Teva exec reveals the AI tools pharma really wants

Teva executive VP of global R&D and chief medical officer Dr. Eric A. Hughes  credit: Teva
Teva executive VP of global R&D and chief medical officer Dr. Eric A. Hughes credit: Teva

Fantasies of an AI revolution in drugs have met the reality of the industry's conservatism. But Teva’s head of R&D offers a glimpse into the AI tools his company actually uses, and is looking for. Startups take note.

The idea of artificial intelligence solutions for drug development has been touted by everyone from startups to giants like Nvidia and Alphabet, telling us how AI will radically change the pharma sector. But, as with other sectors, there is a gap between AI's revolutionary promise and what’s actually happening.

Implementing an AI solution at pharmaceutical companies, which are usually big, heavy entities, requires an endless investment of time and money. Moreover, the sector is characterized by conservatism, stemming from both the caution required when dealing with human life, and the burden of regulation.

So, out of all the possibilities, what are pharmaceutical companies really doing today with artificial intelligence?

At the 22nd BIOMED Israel Conference and Exhibition, held about a month ago in Tel Aviv, Dr. Eric A. Hughes, executive VP of global R&D and chief medical officer at Teva Pharmaceutical Industries, revealed not only his dream for implementing AI at the company, but also what Teva is already doing in practice today with these capabilities. Hughes made his remarks during the session, "Is AI revolutionizing drug discovery and clinical development?" led by Dr. Anat Cohen Dayag, president and CEO of Compugen, and Dr. Dan Goldstaub, scientific co-founder at Phase V.

Teva does not develop AI products itself, but it certainly cooperates with companies, including Israeli ones, that do. "What are we looking for? A product that is usable from day one, that has no regulatory limitations, and about which we understand immediately the value it gives us.

"We generate lots of data these days, and I regret to have to inform you -- and my boss -- that we analyze and understand only 10-20% of it," Hughes said. "During my career, the cost of developing a new drug has risen from $800 million to more than $2 billion. There’s hope we can improve the situation."

1. Locating trials subjects and sites

Today, AI is mainly used for improving clinical trials. Some of these solutions may seem simple, relative to AI’s capabilities, but, nonetheless, they can save a company like Teva a great deal of money. "Let's say I’m now recruiting 2,000 patients for an asthma trial. I have to sign agreements with dozens of trial sites (hospitals or clinics), and train them to perform the trial. Sometimes, after all this work, a particular center will fail to recruit patients, not even one. AI helps me focus on the areas where relevant patients are located, where the site has a history of good performance in these types of trials."

2. Diagnostics with wearables

As a pioneering company in treating movement disorders, Teva is today highly interested in measuring biomechanics using wearable products, says Hughes.

"The conventional diagnosis of movement disorders is made by doctors who look at the patient’s walking gait in the clinic. The diagnosis varies from diagnostician to diagnostician. But today, it’s possible to let a patient wear a device for a week, which will provide us with terabytes of information, and will always measure the same way. We believe that when the data are so abundant, and so accurate, it will be possible to reduce a trial of, let's say, 700 people, to about 100." This means a saving of tens of millions of dollars per trial.

3. Diagnosis via mobile phone

"One application that’s very dear to my heart is software for digital diagnosis of a muscular condition called tardive dyskinesia," says Hughes. (Dyskinesia is a movement disorder that is usually a side effect of taking certain medications - G.W.) "In the early stages of the disease, it’s hard for doctors to detect it, and distinguish it from other conditions. We offer patients who take these drugs the possibility of looking at their phones for a short time, and automatic software can identify the disease based on their facial movements."

According to Hughes, this application is accessible to patients, and also raises awareness of the disease. This is important to Teva, because it believes that this disease is underdiagnosed at present, and that the market could be increased significantly. In addition, the application enables patients to be selected for clinical trials of the drug, and also monitors their improvement, first during the trial stage, and then during ongoing treatment, thus motivating patients to stick to the treatment.

A similar technology is being applied in diagnosing schizophrenia, a disease for which Teva has developed a delayed-release drug that is considered one of the most promising in its pipeline. "Today, schizophrenia is measured using a questionnaire for the patient and the doctor. We are adding to that a technology that analyzes voice intonation of in mobile phone conversations. It is a continuous and easy diagnosis that should also be very accurate. We’re testing it right now, against the questionnaire."

4. Discovering errors in clinical trials

Another "very important even if less exciting" tool, as Hughes puts it, is improving data collection. "Today, the data from trials are collected manually at the sites, and we also look for possible errors manually. But artificial intelligence has the potential to spot errors far more easily. Is there a different, strange pattern at one of the sites? Is there a delay in uploading the data that could indicate a problem? "

Another tool for improving clinical trials is interviewing patients using an avatar instead of medical staff members. "It may sound alienating, but this form of questioning allows patients to be candid, and is more replicable because there’s no variable response to hints or expectations on the part of a particular doctor."

5. Antibody design without mice

Improving clinical trials is considered the "low-hanging fruit" of AI. The "high hanging fruit" is designing improved antibodies to attack existing targets. Teva collaborates in this area with Biolojic Design, a company founded by Prof. Yanay Ofran. (Biolojic has several multimillion-dollar contracts with pharmaceutical companies, with potential for hundreds of millions). "This is amazing technology," says Hughes. "They’ve managed to design, for the same disease, an antibody that is safe to use, with fewer side effects, at a higher dosage, and an easier treatment protocol.

"You can’t do this kind of design using mice, because antibodies work a little differently in them from the way they do in humans. They have an amazing method for doing this, but I say to the AI companies, I don't much care how you do it - but about what you get at the end".

For Hughes, the highlight is the digital biomarker, i.e., a tool that makes it possible to find the 20% of patients for whom the product works effectively. Other areas of interest to him are segmented marketing using artificial intelligence, and using AI for continuous monitoring of treatment out in the real world.

Published by Globes, Israel business news - - on June 26, 2024.

© Copyright of Globes Publisher Itonut (1983) Ltd., 2024.

Teva executive VP of global R&D and chief medical officer Dr. Eric A. Hughes  credit: Teva
Teva executive VP of global R&D and chief medical officer Dr. Eric A. Hughes credit: Teva
Twitter Facebook Linkedin RSS Newsletters גלובס Israel Business Conference 2018