No matter what industry you’re in, artificial intelligence (AI) is all the rage. It’s the shiniest of the shiny and new, and it’s everywhere.
In pop culture alone, it’s the central theme of HBO’s Westworld, where humanoid AI robots pretend to be people, or even the most recent season of Silicon Valley where a major character was an AI-powered robot named Fiona.
AI is also the central, recurring theme at every conference. Even at giant tradeshows like the Consumer Electronics Show (CES) where, this year, we saw autonomous vehicles, voice-enabled bot driving assistants within cars, L’Oreal’s thumbnail-sized UV sensor patch, and hundreds of other AI-enabled “smart” products. At South by Southwest (SXSW), it seemed every other session was about AI. Elon Musk himself, CEO of SpaceX and Tesla, even made a surprise appearance where he fielded questions from the audience and warned against the irresponsible development of AI and the requirement to ‘work safely’ when exploring it.
AI is omnipresent and it’s no shock that it’s even crept its way into the highly regulated pharmaceutical industry, one that is usually risk averse and slow to adapt new technologies. And with increasing R&D costs and healthcare costs in general, paired with larger and more precise data sets, AI may alleviate multiple pain-points across the industry. But will we adapt quickly enough?
"Machine learning (ML) is making the drug discovery process cheaper, faster, and more optimal for all involved"
Pharma’s Recent History of Emerging Tech Adoption
Look at the adoption of social media, for example. A few years ago, social media had become the norm and a crucial tactic for brand engagement in every other industry. Customers were taking brand complaints, praise, and discussions to social, but pharma had no presence. The conversation was happening with or without us, and we had a choice to make – we could either meet our customers where they wanted to engage, or miss out. Clearly we needed to be there. Yet pharma was bound by requirements to provide fair balance, privacy and safety information among all branded promotional materials.
In 2014 the Food and Drug Administration (FDA) came out with draft guidelines entitled Guidance for Industry Internet/ Social Media Platforms with Character Space Limitations— Presenting Risk and Benefit Information for Prescription Drugs and Medical Devices, and everything changed. Today there are hundreds of branded pharmaceutical product pages across social channels like Facebook, Instagram, Twitter, and more, many with open comments. Pharma’s even ventured into Snapchat! These channels allow for precise targeting, but more importantly for opportunities to compliantly engage with customers for customer service purposes, gain invaluable brand and behavioral insights, and provide condition support communities.
So how is regulation affecting the pharmaceutical industry’s adoption of AI? What are the use cases within pharma for AI? With social media, it became clear that pharma had to either adapt or miss crucial patient and customer engagements. AI is no different.
AI is Already Happening in Pharma
Despite being slow to adopt other technologies, we’re already seeing AI come to life across the pharmaceutical industry. Of course, there are dozens of components within the industry where AI is applicable, including therapy discovery, product approval, commercialization, clinical trials, FDA submission strategies, product launch execution, pricing, supply chain management, market penetration and building, awareness, product adherence, clinical development and trials for new indications, and submission in other markets, to name a few.
It would take years to understand use cases of AI for every single aspect of this complex industry, but the common theme is simple: Driven by precise data sets, AI will shorten the amount of time it takes to solve business problems and meet objectives. Let’s explore some of the potential use cases.
Improving Drug Discovery
First, AI has the potential to find new therapies. Machine learning (ML) is making the drug discovery process cheaper, faster, and more optimal for all involved. Startups like Berg and Benevolent Bio have each developed their own AI platforms to analyze obscenely large amounts of biological and clinical data in order to discover new cancer and neurological therapies. Additionally, modern predictive analysis technologies have the potential to improve drug pipelines through computer simulations
Data-Driven & Precisely Personalized Treatment Plans
Highly personalized treatment plans are also on the horizon due to advances in AI and remote patient monitoring. Last year, AiveCor’s Kardia band became the first FDA-cleared Apple Watch band, upgrading the Apple Watch to a medical device, AliveCor, recently named the number one Most Innovative Company in AI by Fast Company, “enables patients and their care teams to easily, quickly and inexpensively detect and manage possible abnormal heart rhythms.” The band functions as an electrocardiogram machine and is 84 percent accurate at detecting one’s normal heartbeat from a trial fibrillation, which can cause stroke. A cardiologist can now remotely monitor a patient versus seeing them once a year, allowing the physician to create precise treatment plans, which in turn affect pharmaceutical sales volumes and provide anonymized patient insights leading to more relevant therapy options.
IBM Watson is also at the forefront of AI, optimizing patient treatment options based on medical history and information. Remote patient monitoring, better data flow, and predictive analyses are all allowing for optimized treatment plans and better outcomes.
Staying on Therapy with AI
Drug adherence is another area where AI is improving patient outcomes. AiCure’s intelligent medical assistant uses a HIPAA compliant visual recognition platform to track patient therapy use. The product provides visual dose confirmation, interactive patient support, and visual diagnostic capabilities. A more basic but effective example is that many manufacturers are launching SMS interventions and ML enabled chatbots to adapt to patient needs, providing dosing day reminders and refill reminders, thus improving product adherence.
Re-Thinking Traditional Governance Models
Of course, before pharma can truly embrace AI in a mainstream way, we need to understand the regulatory considerations and potential hurdles. Data security, patient privacy, accuracy, and lack of infrastructure are just a few initial concerns. Regardless, AI is here and the industry must get comfortable with being uncomfortable. And from a regulatory body standpoint, we’re off to a decent start. Because AI can increase datasets of relevant information that directly support the FDA’s goals, cross-functional groups such as the Digital Health Unit have been created to understand AI’s possibilities and obstacles. The Unit is comprised of AI experts, healthcare industry members, and members of the FDA.
At the end of the day, AI has boundless potential to solve business problems and meet objectives within pharmaceutical organizations, thus improving patient outcomes in a much shorter time span. But the most important thing to remember is to not be distracted by the shininess and newness of AI. Always ask yourself: What business problem am I trying to solve, what outcome am I looking to achieve, and will AI get me there faster?
More importantly, is this what’s best for the patient? If the answer is yes, you’re at an optimal starting point. You will also need to get used to new models internally if you’re going to succeed. Consider a new approach to digital and technology governance internally that embraces agile approaches, testing ideas, and unlikely partnerships such as AI startups.