Report: The Future of Healthcare
Publication Date: 2020
Mark O'Herlihy, Managing Director for EMEA of IBM Watson Health
Mark O'Herlihy explains how and why machine learning will be fundamental to the healthcare of tomorrow, admitting that there is still a great deal of work to be done to demystify the role of new technologies by identifying proven results and best practice.
Could you please start by introducing IBM Watson Health and the fundamental mission behind the company?
IBM Watson Health is committed to building smarter health ecosystems in every corner of the globe. We were created in 2015 for the purpose of enabling healthcare payers, providers and researchers to more effectively respond to the world's most pressing health challenges through the use of advanced data, analytics and enterprise solutions. For more than 100 years, IBM has been transforming industries. As such, our roots are steeped in research – leading with science – to address the world’s most challenging problems.
Watson Health leverages deep industry expertise, data/analytics and actionable insights – all underpinned by security and trust – to deliver solutions that augment human expertise and improve clinical and operational workflows at the enterprise level, allowing government health and human services, payers, providers, and life sciences enterprises to focus on what they do best.
While we work with many different kinds of enterprises, our basic principle always remains the same: to harness and apply the true power of digital technology to enable patients to live longer, healthier and happier lives. We are relentlessly focused on generating ‘actionable insights’ that will tangibly result in improved end outcomes for patients.
What is the added value to be derived from digitalising the healthcare space and plugging in technologies such as artificial intelligence?
Much of the added value involves “the need for speed.” Irrespective of the stakeholder – whether it be a global pharmaceutical firm, healthcare provider or payer – they all need faster insights from data.
Technologies such as Artificial Intelligence (AI) often deliver new layers of visibility that can help condense the timeframe needed to navigate a drug through clinical trials or help inform the process of putting together a treatment plan. Doing so can increase the likelihood that practitioners select the optimal treatment pathway for a particular patient from the beginning, without the risk of multiple false starts.
We have a great working example from the university hospital in Geneva, Hôpitaux Universitaires de Genève (“HUG”). HUG recently agreed to implement Watson for Genomics, becoming the first university hospital in Europe to deploy the platform. Analysing a panel of the genome previously took more than 160 hours to complete manually; now, HUG can use Watson for Genomics to do it in a mere 10 minutes.
Beyond delivering swifter positive health outcomes, these technologies can go a long way in promoting health equity. Some of the biggest issues in health relate to an unwarranted variation in care, including a lack of standardization in determining prognosis and setting treatment plans. By using data-driven insights, we can help overcome this challenge.
Can you give us some concrete examples of the latter?
IBM Watson Health has been working to help to standardise the treatment of lung cancer. Specifically, we provide a system that augments human decision-making by enabling practitioners to track the latest intelligence on prognosis and treatment pathways, as the practitioner is likely unable to remain totally current on the latest research given the vast amount of research that is now published.
Realistically, the reference point for these practitioners will be centred around what they learned during their medical studies. At IBM Watson Health, we help bridge this gap by providing practitioners with evidence-backed insights – this additional information may result in practitioners modifying their decisions.
How do you actually achieve this?
Watson for Genomics scales laboratories’ precision oncology programmes, leveraging AI to extract unstructured data from peer-reviewed literature. These laboratories then use this data to continually grow their knowledge base by providing variant information and up-to-date clinical content, meaning content that is based on the latest approved therapeutic options, professional guidelines, biomarker-based clinical trial options and genomic databases.
For example, oncologists at the University of North Carolina’s Lineberger Comprehensive Cancer Centre tested Watson for Genomics on over 1,000 retrospective patient cases. In more than 300 instances, our tool identified additional potential therapeutic options. Of these, 96 were not previously identified as having an actionable mutation, and these findings were reported to treating physicians. These are precisely the sorts of data insights that can help augment a molecular tumor board process.
While digitalisation has been swiftly embraced by some industries such as FinTech, its application to healthcare seems to have been taking place more slowly than many imagined. Why might this be?
Our understanding of genomics is considerably more advanced than it was two decades ago. I have been in the healthcare and life sciences space for nearly 30 years, spending 15 years in clinical practice with a focus on radiology and cardiology, and many of the problems we faced when I entered the industry still persist today. However, nowadays we have a tsunami of data to help us, combined with the technology to accelerate the process of interpreting this information. Ultimately, this can generate vastly improved insights.
The urgency of needing to adopt and mainstream these technologies is increasing: as public healthcare budgets become overstretched, so does the burden of disease. Cancer is a prime example, as there were 17 million cancer diagnoses globally in 2018.
As a result, the healthcare community is being compelled to rethink the paradigm despite its ingrained caution and conservatism. Business-as-usual is no longer a viable option.
The issue is particularly stark in Europe, where there is limited new money to allocate towards health expenditure. We collectively need to seek out efficiency gains so we can achieve more with less, or ‘addition through subtraction.’ If we are to work effectively within the existing envelope, processes will need to be rationalised and optimised – digital technologies are some of the primary tools with which to do so and unlock this hidden value.
What is your assessment of the state of preparedness of the healthcare sector with regard to the inevitable arrival of disruptive digital technologies?
