Drug design and development is expensive, often inadequate, and prone to failure. Traditionally, it's a haphazard process that can take more than a decade, costing roughly $2.3 billion. This figure has risen even more recently, with costs increasing 15% for the world's top biopharmaceutical companies last year alone. Yet, despite misconceptions about its use, artificial intelligence (AI) is poised to help companies expedite the drug design process through modalities such as machine learning models.
Designing therapies of the future with technologies of the future
The world of chemistry spans more than 10 years60 There are countless potential drug molecules, too many to study in a human lifetime. Current methods select, design, and prioritize molecular structures based on a variety of factors, ranging from desired biological activity to retrosynthetic analysis. However, this process requires solving complex, multidimensional optimization problems and can take a significant amount of time.
Thankfully, AI is revolutionizing how we explore this area of chemistry by speeding up the identification of promising compounds. Machine learning and data analytics techniques can rapidly identify hits and leads, accelerate drug target validation, and optimize drug structure design, shortening the time and dramatically reducing the cost of drug development. As a result, it is becoming possible to test a wider range of potential therapeutics.
One of the biggest barriers in drug design is finding the right protein involved in a disease and accurately creating a drug molecule. AI tools like AlphaFold and MATLAB enable scientists to predict the 3D structure of target proteins from scratch, including molecules never seen before. This significantly reduces time and cost, and enables more accurate and effective drug formulations. These latest machine learning AI models work by predicting protein structure through four models:
- An input module that collects amino acid sequences from various proteins
- Neural networks that use pattern recognition software to convert amino acid sequences into spatial information
- Output model converting spatial information into 3D structure
- A purification process that fine-tunes the structure of drug molecules
In the near future, these machine learning techniques will become even more valuable as more automated drug design models are implemented. Research shows that AI has the potential to not only develop drugs, but also develop them in consistent batches and optimal dosage forms.
AI-designed drugs in clinical trials
AI-designed medicines are already in clinical trials, and preliminary results are very promising: in June 2022, an AI-designed drug molecule by Exscientia successfully entered a Phase 1b/2 clinical trial in certain cancer patients. Similarly, in early 2023, New Zealand-based Insilico Medicine demonstrated the ability of its Pharma.AI platform to target and design drug molecules that are highly integrated with human biology.
That said, AI-designed medicines are still in their infancy, with many innovative companies making claims that have yet to be realized. While it may be years before the first fully AI-designed medicine hits the market, it's clear that this technology is definitely shaking up the pharmaceutical industry for the better.
Application of AI in the medical field
In addition to its use in drug design and development, AI is also being implemented in real-world healthcare to speed up processes and streamline workflows. For example, local infusion centers, which, like the rest of the healthcare industry, face major challenges such as staffing shortages and operational inefficiencies, have seen great benefits from AI. A survey of 100 U.S. cancer center leaders conducted by LeanTaaS demonstrated that AI reduced patient wait times by 30%, reduced staff overtime by 50%, and enabled them to accommodate a 15% increase in patient volume.
As local infusion centers are empowered to take advantage of these technologies, they will inadvertently improve the patient experience and create a more seamless journey for people who have struggled to find streamlined ways to manage their illnesses. While it is true that increasing technology skills among staff will increase stress, in the long run, this technology will enable healthcare providers to better allocate resources and ensure patients receive the right care.
Over the next few years, AI will likely integrate larger and more diverse datasets, including real patient data, to further refine drug design and discovery, enabling new products to be developed in record time. However, it's important to look beyond the hype and take things one step at a time, as past technologies have been expected to revolutionize drug discovery but have fallen short.
Editor's note: The author and his employer have no financial relationship with any companies or tools mentioned in this article.
Photo: metamorworks, Getty Images
Shane Reeves, Chief Executive Officer of TwelveStone Health Partners, oversees all aspects of the business including product, marketing, sales, finance and delivery strategy. Shane began his career in 1994 when he joined the family business and rose through every function within the company. Under Shane's leadership, the organization moving forward as TwelveStone has grown into a broad-based healthcare services company with numerous clients across the entire continuum of care.
This post MedCity Influencer Program. Anyone can have their perspective on business and innovation in healthcare featured in MedCity News through the MedCity Influencers program. Click here to find out how.
