CB Insights (Feb. 3, 2017). https://www.cbinsights.com/blog/artificial-intelligence-startups-healthcare/ 9
AI Investment Highlights13 Insights & risk analytics: Health Core AI companies bring their Virtual Assistants: This category has insights and risk analytics has been algorithms to healthcare: Core AI a relatively low deal count, less than the hottest category for investment start-up Ayasdi, which has developed ten since 2012, but has the potential since 2015. Companies in this category a machine intelligence platform based for increased investment activity. One provide predictive insights about a on topological data analysis, is bringing of the start-ups, Babylon Health in patient’s health using machine learning its solutions to healthcare providers the United Kingdom, was backed last and natural language processing for applications including patient risk year by investors including Kinnevik algorithms. The analytics are based scoring and readmission reduction. and Google DeepMind in a US$25m on factors that include medical history Series A round. and demography. Drug discovery: This category is gaining attention, and venture Mental Health: This is another Imaging & diagnostics: This category capitalists have backed six out of the category with relatively low deal count, has become more saturated with nine start-ups on the map, who are as well as fewer companies. Seed-stage companies in recent years compared using machine learning algorithms to start-up Avalon uses AI to predict brain to other areas of healthcare AI, with reduce drug discovery times. degeneration, and focuses on neuro- more than 80% of deals raised after degenerative disease like Alzheimer’s January 2015. Oncology: IBM Watson Group-backed and Parkinson’s. Pathway Genomics has recently started Remote patient monitoring: Deals to a research study for its new blood test Research: This category includes the category have picked up since 2015, kit, CancerIntercept Detect. China-based iCarbonX, which joined though there are fewer companies here the unicorn club in Q2 2016. Another than in diagnostics or risk analytics. The company will collect blood start-up, Desktop Genetics, helps London-based Babylon Health, backed samples from high-risk individuals who scientist with genome editing and by investors including Kinnevik have never been diagnosed with the CRISPR research. It received funding and Google-owned DeepMind disease to determine if early detection last year from genetic research Technologies, raised a US$25m Series is possible. company Illumina. A round in 2016 to develop an AI-based chat platform. Emergency room & hospital Nutrition: Ireland-based Nuritas uses management: Start-ups here provide artificial intelligence to mine data insights and real-time analytics, and identify compounds in food that specifically in a hospital environment. are beneficial to health. Marc Benioff Gauss Surgical, for instance, uses invested over US$2m in the company image recognition to monitor blood in Q2 2016. loss during surgery in real time using an iPad. 10
Rise of the robots Alongside AI, robotics is becoming one of the hottest new markets in tech. According to New entrants, minimally invasive surgical robots17 IDC, the overall market has been growing Verb Surgical is working with perform procedures within small- at a compound rate of 17% a year and will Johnson & Johnson’s Ethicon to medium-size surgical spaces for be worth US$135bn by 2019, with a boom and Alphabet Inc’s Verily to general abdominal, gynecologic, and taking place in Asia, Japan and China. develop the generation of surgical urologic indications. The company is robots, incorporating advanced pre-revenue. The company recently Markets and Markets estimates that the visualisation, machine learning, data raised CAD$15.1m through a market for healthcare robotics will grow analytics and connectivity. public offering. to US$11.4bn by 202014. Surgical robots comprise the largest component of the Mazor Robotics Renaissance Hansen Medical develops, medical robotics market, according to surgical guidance system transforms manufactures, and sells medical IndustryArc, specifically the use of robots spine surgery from freehand robotics designed for the positioning, to facilitate minimally invasive surgery procedures to guided procedures. manipulation, and control of (see box-out). Mazor’s ADSs trade on the NASDAQ catheters. The company’s Magellan under the symbol MZOR. The Robotic System controls the Changes soon could cause something of a company reported a 54% year-over- proprietary Magellan Robotic shift in power around the operating table. year Q4 2015 revenue increase Catheter, a telescoping and Surgical robots may cease to be dumb, to US$8.8m. robotically steerable