Can AI Tackle Drug Shortages?
The role of artificial intelligence in getting more medicines to market
Jo Varshney | | 4 min read | Hot Topic
It’s been more than a year since the onset of some of the worst drug shortages in US history. You may think I’m referring to the shortage of Adderall – the most commonly prescribed drug for the treatment of attention deficit hyperactivity disorder (ADHD), which was confirmed by the FDA on October 12, 2022. Though the continuation of that shortage (and the multiple failed recovery efforts) is mortifying at best, I’m actually referring to the thousands of other drugs – including lifesaving and life-improving treatments for cancer and hundreds of diseases – that should be undergoing research trials right now. Strangely, this problem hasn’t received anywhere near the same level of attention in the media.
Consider this: in October 2023, there were 16 FDA-approved oncology medications experiencing shortages (up from 11 in July 2023). The situation is a huge problem for patients and the doctors who treat them. In every case of a drug shortage, doctors are forced to choose another means of treatment (a different drug or another type of therapy) or to forgo treatment until options improve. We’ve seen how well that works with drugs like Adderall (it doesn’t). In fact, the Adderall shortage had a cascading effect; when doctors started prescribing other ADHD medications, it resulted in widespread shortages of multiple medications. But when it comes to even more specialized drugs, like cisplatin, the outcome can be even more somber. Cisplatin – a powerful chemotherapy drug frequently used for the treatment of multiple cancers including ovarian, bladder, brain, throat, cervical, and lung cancer – fell into short supply in December 2022, when manufacturer Intas Pharmaceuticals ceased production in the face of FDA concerns over quality. The move immediately suspended US access to 50 percent of its cisplatin supply, leaving cancer patients with suboptimal treatment options, and no other manufacturers to step in.
Quality concerns, capacity-building failures, and cascading effects aren’t the only reasons that drug shortages happen or worsen. In fact, until recently, regulatory restrictions were the most cited precursor to drug shortages.
The bigger issue, and the reason we fight the same drug shortages year after year, is that we simply rinse and repeat the same solutions that never hold. The only way out of the cycle is to solve the real problem: there aren’t enough medications to choose from. And that problem starts long before drugs ever hit the market. In fact, just one out of 10 potential drugs successfully clears preclinical or clinical trials. So far this year, the FDA has only approved 43 novel drugs for the treatment of human medical conditions. We may never know how many drug candidates didn’t make it through clinical trials or how many lives they might have saved. Perhaps more sobering is the reality that most drugs fail for completely preventable reasons, including human error, flawed study design, or underpowered clinical trials with too few participants.
There are ways to fix the problem and give every useful drug candidate a chance. Translational medicine concepts are already being applied to drug trials to help reduce human error. And AI is being deployed to make the drug development process more efficient than ever. My own company, VeriSIM Life, is deploying hybrid AI methodologies to identify the most effective compounds and combinations with the fewest side effects. Importantly, we’re making it happen before ever involving a human patient – a move that de-risks the experimentation process. When applied correctly, our approach could significantly reduce time-to-market (a process that presently takes 10–12 years on average to complete).
But even a shortened timeline for drug development is a long time to wait when current treatments are already in short supply. And that’s why applying AI to reformulation – a capsule to a patch or a caplet to an injection – could come in handy right now. AI methodologies can speed up the process of developing new applications for already-existing medications by running simulations that perfect dosing across multiple application types.
Although the focus right now is on moments of hope, such as the recent approval of multiple generics for Vyvanse (an alternative to Adderall), shortages for hundreds of other medications will not only continue this year but also circle back around unless we do something about it. In my view, that means thinking outside of the box to change the drug industry from the bottom up.
AI can really help. So the big question is whether the industry will move quickly enough to change a paradigm in desperate need of a technology-led evolution.