Would you take a drug created with AI?

How would you feel if a pharmacist offered you medication developed with the assistance of artificial intelligence?

Apprehensive? You’re not alone. According to studies by QBE and Customertimes, more than half (57%) of the UK population and a third of the US population feel unsure about the concept. Meanwhile, 15% of the US remains ethically opposed.

However, developing a single drug currently takes 10-15 years and costs pharmaceutical companies up to $2.8 billion. During this time, 80-90% of candidates fail in the clinic. As such, there is much desire to speed up the process.

McKinsey has estimated that enlisting AI’s help could generate between $60-$110 billion a year in economic value for the pharmaceutical and medical-product industries. That’s largely because AI can boost productivity by accelerating the process of identifying new drugs and speeding up development and approval.

According to the data analytics firm Elsevier, 93% of researchers and clinicians believe that AI will lead to cost savings for businesses. Nevertheless, even those within the industry approach the technology with caution.

The same study also found that 96% of clinicians and 94% of researchers believe AI will help accelerate knowledge discovery. However, similar numbers are concerned that it will be used for misinformation, and 86% believe that AI has the potential to cause critical errors.

Why are people apprehensive?

Public hesitation about drugs produced at a faster rate may come as no surprise after some of the reactions to the “quick” development of COVID-19 vaccines. Scientific articles conclude that a lack of credible information about the vaccines contributed to concern.

Tim Galloway, portfolio manager from QBE — which conducted the UK study on public perception in AI — believes “people are generally concerned because there’s a lack of trust on AI.”

Mirit Eldor, managing director of Life Sciences Solutions at Elsevier, agrees that a lack of transparency in AI development may be of concern.

“There’s a lot of concern around accuracy and validating what’s right,” says Eldor. “There’s no transparency of where the answer originates, what the source is, if this is a prediction, and what it’s based on.”

“That’s particularly important in regulated industries like life sciences.”

Working in the life science portfolio at QBE, Galloway explains that the insurance firm conducted the study to assess potential risks and concerns regarding the use of AI in life sciences.

“Major companies using AI could be significant for us if something goes wrong. We wanted to understand exposures and how we can manage them as an insurance company,” explains Galloway.

One issue the industry is looking to prepare for is an increase in data protection.

“Data breaches are a real threat,” says Galloway. “People are concerned about their own health records.”

“Our cyber underwriters are seeing growth due to these concerns. People are more aware of their data privacy, and as AI progresses, ensuring data security must be a priority,” he explains.

Slow off the mark

On hesitation regarding the actual drugs, Customertimes’ managing partner Max Votek says, “Caution is typical in the adoption of any technology, including AI.”

“We are at the infancy stage of AI progression, so I truly believe that we are very far from wide adoption.”

On the contrary, the CEO and founder of Unlearn, a firm using generative models to advance clinical trials, finds that most patients are open to the idea: “With clinical trials, patients want to get access to experimental therapies. That’s why they’re participating in the first place, typically.”

Yet, he says the firm encounters more scepticism from the pharmaceutical industry itself.

“It’s really technophobic. It’s near the slowest to adopt new technologies of almost any industry that exists,” Fisher says. “People are still faxing clinical records around as part of clinical trials, for instance.”

He says that regulators and patients are the most supportive, yet pharmaceutical companies are more conservative because computer simulations, machine learning, and AI are unfamiliar to those in the industry.

According to Elsevier, just over a third of researchers and clinicians from the US, China, and India have actively used AI for work-related purposes. Only 28% of US respondents feel positive about AI’s future impact on their area of work.

The true benefits of AI in pharma

While Unlearn’s founding principle is to use generative models to solve scientific problems, the San Francisco-based firm — founded in 2017 — currently focuses on using the tech in clinical trials.

It uses digital twin technology to discover what drug will work best on Alzheimer’s patients.

“In clinical trials, the goal is to compare what happens to a patient if they receive a treatment versus a placebo,” explains Fisher.

“Using a digital twin, we can forecast what would happen if a patient received a placebo and compare it to the actual outcome when they receive the treatment.”

This allows Unlearn to estimate drug effectiveness individually and reduce the need for large placebo groups.

For example, instead of recruiting 2,000 patients with 1,000 in the placebo group, you could have 1,500 patients with only 500 in the placebo group — and when a clinical trial can cost up to $100,000 per patient, this means significant savings, says Fisher.

“We can then estimate drug effectiveness are for each person using the digital twin. This reduces costs and accelerates the trial process.”

Most of the public and those in the industry are aligned, believing that AI working in drug design will help reduce costs.

56% of those surveyed by Customertimes think AI-designed drugs will be cheaper than human-designed drugs. For nearly half of Americans facing unmanageable medical bills and worrying about unexpected healthcare expenses, there is hope.

Similarly, the drug-discovery firm Vivan Therapeutics is using AI to help tailor treatments to individual patients, with the aim of improving outcomes and reducing side effects.

“Our goal is to build trust by demonstrating the value and safety of AI technologies,” says Vivan Therapeutics CEO Laura Towart.

The firm is based in White City Place in London, owned by investment firm Stanhope PLC.

“White City Place is where groundbreaking ideas are transformed into life-changing solutions, like Vivan’s use of AI to develop personalised medical treatment in oncology,” says Liam Le Roux, asset management director of Stanhope PLC.

Vivan uses the technology to model the complexity of a genetic disease, with an existing focus on cancer patients.

An oncology patient will get a tumour biopsy and blood sample, and Vivan will coordinate their sequencing and identify the mutations or genetic alterations driving a patient’s tumour.

It will then be engineered into a fruit fly and developed into a tumour in the same location as the patient’s.

“We test all approved drugs and investigational drugs to identify the right combination that will rescue the fly,” explains the CEO, Towart.

“So we might find a melanoma drug that works for a lung cancer patient, sometimes even incorporating asthma medicine.”

The firm is currently collaborating with the Institute for Cancer Research, led by Paul Workman, a scientist working on the discovery of small-molecule cancer drugs, and pharmaceutical firm IDIBELL’s Albert Antolin, who is developing drugs using big data and AI.

The collaboration is using the fruit fly model to find more effective cancer drugs that are less prone to drug resistance.

As far as patients seeking help, Towart says that the firm has not experienced any hesitation to try drugs recommended by AI.

“I’ve seen several innovative AI solutions in healthcare, and it’s really exciting. Technologies will improve early detection, management, and treatment direction of diseases. If used correctly, AI can significantly improve our way of life and ageing,” says Towart.

Plus, Elsevier offers an AI-driven platform, ‘PharmaPendium,’ used by clinical researchers, toxicologists, and during the early drug development stages to mitigate clinical trial risks based on data from earlier trials.

“AI can help identify potential failures, toxicity, and adverse byproducts earlier in the process,” explains Eldor.

“We established responsible AI principles early on, addressing issues like real-world impact, bias prevention, explainability, human oversight, and data privacy,” she adds. “Fully transparent and ethical practices and subject matter expertise are crucial.”