catheter. In programmable machines and transform 2016 the FDA approved Hansen’s into smart assistants. TransEnterix is a medical device Magellan Robotic Catheter eKit company that is pioneering the use and a few days later, the company Yet there is also growing opportunity of robotics to improve minimally announced its first clinical procedure in the home health and end of life care invasive surgery. They are developing with the Magellan in the US. space for robots. The global personal and commercialising its ALF-X robot market, including ‘care-bots’, could System, a multi-port robotic surgery Stereotaxis designs, manufactures, reach US$17.4bn by 2020, according system that enables up to four arms and markets robotic systems and to Frost & Sullivan, driven by rapidly to control robotic instruments and a instruments for the treatment of ageing populations, a looming shortfall camera, and its SurgiBot System, a abnormal heart rhythms. Its products of care workers, and the need to enhance single-incision, patient-side robotic- include the Niobe ES robotic system performance and assist rehabilitation of assisted surgery system. and Vdrive system. Stereotaxis has the elderly and physically handicapped. strategic alliances with Siemens Titan Medical is a development and Philips Medical Systems. The Japan is leading the way with one-third of stage company that is developing its company recently reported its first the government budget on robots devoted SPORT (Single Port Orifice Robotic quarter of positive free cash flow – to the elderly. The Japanese ‘care-bot’ Technology) Surgical System of US$1.6m – which management market alone is estimated to grow from composed of a surgeon-controlled indicated helped drive a 71% US$155m in 2015 to US$3.728 billion robotic platform with 3D vision reduction in 2015 annual cash burn, by 2035 (source: Ministry of Economy, and interactive instrument control For FY 2015, Stereotaxis registered Trade & Industry)15. for performing minimally invasive total revenue of US$37.7m, which surgery. The company believes represented an 8% year-over- Bionics, exoskeletons and the next SPORT will enable surgeons to year improvement. generation wearable robot are also becoming a reality and could revolutionise how we treat and care for The sheer scope of opportunity raises Markets and Markets (Feb 2017). 14 those with diminished or lost functions many questions for investors. There http://www.marketsandmarkets.com/ due to ageing or physical challenges and are risks – competitive, regulatory and PressReleases/medical-robotic-systems. injuries. The rehabilitation robot market market risk, among others. But with asp was estimated at US$203m in 2014 and is healthcare undergoing such massive 15 Bank of America (2016). http://about. expected grow to US$1.1bn by 202116. change, and the rapid progress of AI and bankofamerica.com/assets/davos-2016/ robotics, how safe are current, traditional PDFs/robotic-revolution.pdf healthcare investment strategies? We 16 Bank of America (2016). http://about. believe that healthcare is fundamentally bankofamerica.com/assets/davos-2016/ about the most effective application PDFs/robotic-revolution.pdf of knowledge in order to keep people healthy. As knowledge progresses, it is 17 MedDeviceOnline (February 12, 2016). New the investor who has the power to bring Entrants to Open Up Surgical Robotics that knowledge to life. Market. https://www.meddeviceonline.com/ doc/new-entrants-to-open-up-surgical- robotics-market-0001 11
Findings The public is ready Trends did emerge across regions – people in those countries with well- In November 2016, we asked YouGov established, and therefore more rigid, Research to conduct a survey of the healthcare systems were willing to general public across Europe, the engage with a non-human healthcare Middle East and Africa (EMEA) to provider, but less so than those in understand 3 things: countries where healthcare still has the • if there was the appetite to engage ability to shape and form. with AI and robots for healthcare; • the circumstances under which In fact, for all questions throughout the there would be greater or lesser survey a pattern can be seen between willingness to do so; and, developed and emerging economies: • the perceived advantages and • Developed economies with disadvantages of using AI and entrenched healthcare systems– robots in healthcare. characterised by a high per capita health spend and a long legacy We surveyed over 11,000 people of universal healthcare coverage across 12 countries and the evidence delivering good overall value – strongly indicates that there is a occupy the bottom half of the scale growing enthusiasm among consumers where respondents were less willing to engage in new