In late October, the Milken Institute held its annual Future of Health Summit. The event is an opportunity for those in all areas of healthcare – scientists, pharmaceutical companies, policy makers and technology providers – to come together and discuss what’s on the horizon for medicine, biotech and patient care.
CAS President Manuel Guzman spoke on a panel highlighting the critical role of data investments and the power of partnerships to accelerate innovation in healthcare. CAS was proud to also host a roundtable discussion at this year’s event, focusing on how AI technology can help accelerate R&D to find cures more quickly. I had the pleasure of hearing from colleagues across the healthcare industry about the challenges and opportunities they see in implementing AI technology, and their wider perspectives on this crucial area of innovation.
The AI hype curve
Whenever a technology becomes especially popular or trendy, stakeholders begin to wonder if the conversation is driven just by hype? Our discussion kicked off with attendees’ thoughts on this “hype curve” around AI in healthcare, as well as the possibility of an impending “AI winter” – the idea that the promise of AI will have outweighed the actual benefits and the technology will be abandoned.
The panel participants all agreed that healthcare tends to lag behind most other industries when adopting new technology and that there will always be peaking and troughing when embarking on technological innovation. Participants pointed out that the financial sector has been successfully utilizing AI solutions for years, but these changes still need to be brought to healthcare. The consensus was that the idea of an AI winter happening purely because healthcare hasn’t caught up yet is likely premature.
One reason AI is expected to bring its promise to fruition in biotech R&D is because there is exponentially more data available to inform algorithms. For example, at CAS, it took 40 years to register the first 25 million chemical substances in the CAS REGISTRYSM database. However, over the last 4 years we have registered more than 50 million molecules. We’ve seen explosive growth in the amount of scientific information available, and with that growth comes an increased need for programs and people that can sort through it and make it useful for scientific innovation.
The importance of a diverse team
Any organization can try to implement an AI solution, but an important aspect that many overlook is the human element involved in the process. As our discussion addressed, in the field of biopharma R&D, teams should comprise diverse perspectives and areas of expertise, including data scientists, clinicians and patient advocates working together to provide a holistic view of the information.
The reason this team composition is so important is because of the context diverse expertise brings to the data that is driving AI solutions. For example, participants in the discussion described diagnostic tools using AI, and noted that data scientists realize this requires pattern recognition and signal detection in a successful program. The attendees explained this need for taxonomic discussions with data scientists when designing AI solutions for R&D to ensure these processes are in place. They also agreed that this type of collaboration and team diversity can unlock siloes in R&D organizations and bring many voices together to build better solutions.
Quality in, quality out
Another benefit to leveraging expertise across disciplines is that it can produce the high-quality data needed to drive effective AI solutions. As participants noted during the discussion, financial companies publish quarterly reports and the data is presented in a clean, orderly way. In healthcare, immense amounts of important data are never used because the organizations that hold it don’t have the time or ability to curate it. The National Institutes of Health, for instance, holds terabytes of data about hundreds of diseases, but if it’s not adequately curated, it can’t be entered into an AI program.
By bringing in different expertise, and often through leveraging partner organizations, biotech firms can develop harmonized, clean data repositories that are effective for AI-driven R&D tools. And the benefits could be massive—for example, as our group discussed, only about 10% of clinical trials are successful and finish within budget. These trials are the lifeblood of biotech companies, and AI solutions can increase the percentage of successful, on-budget trials by finding breakthroughs faster or overcoming roadblocks sooner. Participants all agreed that clinical trial design and planning is another area within healthcare where AI has proven to be very effective. With the power of advanced technologies and quality data to drive them, biotech companies can overcome that hype we keep hearing about and begin to see real growth in innovation.
As a scientific information solutions specialist, CAS understands the problem of data quality and is aiming to solve it by 1) partnering with other organizations that have large repositories of data, like universities and pharma companies, to harmonize and improve their data assets and 2) building, curating and managing our own world-class collection of scientific data extracted from the global body of published scientific literature over the past 100 years.
The value of these resources was made abundantly clear at our Milken Future of Health discussion, and we’re excited to be a part of this evolution in the biopharma R&D industry. To find out more about how CAS is creating a path forward for healthcare AI, read our recent case study that quantifies the impact of high quality data on machine learning outcomes.