The healthcare sector is intensely regulated and directly connected to wellbeing. So, it is only natural that we should exhibit an ingrained conservatism and scepticism of technologies such as artificial intelligence. Moreover, healthcare is a highly emotive issue, particularly around data security and privacy of medical records.
This is why we go to great pains to win over the hearts and minds of the medical community when it comes to technology. We accomplish this by demonstrating the material benefits to patients from using algorithmic software, big data and machine learning.
We are always quick to mitigate fears about the future uses of machines. IBM and Watson Health are focused on augmenting human activity – not replacing it. Therefore, far from seeking to substitute physicians for machines, we are actually trying to empower medical professionals by optimising their workflow to support enhanced decision-making.
Much of the healthcare industry’s reticence can ultimately be linked to a lack of understanding. There is still a great deal of work to be done on our part to demystify the role of artificial intelligence and machine learning in healthcare by identifying practical, proven results and clear-cut best practices. We are transparent with the use of data and believe a client’s data should remain its own.
The sector is becoming more digitally savvy with the rise of the ‘data scientist’ and the increasing inclusion by many firms of a ‘chief digital officer’ (CDO) position within the c-suite. However, there is a great need for more professionals within the industry who possess both a medical background combined with competencies in software development, advanced mathematics, quantum physics and related disciplines.
It is great to see that many organisations now have a ‘chief information officer’ (CIO). However, the nature of that role needs to evolve with the times to ensure that those occupying those roles have familiarity with digital infrastructure, as well as additional hard-core competencies in data science.
When you engage with clients, how do you go about managing cultural differences in how companies interact with technology?
We completely realise companies cannot leap directly from paper to AI. Additionally, many clients don’t understand their data environment or how rich the information they already have, or could gather, would be. Ultimately, we seek to use data and machine learning to augment physicians’ workflows and decision-making processes –to empower physicians to make better, more informed decisions about their patients care.
So, as a first step, we take stock of their data accuracy and determine what system integrations and extra fields will be required so the data can be rendered actionable. Then, we frame the journey of what will be needed to get there.
Next, we educate the companies about the importance of maintaining good data hygiene and the potential it has to strengthen these companies. This needs to be executed using a step-wise, sequential approach – there is little point in trying to run before you can walk.
Right now, less than 10% of the healthcare data worldwide is actually analysed. There is a very long way to go, but just imagine what benefits we could reap merely by raising that level a bit more?
IBM Watson Health has been a real trailblazer in pioneering the deployment of AI in support of evidence-based medicine for health care. What have been the primary lessons learned and challenges encountered?
During our four years of existence, we have been systematically fine-tuning and readjusting our approach. Initially, we trained Watson by analysing millions of CT scans and cross-checking the images. Then, we realised that wasn’t exactly what the medical community needed most; instead, it was better to focus on alleviating specific bottlenecks.
If we take the example of breast screening, a radiologist will typically read 30,000 mammograms a year, out of which a high percent is normal. Only a very small percent exhibit anomalies, meaning a great deal of their time is spent trying to locate the abnormals. Watson could be supportive by building an algorithm that pre-reads the mammograms and prioritises all the clearly abnormal ones, redirecting a radiologist’s workload to where it is most needed, and helping patients requiring treatment to gain access earlier.
How do you see the future of algorithmic medicine and machine learning-backed precision medicine progressing?
Right now, we are witnessing a proliferation of digital health start-ups and AI firms offering solutions in the healthcare space.
In many instances, AI experts are already providing point solutions in hospitals. But my concern is that many are delivering very discreet insights based on a small portion of the population. Where the deployment of AI will really take off is when a critical mass is attained that allows for rigorous appraisal of scientific evidence available. The widespread implementation of proper data hygiene will also be essential, as data represents the lifeblood upon which our systems run.
I believe we've already reached a juncture, both in technology and medicine, where digital tools can help abbreviate timeframes and inform clinician decision-making. What comes next will be even more exciting. The future of healthcare will be about treating mutations well before they become an issue, and physicians will be able to manage diseases such as cancer that today are considered killers.
Cancer is fundamentally a DNA issue, not just something that happens by chance overnight. Therefore, our predictive capabilities play a crucial role in countering this disease. The days when a patient’s stage-four cancer is only discovered during an unrelated surgery absolutely have to stop. Meanwhile, the digital revolution will also be a handmaiden to the rise of phenomics, whereby better understanding of the unique metabolic signature of individuals will enable improved prediction, prevention and minimally invasive treatment.
Do you, perhaps, have a final message for our international readers?
The healthcare space is already ripe for digital disruption. Only in the world of health do we have a paradox in which more data than ever is flowing through physicians’ hands, but the true value of that data has gone largely untapped because it is unstructured and silo-ed in systems that don’t talk to one another. In healthcare, consumers and patients have more service options than ever, yet very little visibility into what those services cost due to a misalignment of incentives between various stakeholders.
IBM Watson Health is committed to building smarter health ecosystems with simpler processes, better care, faster breakthroughs, and improved patient experiences. We have the essential capabilities necessary to help our clients drive their digital transformations: deep industry expertise, data and analytics, and actionable insights, all underpinned by security and trust.