ways with new to rely on AI. technology for their health and • Emerging economies are wellness needs. significantly more willing to engage with the new technologies. These The message is clear; the public is countries have mixed degrees ready and willing to substitute AI and of healthcare coverage (Turkey robotics for humans. enacted universal healthcare in 2003, for example) but lower per But behind that message is a lot capita health spend compared to of important information that has developed economies. Some still implications for how these new have access constraints and quality technologies will shape New Health. disparities between public and The findings of the survey suggest private care. three key themes that are impacting • Countries in the Middle East sit in how consumers are willing to engage the middle of our survey results. with AI and robotics: These countries have a longer • People are increasingly willing history with universal coverage to engage with AI and robots if it and spend two to three times more means better access to healthcare; per capita on healthcare than • Speed and accuracy of diagnosis other emerging markets, but are affected by workforce issues and 18 Total health expenditure (THE) in the and treatment is a critical factor capacity constraints18. developed economies in our survey (e.g. for this willingness; and UK, Northern and Central Europe) ranges • Trust in the technology is vital for However, there is generally a between $3,377 and $6,347 per capita. wider use and adoption; the ‘human surprisingly high willingness among Middle East countries range between touch’ remains a key component of all respondents to engage with AI and $2,400-3,000 per capita. South Africa and the healthcare experience. robots, regardless of country, gender or Turkey’s THE per capita is around $1,000, age. From fitness counsel, monitoring while Nigeria has the lowest in the sample and advice on diabetes and heart at $217 per capita. All amounts are Int$ conditions, to both minor and major (PPP) for 2014. Source: WHO Global surgery, we found that consumers Health Expenditure Database across demographics are willing to consider non-traditional options for managing and treating their health. 12
Will AI replace the As a whole, across EMEA there is more willingness than unwillingness consultation? to engage with AI and robots, with The human ‘face-to-face’ element of more than half willing (55%), one- healthcare is traditionally cited as vital third unwilling (38%) and 7% neither to healthcare. And yet most doctors willing nor unwilling. probably spend more time going over medical records and interpreting Yet when we looked at the individual data from tests and other medical country data, there were interesting interventions than actually spending findings. From almost four out of ten time with their patients. respondents in the UK (39%) to more than nine out of ten in Nigeria (95%), As we’ve noted, there is already willingness to talk to and interact existing technology that can perform with a device, platform or robot with these functions. This may well evolve artificial intelligence varied widely. into a human-like interface for patients Figure 2: Percentage of respondents willing/unwilling to engage and consumers of healthcare services Figure with2:AIP ercentage of respondents and robotics willing/unwilling for their healthcare to engage needs (total with AI and sample) – either remotely or in a clinical robotics for their healthcare needs (total sample) environment – where the experience will feel like an interaction with a ‘real’ doctor. How willing are patients 55% 38% 7% to interact with such technology? Willing Unwilling Neither willing We asked our survey participants nor unwilling to consider advanced computer Source: PwC survey technology or robots with AI that had the ability to answer health questions, perform tests, make a diagnosis based Figure 3: Percentage of respondents willing/unwilling to engage with AI and on those test and symptoms, and robotics for their healthcare needs (by country) recommend and administer treatment. We wanted to understand how willing Figure 3: Percentage of respondents willing/unwilling to engage with 94% 100 they would be to engage with this Net: Willing Net: Unwilling 85% technology if it was more accessible and 82% 90 could process health information faster and more efficiently than their doctor or other health professional. 80 66% 65% AI and robotics for their healthcare needs (by country) 62% 70 55% 60 51% 51% 50% 50% 48% 44% 43% 50 41% 39% 39% 39% 40 26% 24% 24% 30 16% 20 11% 6% 10 0 Nigeria Turkey South Africa Saudi Arabia Qatar UAE UK Netherlands Belgium Norway Sweden Germany Source: PwC survey 13
