Credit: Google News
Healthcare, growing to $10 Trillion US globally, lags behind nearly 80% of other industries in innovative technology adoption, and in fact is at the bottom five across industries. However this is about to change in the next three years. From the WEF Future of Jobs Report, by 2022, adoption forecasts include: 87% big data, 87% biotech, 80% machine learning (a form of AI), 73% wearable technologies, 67% each for blockchain / IoT / AR-VR, 53% 3D printing, 47% stationary robots, and 40% non-human robots.
This became a focused area of discussion at the W2O hosted session in March at SxSW “EQ in the ER: Building Empathy Through XR” which is now available on video and follow-up podcast. Anita Bose, Chief Business Development Officer, W2O Group moderated a highly interactive dialogue with Rasu Shrestha MD MBA Chief Strategy Officer and EVP Atrium Health and myself. We explored: what are the opportunities, challenges, investments, frontier examples and issues in healthcare such as ethics, genetic editing, big data, extended reality, AI and machine learning, biotechnology, wearables, blockchain, IoT, robots, quantum computing, 3D printing, empathy in the medical context, opioid management, aging, and much more.
Dr. Shrestha provided a multi-faceted and nuanced discussion on how enabling technology can be in providing enhanced caring and empathy (deep EQ / emotional quotient) in healthcare with specific examples from his considerable insights in the medical system. This addressed the title about EQ in the ER: Building Empathy Through XR. EQ meaning emotional quotient and ER equating to the emergency room however in all contexts during the SXSW session.
Getting into more specifics. XR (extended reality) is the umbrella term for Virtual Reality (AR) + Augmented Reality (VR) + Mixed Reality (MR). The XR marketplace is about $30B US for 2018 and more than $200B in three years. VR are the wearable goggles where you are totally immersed in a virtual (imaginary) environment and available widely in gaming for several years. AR are the games (Pokémon) and graphics tools (Snapchat) you see on most smartphones where you can overlay virtual objects with real objects. An example of MR would be the new Microsoft HoloLens 2 released in this year where the virtual objects are anchored and seamless with physical ones where you can interact directly with both. XR provides for training of patients and medical staff, can be used in pain management and in operating rooms. Moreover, good healthcare can be threatened by bias, where implicit prejudices could be a matter of life or death. But the extended reality (XR) spectrum – AR, VR and MR – can allow us to see beyond our own biases and foster empathy with people who are different than ourselves.
What are the opportunities to use other extended reality technology within the medical arena and what do we hope to gain from those uses? I posed this question to Prof. Dr.-Ing. Thomas Wiegand, Executive Director, Fraunhofer HHI, Information Technology, TU Berlin. Thomas states: “The use of virtual reality as a treatment method for stroke patients is something that we are currently investigating in a project together with the Max Planck Institute for Human Cognitive and Brain Sciences in Leipzig, Germany.”
The largest impact however is from big data and machine learning (a form of artificial intelligence or AI). This sentiment is supported by experts globally.
Dr. Niranjan Kissoon (VP Medical Affairs, BC Children’s Hospital in Canada) states, “We are now understanding the potential to harness big data and use computing learning and modeling in many areas in medicine. For instance, simulation of scenarios for problem-solving in critical care is gaining traction as well as our ability to build algorithms and models for many complex processes. The future has limitless possibilities. I look forward to a brave new world where (AI) technology can save healthcare providers time to show empathy to their patients while providing clear data-driven evidence of the optimal individualized pathway to health.”
Anita Bose, who hosted the SXSW session and as Chief Business Development Officer of W2O, commented:
“Digital advances and big data are poised to revolutionize the healthcare industry, an industry that has historically lagged behind. AI and machine learning, in particular, should have a significant impact since medical care is reliant on pattern recognition and predicting what may happen based on those patterns. In a recent Intel survey, 54% of healthcare professionals said they expected widespread AI adoption in the next 5 years and 83% anticipated more accurate diagnostic capabilities through AI. But the question then becomes, how do we accelerate innovation through these exponential advances while still maintaining the human touch that’s required in the effective delivery of personalized health care.”
