A labyrinth of pills leading to a blossoming flower, symbolizing drug repurposing.

Unlock Hidden Potential: Repurposing Existing Drugs for New Breakthroughs

"Discover how machine learning is revolutionizing drug development by finding new uses for old medications, offering faster and more reliable treatments."


Drug discovery is traditionally a lengthy and expensive process. It often takes years and requires a substantial investment to bring a new drug to market. Moreover, predicting potential side effects remains a significant challenge. However, an emerging strategy known as drug repositioning, or drug repurposing, offers a promising alternative. This approach involves finding new uses for existing drugs, potentially shortening the approval process and reducing development costs.

Drug repositioning hinges on the idea that a drug effective for one disease might also work for another, especially if the two diseases share common underlying mechanisms. Recent advances in biomedical informatics have made it easier to systematically search for potential drug repositioning candidates, opening up new avenues for treatment discovery.

This article will dive into a machine learning approach developed to predict new uses for existing drugs and evaluate the reliability of these predictions. We'll explore how this innovative method uses data analysis to identify promising candidates for drug repositioning, potentially leading to faster and more effective treatments for a variety of conditions.

AI Search Multiple angles on this topic

Why Repurposing Matters Today

Drug repositioning (also called drug repurposing) involves investigating existing drugs for new therapeutic purposes. Its impact is most visible on the regulatory side: repositioning can simplify the procedures for introducing a previously approved drug to the market, particularly in the United States. Because the compound has already cleared regulatory hurdles for an original use, sponsors can build on established data rather than starting from scratch. These interrelated advantages help explain why the strategy is attracting growing attention across drug development.

The Core Repurposing Process

Drug repurposing, also known as drug repositioning, is a process of identifying new therapeutic uses for old, existing, or available drugs. It is framed as an effective strategy for discovering or developing drug molecules with new pharmacological or therapeutic indications. Rather than designing a molecule from scratch, the approach starts from compounds already in use or on the shelf and searches for additional clinical value. As an overview of the strategy, the chapter positions repurposing as a recognized pathway within the broader drug discovery and development landscape.

From Many Names to Proven Wins

Drug repurposing has accumulated a remarkable number of names over the years — drug repositioning, drug re-tasking, drug rescuing, drug recycling, drug reprofiling, and therapeutic switching, with the literature using at least a dozen synonyms in total. That linguistic sprawl reflects how long the practice has circulated across research communities. The strategy is also backed by a series of success stories, including notable wins in oncology. These milestones helped turn repurposing from an occasional accident into a recognized, nameable research strategy.

Machine Learning: The Key to Unlocking Drug Potential

A labyrinth of pills leading to a blossoming flower, symbolizing drug repurposing.

The core of this approach lies in using machine learning, specifically a support vector machine (SVM), to analyze various data points related to existing drugs. This data includes chemical structures, side effects, and drug targets. By feeding this information into the SVM, researchers can predict whether a drug might be effective for a new, different condition.

However, simply predicting a potential new use isn't enough. It's crucial to evaluate the reliability of these predictions. The method does this by calculating a 'reliability score,' which combines two key factors: the distance of the data point from the separating hyperplane in the SVM (indicating the strength of the prediction) and the similarity between the diseases targeted by the drug and the candidate disease.

To improve the accuracy and reliability of predictions, several key factors are considered:
  • Chemical Structure: Analyzing the molecular makeup of drugs to find similarities that suggest similar actions.
  • Side Effects: Leveraging known side effects to predict efficacy in related conditions.
  • Drug Targets: Identifying common biological targets between different diseases to repurpose drugs effectively.
  • Data Integration: Combining diverse data sources to train robust machine-learning models.
AI Search Multiple angles on this topic

Repositioning as Efficient Discovery

A 2015 review in the International Journal of Pharma Research & Review describes drug repositioning as an efficient approach to drug discovery, defined as the process of finding new therapeutic indications for existing drugs. Complementary research highlights how repositioning approaches may find suitable drug candidates for disease targets with low-frequency mutations, where conventional pipelines can struggle. To support this work, reviewers stress that it is essential to have a physical collection of drugs available for high-throughput screening. Together these reviews position repositioning as a practical complement to, rather than a replacement for, conventional discovery.

Efficacy Remains the Hard Test

Repositioned drugs still have to clear the same efficacy bar as any other candidate. When such drugs enter clinical trials, they compete with non-repositioned drugs not in terms of safety but in terms of efficacy. Because safety accounts for approximately 30% of drug failures in clinical trials, this is a significant development advantage that repositioned drugs enjoy. The flip side is that efficacy failures remain an open challenge: a repositioned drug that cannot outperform existing options clinically will not succeed, regardless of its safety record.

Repurposing Versus Repositioning

One line of analysis argues that the language matters: better distinguishing between repurposing drugs and repositioning them might sound pedantic, but the distinction carries regulatory weight. For drugs that failed due to safety or clinical concerns, repositioning may require more extensive regulatory scrutiny compared with already-approved drugs being repurposed for new indications. The case of TGN1412 is cited as an illustration of the risks of drug repositioning. In this framing, the two terms describe pathways that differ meaningfully in risk profile and oversight burden.