100 Figure 4: Percentage of respondents willing/unwilling to engage 90 80 While there was generally high Figure 4: P ercentage of respondents willing/unwilling to engage with AI and with AI and robotics for their healthcare needs (by age) willingness to engage with AI and robotics for their healthcare needs (by age) 70 robots across all countries, a clear Net: Willing Net: Unwilling 57% 56% pattern emerged: 54% 60 • Nigeria, Turkey and South Africa 47% 45% are significantly the most willing 44% 50 (94%, 85% and 82% respectively). 37% • Two-thirds of the Middle East 40 32% are willing, with percentages 28% statistically the same across 26% 30 the region. • The Netherlands (55%), Belgium (51%), Norway (50%) and Sweden 20 (48%) are less willingness yet still represent a significant percentage of 10 the sample. • Germany and the UK are the only 0 countries where unwillingness 18 – 24 25 – 34 35 – 44 45 – 54 55+ (51% and 50% respectively) is greater than willingness (41% and 39% respectively). Source: PwC survey But the variation across countries should We then wanted to understand how taking pulse and blood pressure, testing not be taken to mean there are not people felt about receiving health advice blood sugar levels, checking kidney opportunities in every country for AI and and information remotely, from an function, and monitoring weight and robotics to be adopted. In every country, ‘intelligent healthcare assistant’, via a level of exercise. a highly significant proportion of the smartphone, tablet or personal computer population is willing to engage with AI – in a situation that involved their own The patterns remained consistent with and robotics for healthcare services. health and also for a loved one (such as the willingness to engage face-to-face, Yet the findings clearly suggest that parent, spouse or child). We asked them with the exception of Germany and implementation is key – the introduction to imagine a specific health situation the UK, where respondents were more of AI and robotics for these services in each case. For their own situation, willing than unwilling in this specific will need a different approach in the we asked them about remote diabetes situation, even though they remained the different countries. monitoring and recommendations for least willing and most unwilling of all any treatment or lifestyle changes that the countries. There were some distinctions with were necessary – this would involve regards to gender – men were generally more willing than women (61% versus use an 'intelligent healthcare assistant' via a smartphone, Figure 5: Percentage of respondents willing/unwilling to use an ‘intelligent Figure 5: Percentage of respondents willing/unwilling to 55%), yet women were significantly more healthcare assisstant’ via a smartphone, tablet or personal computer willing than men in Turkey (91% versus Net: Willing Net: Unwilling 95% 80%), Norway (57% versus 42%) and 89% 84% Sweden (54% versus 41%). 0 10 20 30 40 50 60 70 80 90 100 74% 72% 67% This was the pattern across all countries 60% 59% 58% with the exceptions of Nigeria, where the 53% 51% most willing were aged 18-24 (95%) and 47% 41% 40% over the age of 45 (100%), and Turkey, 38% 34% 36% where willingness increased with age 28% 24% 23% (from 82% to 89%). tablet or personal computer 16% 15% 9% 5% Nigeria Turkey South Africa Saudi Arabia Qatar UAE UK Netherlands Norway Belgium Sweden Germany Source: PwC survey 14
an 'intelligent healthcare assistant' via a smartphone, tablet or Figure 6: Percentage of respondents willing/unwilling to use When the situation involved a loved Figure 6: Percentage of respondents willing/unwilling to use an ‘intelligent one, the respondents were less willing, healthcare assisstant’ via a smartphone, tablet or personal computer though not substantially so. Specifically, for a loved one (e.g. child, parent) personal computer for a loved one (e.g. child, parent). we asked them to imagine that a loved Net: Willing Net: Unwilling 91% one was feeling unwell and they needed 85% 0 10 20 30 40 50 60 70 80 90 100 advice about treatment. This would 79% involve making a diagnosis based 70% 64% on their symptoms, medical history 56% 54% and vital signs (pulse, blood pressure 53% 48% 49% 45% 48% 46% 46% 46% and temperature). 