The top medical journal, The Lancet, published a comment March-end “WHO and ITU establish benchmarking process for artificial intelligence in health.” From the comment, “Growing populations, demographic changes, and a shortage of health practitioners have placed pressures on the health-care sector. In parallel, increasing amounts of digital health data and information have become available. Artificial intelligence (AI) models that learn from these large datasets are in development and have the potential to assist with pattern recognition and classification problems in medicine—for example, early detection, diagnosis, and medical decision making. These advances promise to improve health care for patients and provide much-needed support for medical practitioners.”
This moves healthcare to high-quality evidenced based medicine where figures vary but are astonishing that often clinical recommendations are not based on high-quality evidence and this extends to clinical procedures as well. The application of technology can move the needle in a meaningful way to high-quality evidence.
I then posed these questions to Naomi Lee, Executive Editor, The Lancet. “How are clinicians using AI in the clinical setting? What are additional opportunities to use AI/machine learning in the clinical setting? And what are the expected outcomes of these uses?”
Naomi states: “Currently AI in the clinical setting is still at a very early phase, and most clinical AI is being used in a research setting, rather than in routine clinical care. The potential of AI in health, however, is very exciting, and many feel that it will be AI that transforms healthcare in a way other digital technologies have so far failed to do. Much of medical practice is about pattern recognition for diagnosis and prediction, so AI should present an advantage here over humans alone, particularly now there are large amounts of digital data available. There is also interest in using AI to solve logistic problems in healthcare, for example around patterns of service use, which could realise some of the potential with less risk.”
In asking Naomi this follow-up question, “How do we ensure there are clear guidelines for use and some level of standardization?”. Naomi responded: “There is a clear framework for medical research of a new drug that goes from laboratory work, to a first in human study, through clinical trials phase 1-3, and then post marketing surveillance. However, this framework is challenged for many reasons when we try to assess AI models, and we need to create a new system that builds on this, but addresses the benefits and challenges of AI models in health. The World Health Organisation and International Telecommunications Union have together formed a focus group that will create a benchmarking process for AI models in health. This is a vital first step to understanding which models perform well enough, in order that we can then examine the effect the model’s use has on hard clinical endpoints like survival.”
Prof. Dr.-Ing. Thomas Wiegand added the following to the AI impact in healthcare. “The potential for AI assistance in the health domain and advancing the field of digital health is immense because AI can support medical and public health decision making at reduced costs, everywhere. For example, a significant amount of recent work on AI in health has gone into applications that revolve around image interpretation and natural language understanding. The topics covered include the analysis of X-ray, CT, MRI, digital pathology, cardiac, abdominal, musculoskeletal, foetal, dermatological and retinal images. In language understanding, the areas of biomedical text mining, electronic health record analysis, sentiment analysis on internet-derived data, and medical decision support systems have shown promising results. Other AI research addresses clinical processes, public health topics and prevention. Many of these advances of AI in these areas hold the promise to substantially transform the health space.”
On asking this follow-up question to Thomas, “How do we ensure there are clear guidelines for use and some level of standardization?” He commented: “The potential of AI for health also faces a number of challenges. In particular, deep learning models are famously hard to interpret and explain – which may substantially hinder their acceptance when facing critical or even vital decisions. Hence, interpretability, explainability, and proven robustness (e.g. to outliers and to adversarial attacks) are crucial aspects that have to be considered for trustworthiness. This problem is complicated further because most modern AI applications are based on supervised learning and rely on data that are labeled. In the health domain, labels can typically be given only by qualified specialists. In addition, machine learning approaches must take into account the biases. Hence, in machine learning, algorithms and training data have to be considered in combination. The Focus Group on “Artificial Intelligence for Health” (FG-AI4H) established by ITU in partnership with WHO aims to meet the challenges by providing open, transparent and standardised processes for the evaluation of AI algorithms in the health space.
I added to the exchange with these additional questions to Thomas. “With AI and machine learning, are we actually creating national data sets of medical information? Is that the future? What are the benefits and risks associated with that?” Thomas responded, “Data sets are a necessity for AI methods. Health data are particularly sensitive. Creating and assessing data sets in this space that are of high quality for AI algorithm training and testing is a challenge. The Focus Group on AI for Health that is supported by ITU and WHO which are both United Nations organisations, offers a great opportunity here to provide such data sets.”
In summary, the SXSW session is a catalyst for a global conversation and collaboration to move to implementation of transformational technologies to improve healthcare while still emphasizing the human element with a focus on caring and empathy.
Credit: Google News