Imagine a scenario where a drug initially designed to treat high blood pressure is found to be effective in managing a specific type of arthritis. This is the power of drug repositioning. By leveraging existing data and advanced machine learning techniques, we can unlock the hidden potential of existing drugs and bring new treatments to patients faster.

The Future of Drug Development

This machine learning approach represents a significant step forward in drug development. By efficiently identifying new uses for existing drugs, it has the potential to accelerate the availability of effective treatments, reduce development costs, and ultimately improve patient outcomes. This innovative method paves the way for a more efficient and targeted approach to tackling a wide range of diseases.

AI Search Multiple angles on this topic

Experts See Deeper Integration Ahead

Experts are of the opinion that repositioning drug technology will have better integration in pharmaceutical research over the next decade. Beyond the laboratory, drug repositioning has also been analyzed as a business opportunity for pharmaceutical companies, with researchers weighing both the challenges and the commercial appeal. The convergence of scientific endorsement and business interest suggests the strategy is being taken seriously as a mainstream R&D channel. Viewed together, these perspectives position repurposing not as a fringe shortcut but as an increasingly institutionalized part of pharma strategy.

A Systematic, Data-Driven Frontier

The global drug repositioning market is anticipated to experience steady expansion during the 2026–2033 period, driven by broader adoption across diverse end-use industries. Underpinning that growth is a shift in method: drug repurposing has evolved from a serendipitous discovery process into a systematic, computationally driven branch of modern pharmacology. This evolution — from chance observations to algorithm-guided screening — is central to accelerating the pipeline of repurposed candidates. As computational tools mature, the outlook is for repositioning to keep feeding the development pipeline from molecules that already exist.

A Strategy Anchored in the System

Drug repositioning: a brief overview, authored by Jean-Pierre Jourdan, Ronan Bureau, Christophe Rochais, and Patrick Dallemagne, surveys the field's core concepts and current context. As a review document, it reflects how repositioning has grown from a niche idea into a recognized research stream within drug development. The very existence of such a dedicated overview signals the strategy's systemic relevance across the research community. For readers, the piece serves as a structured entry point into how existing drugs are being re-examined for new uses.

Case Studies With Real Patients

Real-world case studies show repositioning moving from theory to practice across very different disease areas. A network-based model (DRGBCN) has confirmed practical application in real-world drug repositioning scenarios involving bladder cancer and acute lymphoblastic leukemia. In neurology, researchers present real-world repositioning case studies for movement disorders, spanning serendipitous discovery and rational strategies. The same literature notes that the current model of drug discovery and development implies high investment by the pharmaceutical industry, which ultimately drives high drug prices — a burden that repositioning's lower-risk economics are meant to ease.

About this Article -

Written with AI assistance from published research, and reviewed by the Mystum team. See our About page for more information.

This article is based on research published under:

DOI-LINK: 10.1007/978-1-4939-8955-3_16, Alternate LINK

Title: Machine Learning Approach For Predicting New Uses Of Existing Drugs And Evaluation Of Their Reliabilities

Journal: Methods in Molecular Biology

Publisher: Springer New York

Authors: Yutaka Fukuoka

Published: 2018-12-14

Everything You Need To Know

1

What is 'drug repositioning,' and how does it differ from traditional drug discovery?

Drug repositioning, also known as drug repurposing, identifies new therapeutic uses for existing medications. This is achieved by exploring whether a drug approved for one condition can be effective for another, particularly if the diseases share common biological pathways. This strategy can significantly shorten the drug development timeline and reduce costs compared to developing a new drug from scratch.

2

How is the 'reliability score' calculated in the machine learning model, and what does it indicate?

The reliability score in the machine learning model is calculated using two key factors: the distance of the data point from the separating hyperplane in the Support Vector Machine (SVM), indicating the prediction strength, and the similarity between the diseases targeted by the drug and the candidate disease. This score assesses how confident the model is in its prediction for a new use of an existing drug.

3

How does a 'Support Vector Machine (SVM)' function in predicting new uses for existing drugs?

Support Vector Machine (SVM) analyzes data points, including chemical structures, side effects, and drug targets, to predict if a drug might be effective for a new condition. By integrating diverse data sources, the Support Vector Machine identifies patterns and relationships that suggest new potential uses for existing drugs. The Support Vector Machine uses the reliability score to improve the accuracy.

4

What key factors are considered to improve the accuracy and reliability of predictions made by the Support Vector Machine (SVM)?

Several factors are used to improve the accuracy, including analyzing the chemical structure of the drug molecules, known side effects, and drug targets. By integrating diverse data sources into the Support Vector Machine, it enhances the ability to make predictions. The chemical structure helps the Support Vector Machine find drugs with similar actions. Side effects help the Support Vector Machine predict efficacy in related conditions. Drug targets helps the Support Vector Machine identify common biological targets between different diseases.

5

What are the potential implications of drug repositioning, particularly regarding the future of drug development and patient care?

Drug repositioning holds the potential to significantly accelerate the availability of effective treatments and improve patient outcomes. The Support Vector Machine is used to analyze the drug and diseases. By efficiently identifying new uses for existing drugs, development costs can be reduced, and the time it takes to bring new treatments to patients is shortened. The use of machine learning is expected to lead to more targeted treatments for a wider range of diseases.

Newsletter Subscribe

Subscribe to get the latest articles and insights directly in your inbox.