42% 37% 33% 34% 32% While the patterns from country-to- 21% 20% country remained consistent, with 12% significantly higher willingness for Nigeria, Turkey and South Africa, and 9% high willingness across the Middle East, respondents across Northern and Western Europe expressed significantly Nigeria Turkey South Africa Saudi Arabia Qatar UAE Norway Netherlands Belgium Sweden Germany UK less willingness. In Sweden, Germany and the UK those who were reluctant to use this technology for their loved ones outnumbered those who would be willing. Source: PwC survey Again, percentages of willingness remained high, reinforcing the indication for different implementation strategies. Figure 7: Procedures respondents were most willing to recieve from an AI/robot, when asked to rank three Monitor my heart condition Based on test (e.g. pulse, results, my blood Administer a Provide preferences pressure, ECG test that checks customised and drawing etc.), take my heartbeat’s advice for on medical note of my rhythm and fitness and Take and research, symptoms, make health based on test a blood advise me on Prepare and Stitch and and advise on recommenda my personal sample and the best give an injection bandage a Set a bro heart condition tions based on preferences and provide me treatments for of medicine/an minor cut or bone and treatment the results health records with results cancer immunisation wound into a cas 41% 37% 37% 32% 17% 11% 10% 7% Source: PwC survey 15
Service preferences • Provide customised advice for Furthermore, it’s worth noting that fitness and health based on my a remarkably high percentage of A significant percentage of respondents respondents in a majority of countries personal preferences and health were highly willing to choose would be willing to receive advice on records (37%). certain treatments, tests or services treating cancer. And even though the UK • Take and test a blood sample and administered by an AI or robot and there and Germany had a significant number of provide me with the results (32%). was widespread agreement about these people unwilling to use such technology Consistent with our 2016 report services across the countries surveyed. for any type of health service or Care Anywhere: Moving health and wellness out of the hospital and into procedure compared to other countries, Specifically, consumers across the there is still greater willingness than the hands of the consumer, these region were most willing – by a unwillingness; again, important to note findings suggest that the public is significant margin – to receive four when it comes to implementation. most willing to use AI and robotics for types of services: monitoring and providing advice. But • Monitor my heart condition this survey has also shown they are (e.g. pulse, blood pressure, willing to replace a human for these electrocardiography (ECG), etc.), same services. In a world increasingly take note of my symptoms, and dominated by chronic disease this has advise on heart condition treatment a major implication for ambulatory or (41%). clinic-based care. • Administer a test that checks my heartbeat’s rhythm and make However, while every country surveyed recommendations based on the agreed that these four services were results (37%). the ones they would be most willing to receive, the priorities were slightly different from country to country. ? X Not applicable – I would not be willing to use an advanced computer/ robot with Monitor me and artificial d provide general intelligence for a Set a broken care/offer any type of t or bone and put it advice during Deliver my procedure/ into a cast pregnancy baby Other Don’t know service 0% 7% 5% 1% 1% 10% 20% 16
Figure 8: H ealthcare Figure service service 8: Healthcare priorities, by country priorities, by country Monitor my heart condition Based on test (e.g. pulse, results, my blood Administer a Provide preferences pressure, ECG test that checks customised and drawing etc.), take my heartbeat’s advice for on medical note of my rhythm and fitness and Take and research, symptoms, make health based on test a blood advise me on Prepare and Stitch and and advise on recommenda my personal sample and the best give an injection bandage a Set a broke heart condition tions based on preferences and provide me treatments for of medicine/an minor cut or bone and p treatment the results health records with results cancer immunisation wound into a cast Belgium 34% 42% 34% 29% 13% 10% 9% 5% Germany 35% 29% 33% 16% 12% 10% 8% 6% Saudi Arabia 41% 32% 36% 31% 22% 12% 11% 10% Netherlands 33% 38% 30% 27% 15% 10% 9% 4% Nigeria 62% 46% 53% 42% 28% 8% 17% 12% Norway 33% 32% 38% 39% 17% 9% 6% 3% Qatar 44% 40% 46% 42% 19% 9% 9% 7% South Africa 53% 47% 46% 43% 21% 11% 11% 9% Sweden 37% 32% 31% 33% 9% 5% 8% 4% Turkey 39% 35% 38% 40% 32% 13% 13% 10% UAE 38% 31% 33% 30% 19% 11% 15% 11% UK 34% 33% 29% 28% 9% 9% 8% 4% All 41% 37% 37% 32% 17% 11% 10% 7% 1st choice 2nd choice 3rd choice Source: PwC commissioned survey (2017) 17
