Joe Corrigan, head of smart healthcare at Cambridge Consultants, part of Capgemini Invent, examines the “collision” of the medical and technological worlds.
With the inevitable slowness of tectonic plates, the medical and technological worlds collide. From a distance, it seems that nothing is happening. But the ground is rising. New ways of working are made possible by artificial intelligence (AI), data and edge computing. Doctors, patients and researchers are – like all of us – immersed in digital consumer-centric services, adding to the relentless pressure. This means that when a new user-centric medical service is launched, there can be sudden and seismic changes in behavior. Being able to predict or cause these earthquakes will determine who ends up on top of the mountain.
Let’s start with the worldview from a pharmaceutical point of view. It is characterized by prolonged R&D cycles, ultimately leading to the arrival of successful drugs that have successfully negotiated clinical trials, with their costs constantly increasing. Compare that with technology, the realm of agile development. Test fast, fail fast, learn fast, repeat. User experience is paramount and the ambition to forge personalized customer relationships drives continuous feedback and the domination of social networks within the almighty digital ecosystem. Nonetheless, change is happening – against a backdrop of radical innovation in medical technology and the growing craze for digital adoption accelerated by the pandemic.
More of that in a moment, but first a word on drug discovery. For a while, AI has been at the forefront of transforming this expensive business. In 2018, the journal Nature hinted at the $ 2.6 billion cost of developing a treatment. Much of this is actually lost on the 90% of candidate therapies that fail between Phase 1 trials and the regulatory green light. Today, large biopharmaceutical companies trust AI-led initiatives, while startups use it to identify patterns hidden in large volumes of data. Faster, cheaper and more successful drug discovery is of course the goal.
At Cambridge Consultants, we conducted an AI-based research project on proteins, which are increasingly reused for use in drugs, antibodies, vaccines and viral vectors for cell and gene therapy. The challenge is to alter a protein sequence, subtly altering the structure, to achieve the desired performance. We have successfully applied natural language processing AI models to protein sequences to improve the likelihood of predicting their specific functions. The technique allows efficient optimization of protein performance in a variety of applications, including drug development.
Clinical trials are another area of interest right now. We have created a digital platform that uses AI in the system design of a test to derive new digital endpoints that can take advantage of real-time continuous monitoring. Currently, positive subjective responses to an investigational medical product (IMP) lack credibility without clinical data on biomarkers, which is costly and invasive. But AI can eliminate subjectivity by tracking digital and chemical biomarkers in a non-invasive, continuous, and contextual way and linking them to results. Such a decentralized testing approach enables continuous remote monitoring – reducing clinic visits, improving data quality and ultimately patient outcomes.
AI is also starting to take center stage in patient care – and patient choice. Interface traditional diagnostics with a patient’s electronic health registry (EHR). Earlier this year, our client Ellume’s device became the first fully-connected, over-the-counter COVID-19 test to gain FDA clearance. Antigen test results are shared through the user’s smartphone to provide real-time reports to the EHR, allowing healthcare professionals to optimize treatment.
The choice of the patient and the influence of the consumer are of course essential. Abbott’s FreeStyle Libre System, for example, is designed to free people with diabetes from traditional finger-prick blood sugar monitoring. ‘Why sting when you can scan?’ as the marketing proposition says. For consumers, it’s all about the user-friendly experience. They prefer it and ask for it. For them, usability is king – and the industry must change to allow this trend.
Many players are reacting. They realize that behavioral knowledge and design innovation are becoming increasingly important where clinical adherence is low, such as asthma with a rate below 50%. If adherence has the greatest impact on results, surely creating a service that improves it is better than a better drug? It may sound provocative, but one could argue that the needs of patients are underserved by pharma. Traditionally, once a company has negotiated regulatory approval, the reward comes from as many doctor scripts as possible. Does the company want behavioral and contextual data from the real world? Once the drug has been shown to be effective, it can no longer be approved than it already is.
Nonetheless, as we move forward, it will no longer be enough to simply ship the drug – companies will need to consider how it fits into the patient’s lifestyle and shape a service accordingly. This means coming to terms with the interpretation of real-time data. The ability to develop AI models that continually ingest data from devices to quickly learn and iterate a product is familiar to Amazon, but unknown to pharma. For the moment.
The focus will be on demonstrating results that are not set in stone by clinical trials alone, but are rethought and corrected by consumer behavior. The results are influenced by whether the patient is taking the drug or experiencing any real benefit. The key opinion leaders – doctors and other health care professionals – will be kingmakers here, along with the patients. They may well prefer the service to a competing drug because it takes into account behaviors that improve outcomes.
The enthusiasm of healthcare professionals to explore AI in areas such as imaging or data processing continues to drive digital adoption. There is an extremely important market dynamic at play here. Once an interconnected platform for easily accessible standardized data – and where the user benefits from adding their data to that platform – then everything changes. Network effects dominate and new de facto monopolies appear. Web search algorithms developed relatively slowly until Google used PageRank to exploit metadata created by users when sharing links to create such a network effect.
But advances in AI now mean that when the data, interconnections and interoperability are in place, user-created value could create a ‘Google moment’ much more suddenly and with surprising consequences. This tipping point could very well create a winner in medical data – or at least an Amazon, Strava, or Zoom for several niche areas.
What would be the implications of such a change for medical technology manufacturers and technology developers? The commodification of hardware comes to mind, as evidenced by the Apple vs. Android example. There is a difference between the Apple infrastructure, which derives its value from the vertical integration of hardware and software, and the Android platform which offers compatibility and interoperability but very thin margins. This has driven out key players in the market, including LG which stopped making smartphones in July.
Medical devices are totally different in that for each condition there are unique biomarkers that can be used to obtain user actionable information. And as I mentioned above, biomarker support comes from KOLs in these markets. And when it is possible to demonstrate that this user information brings value to the end user through results, it becomes a truly differentiating opportunity for a manufacturer. Each patient group and subset of that group becomes a market with independent platform value. So, we are likely to see some services such as data sharing and secure storage becoming commoditized, but a clear differentiation and value that comes from the niche applications that are found on these platforms.
In its article – “Six Winning Roles for MedTech to Thrive in the Future of Healthcare” – Deloitte identifies its most influential trends: data sharing, interoperable data, access, consumer empowerment, behavior change and scientific breakthrough. Most of them are fueling the tipping point fire of the data transformation capabilities I’ve described. The coming advancements in AI, technology and interoperability will enable the next evolutionary step… the great shift to a data structure that enables the network effect, the “Google moment” and the dominance of the. Marlet. It is time to listen to the tremors and act on them.