Have you heard the news? Artificial intelligence is about to save the world. If you believe the headlines, AI is the latest magic bullet set to conquer the deadly pandemic that stopped the world in its tracks just a few months ago. It’s receiving significant attention from media and online. From startups hosting informational webinars to large pharmaceutical companies speaking at conferences, AI is being touted as a superpower in the fight against Covid-19 – the key to new treatment discovery and development. The problem being – this isn’t a realistic promise.
I have no problem putting resources toward AI to help in the fight against Covid-19. There are places where it can help. It can help us better track the spread of the virus, including contact tracing. It can help speed up diagnosis and keep track of individual symptoms to help us understand the effect of the virus on human biology. These are all big uncertainties even seven months into this pandemic.
Behavioral Health, Interoperability and eConsent: Meeting the Demands of CMS Final Rule Compliance
In a webinar on April 16 at 1pm ET, Aneesh Chopra will moderate a discussion with executives from DocuSign, Velatura, and behavioral health providers on eConsent, health information exchange and compliance with the CMS Final Rule on interoperability.
The most effective solution to Covid-19 will be the wide availability of safe and effective vaccines. AI is not the key to achieving this goal. As a society, we know how to create a vaccine for Covid-19. We know how to test it, manufacture it and distribute it. AI has little to offer when the challenge has little to do with scientific uncertainty and more to do with the mobilization of resources. AI does its best work where we don’t otherwise know how to uncover uncertainties. AI has more meaningful applications where analytical insight is required to make decisions under uncertainty.
But this overhyping of AI is nothing new.
When it comes to healthcare, artificial intelligence has been overhyped and overpromised for the greater part of 20 years. AI is attractive. Most industry experts will agree that AI technologies, such as machine learning, have tremendous potential in the healthcare space. With the power to bring down cost, improve accuracy and save time, it’s clear that these tools should and must be utilized. But they’re just that – tools. And overpromising can have unintended consequences.
I recall in 2017 when a string of Alzheimer’s failed clinical trials, focused on beta amyloid and tau proteins, really set back our progress in finding a treatment and cure, despite sound science. A slew of companies used this as an opportunity to apply broad AI to find treatments, which of course was exciting for some. A lot of attention and funding was placed on tech companies that lacked true drug discovery expertise. As a result, nothing of value has been delivered.
A Deep-dive Into Specialty Pharma
A specialty drug is a class of prescription medications used to treat complex, chronic or rare medical conditions. Although this classification was originally intended to define the treatment of rare, also termed “orphan” diseases, affecting fewer than 200,000 people in the US, more recently, specialty drugs have emerged as the cornerstone of treatment for chronic and complex diseases such as cancer, autoimmune conditions, diabetes, hepatitis C, and HIV/AIDS.
This is because there is a lack of education on, and true understanding of, what AI can achieve. That poses a danger to progress, especially in a disease area where we need to focus all of our attention on the right spots. Right now, that is Covid-19 and the AI hype is playing out all over again. We cannot afford a repeated cycle where AI is touted as the solution and never delivers because the companies applying realistic and purposeful efforts will lose credibility and funding.
That is a very real, albeit unintended, consequence. As is the halted cascade of innovation that should have been driving new drug discovery for hard-to-treat therapeutic areas. Artificial intelligence can, and is, being used to aid in efforts where it truly matters – making predictions using real world data, identifying new target biomarkers – but only when we use it properly as a focused tool. Only then can we recognize the impact and translate those insights to the broader picture.
When the project calls for a carpenter, you don’t simply hand someone a hammer and wish them luck. Further, if we measure the potential of that hammer on its ability to build an entire house, we will lose faith in its effectiveness. The same must be applied to artificial intelligence. We need to look at AI for what it can accomplish in its role. That has to start with educating the industry on what is possible with AI and direct efforts appropriately. For example, AI can be applied to the research and development process, including the identification of biomarkers and more quickly identifying and screening potential drug targets or to help with patient selection or clinical trial results analysis.
The global nature of this pandemic brings an unprecedented level of attention and, with it, misinformation. When discussing the role of AI in Covid-19, we must focus on what is achievable. We have to reframe this conversation and look at AI as an important approach that can provide better analysis and ultimately better outcomes for patients, not as replacing the entirety of the drug discovery process or as a magic bullet for a vaccine. By approaching the problems we’re trying to address in a focused manner, we plunge a ripple effect through the entire research and development process.
The truth is that we desperately need an approved vaccine and treatments for Covid-19 and countless other diseases like Alzheimer’s, glioblastoma and lupus. But the next time you read an article that promises AI to rapidly solve a healthcare problem, I encourage you to educate yourself on not just the potential, but the limitations. Overestimating AI’s capabilities in the complex world of drug discovery and development could result in an abandonment of its proper use, and worse, may cause us to miss out on exciting new targets and medicines that make a difference to patients.
In summary, the proper use of AI in medicine is best applied when there is scientific uncertainty. We can use AI in early-stage drug development as a tool to improve hit predictions, giving us a better understanding of whether a set of compounds being tested will have a desired clinical effect. This is an area of high uncertainty and we’ve seen the ripple effect that AI can have with this focus. We know AI’s potential for this specific application, but we also know that complex diseases require a team of drug discovery and development experts working in collaboration to discover breakthroughs. And there’s no uncertainty about that.
Photo: Blue Planet Studio, Getty Images