? X Not applicable – I would not be willing to use an advanced computer/ robot with Monitor me and artificial and provide general intelligence for ge a Set a broken care/offer any type of cut or bone and put it advice during Deliver my procedure/ into a cast pregnancy baby Other Don’t know service 9% 5% 8% 2% 1% 10% 19% 8% 6% 3% 0% 0% 11% 25% 1% 10% 0% 0% 1% 14% 6% 9% 4% 4% 1% 1% 12% 20% 7% 12% 9% 3% 1% 1% 2% 6% 3% 4% 0% 1% 9% 19% 9% 7% 1% 1% 0% 7% 12% 1% 9% 11% 2% 2% 2% 7% 8% 4% 1% 1% 1% 10% 22% 3% 10% 11% 4% 1% 3% 8% 5% 11% 5% 2% 0% 11% 8% 8% 4% 3% 1% 0% 9% 31% 0% 7% 5% 1% 1% 10% 20% 18
Minor, minimally-invasive surgery (e.g. cataract surgery or lase eye surgery etc.) 73% Nigeria 25% Will robots replace Minor surgery Turkey Major surgery 66% surgeons in the 28% Nigeria, Turkey and South Africa were the Even so, Nigeria, Turkey and South operating theatre? most willing to undergo minor surgery 62willing % South Africa Africa were35 far more % (69%, performed by robots (73%, 66% and 62% 60% and 51% respectively) than In the three decades since robots respectively), with the UK the least unwilling to undergo55%major surgery first began performing surgery, the willing (36%). In Belgium, the UK, Qatar performed by 38 a robot. % Again, the exponential advances in robotic Germany and Sweden the unwilling UK was the least willing 55% of all the technology are enhancing a surgeon’s respondents outnumbered the Saudi Arabia willing, 32% (27%) and Belgium yet countries surveyed ability to see, feel and do. Autonomous those countries still have over one-third of (60%), Germany (59%), the UK (58%), surgery performed by robots, replacing 50% the public still willing to undergo suchUAE Sweden (55%), the Netherlands (53%) the surgeon, is likely to further the a procedure. 38 % and Norway (51%) were all much more effectiveness, safety, consistency, and Figure 9: Percentage unwilling than47 willing/unwilling willing % to to have minor surge undergo accessibility of surgical techniques. For Netherlands 46 Perhaps unsurprisingly, the performed by a robot instead of a doctora robot. situation major surgery % performed by example, experimental studies have changed dramatically when it came to Minor, 46% already shown autonomous robots major surgery. We defined major minimally-invasive Norway surgery surgery 42%(e.g. cataract surgery or lase to perform higher quality suturing as invasive surgery – such eye surgery etc.) as replacement than surgeons. 43% of a knee or hip joint, removal ofGermany a tumour, 47% or heart surgery – and found much higher 9: Percentage willing/unwilling to have minor surgery Are consumers ready? 43% 73% minimally-invasive surgery (e.g. cataract surgery or laser unwillingness for a robot to perform a Sweden Nigeria major procedure instead of a doctor. 25 % 47% We asked our survey participants if they 39% 66% would be willing for a robot to perform a Belgium Turkey Figure 9: Percentage willing/unwilling to have minor 28 % 53 surgery % performed by a minor or major surgical procedure instead Net: Unwilling robot instead of a doctor 36% 62% of a doctor if studies showed that they South AfricaUK 49% or laser eye could do it better than a doctor (e.g. more 35 % Minor, minimally-invasive surgery (e.g. cataract surgery quickly, more accurately, with a faster surgery etc.) 55% Qatar Net: Willing % Net: Unwilling 0 10 20 30 40 50 60 70 80 90 100 recovery time). 0 10 20 30 40 3850 60 70 80 90 100 performed by a robot instead of a doctor 73% 55% 66% Source: PwC survey 62% • We defined a minor surgical Saudi Arabia 55% 55% 32% 53% 50% 49% procedure as a non-invasive or 47% 46% 47% 47% 46% 43% 43% 50% 39% minimally-invasive surgery, such 42% 38% 38% 36% UAE 35% 38% 32% Net: Willing as cataract surgery or laser eye 28% 25% surgery, and found that, overall, Netherlands 47% close to half and up to 73% of all 46% respondents would be willing to 46% undergo minor surgery performed Norway 42% by a robot instead of a doctor. 43% eye surgery etc.) personal preferences and health Germany Nigeria Turkey South Africa Qatar Saudi Arabia UAE Netherlands Norway Germany Sweden Belgium UK records (37%). 47% surgery • Take and test a blood sample and 43% remove Sweden 47% provide me with the results (32%). Figure Minor, 39% Figure 10: Percentage willing/unwilling to have major Belgium 53% Major, invasive surgery (e.g. replace a knee or hip joint, Figure 10: P ercentage willing/unwilling to have minor surgery performed by a 90 100Unwilling robot instead of a doctor 36% UK 49% ajor, invasive surgery (e.g. replace a knee or hip joint, remove a tumor, M heart surgery) Net: Willing Net: Unwilling performed by a robot instead of a doctor 40 50 60 70 80 Net: 0 10 20 30 40 50 60 70 80 90 100 69% 60% Source: PwC survey 60% 59% 58% 55% 53% 51% 51% 45% 45% 45% 44% 44% 44% 41% 40% 35% 35% 32% 34% 20 30Willing 30% 29% 27% 0 10 Net: a tumour, heart surgery) Nigeria Turkey South Africa Qatar Saudi Arabia UAE Netherlands Sweden Norway Belgium Germany UK 19
You can also read