Could AI drug repurposing systems reshape healthcare delivery over the next decade?

How does AI drug repurposing actually work, and why could it reshape healthcare?

Drug repurposing means finding new medical uses for existing drugs that are already approved and have known safety profiles. Instead of starting from scratch—which can take over a decade and cost billions—AI systems can rapidly scan massive datasets of genetic, chemical, and clinical information to predict which existing drugs might work for a different disease. A 2025 review explains that machine learning and deep learning algorithms have 'significantly impacted drug discovery, repurposing, and designing new molecules or drug delivery carriers' [2]. This matters for healthcare delivery because it could dramatically shorten the time from idea to treatment, lower costs, and open up options for rare or complex diseases where new drug development is often financially unviable [5].

The practical impact on your doctor's office could be profound. Instead of a one-size-fits-all approach, AI repurposing systems could help identify which existing drug is most likely to work for your specific genetic profile or disease subtype. A 2025 study on neurosymbolic AI—which combines neural networks with logical reasoning—shows that these systems can 'analyse complex medical data, make accurate diagnoses, and develop effective treatment plans personalised to patients,' with drug repurposing as a key application [4]. This means faster access to effective treatments, fewer side effects from trial-and-error prescribing, and potentially lower healthcare costs overall.

What's the gap between the best-case promise and the typical evidence so far?

The best-case evidence is impressive: a comprehensive 2025 review reports that AI and ML have delivered 'significant advancements' in target identification, accelerated drug discovery through generative models, and enhanced structure-based drug design [3]. Another study notes that AI-driven predictive models can streamline patient recruitment in clinical trials and enable real-time monitoring, which could cut years off development timelines [3]. But the typical evidence is more cautious. The same studies that tout AI's potential also list major hurdles: data privacy concerns, lack of algorithmic transparency, ethical questions, and regulatory frameworks that haven't caught up with the technology [1][3][5].

For example, while AI can suggest a repurposed drug candidate, getting it to your pharmacy requires clinical trials, regulatory approval, and manufacturing—steps that AI can accelerate but not eliminate. A 2025 review on drug repurposing in personalized medicine explicitly states that 'challenges persist across regulatory frameworks, intellectual property protection, data integration, and the management of biological variability among patients' [5]. Another study on AI in pharmacy operations warns that 'algorithmic transparency and ethical considerations' remain unresolved, and proposes 'actionable strategies to overcome these barriers'—acknowledging they aren't solved yet [1]. So while the technology is ready, the healthcare system's infrastructure, regulations, and trust in AI are not. The next decade will likely see AI repurposing systems used in research and specialized clinics first, with broader healthcare delivery changes following only as these systemic issues are addressed.

Which areas of healthcare will AI repurposing transform first—and where will it struggle?

AI drug repurposing is most likely to reshape healthcare delivery in areas where traditional drug development has failed patients: rare diseases, complex chronic conditions, and personalized medicine. A 2025 study on psoriasis treatment—a chronic inflammatory skin disorder—shows how AI combined with nanotechnology can identify existing drugs that could be repurposed and delivered topically, avoiding the side effects of oral medications [2]. This approach could be a game-changer for patients who have few treatment options. Similarly, the neurosymbolic AI study demonstrates successful applications in predicting lung cancer tumor recurrence and classifying chronic kidney disease from clinical notes, suggesting that AI repurposing can work hand-in-hand with diagnostics to deliver targeted treatments [4].

However, AI repurposing will struggle in areas where data is scarce, diseases are poorly understood, or regulatory pathways are unclear. The same psoriasis review notes that drug repurposing 'is in its early stages' and faces challenges like 'adverse effects, high costs, and individual response variability' [2]. Another study emphasizes that integrating AI into personalized medicine requires 'robust validation pipelines' and 'open-source, collaborative models' to ensure the predictions are reliable [5]. So while AI can flag promising candidates, the final decision to use a repurposed drug in a patient still depends on human clinical judgment, regulatory approval, and real-world evidence. The biggest transformation over the next decade will likely be in how quickly researchers can generate and test hypotheses—not in overnight changes to your local clinic's prescribing habits.

About These Sources

This answer is built on 5 peer-reviewed studies — published in 2025, 5 from 2024 or later, 1 in Q1 journals — selected as the most relevant from 5 studies that passed quality screening, drawn from 59 papers retrieved from a database of over 500 million.

Sources used in this answer

Related research questions