top of page
Launch Excellence Partners
  • LinkedIn
Writer's pictureHettie Stroebel

Is AI/ML Bringing Medicines To Patients Faster And Cheaper?


September 1, 2022

Dan Schell, Editor



 

Mathematician Alan Turing changed history in 1950 with a simple question: “Can machines think?”

The phrase artificial intelligence (AI) was coined in 1956, yet the pharmaceutical industry embraced artificial intelligence and machine learning (ML) only about 15 to 20 years ago as a technology for use in drug discovery and drug development. It is critical to move beyond simply using AI as

a buzzword and instead determine if AI/ML does indeed change the discovery and clinical development process by bringing innovation to patients faster and at a lower investment.



WHAT ARE AI, ML, AND DL?


Let’s start by getting a basic understanding of the differences between AI/ML and DL. AI is a technology- based system used to create systems with human- like behavior. ML is an application of AI, where AI is achieved by using algorithms that are trained with data. Deep learning (DL) is a type of ML vaguely inspired by the structure of the human brain, referred to as artificial neural networks.


Developing a new drug is a long and expensive process with a low success rate, as evidenced by the following estimates: the average R&D investment is $374 million to $1.3 billion per drug, the median development time for each drug ranges from 5.9 to 7.2 years for non-oncology and 13.1 years for oncology, and the proportion of all drug development programs that eventually lead to approval is 13.8%.


AI/ML is therefore attractive to the drug discovery and development industry as a means to help overcome these barriers on the journey to getting medicines to patients. The hope is that they would result in more efficient drug development and, thereby, reduce costs, shorten development times, and drive a higher probability of success.



HOW HAVE AI/ML BEEN USED IN DRUG DISCOVERY?


AI/ML have been utilized in drug discovery to inform drug screening by some of the large market cap pharmaceutical companies. AI has played a significant role in this area to further predict physicochemical properties, desired bioactivity, and drug toxicity.


Some examples of AI tools that are currently used in drug discovery include: AI software that predicts the toxicity of about 12,000 drugs, an AI system that finds a suitable candidate in drug discovery, a molecular generation tool that helps to create molecules with desired properties, and a neural graph fingerprint to predict properties of novel molecules.


AI has been used in the design of molecules to predict the target protein structure, as numerous proteins are involved in the development of a disease. One of the most important safety elements in drug discovery is the risk of drug-drug interaction; this is pivotal to the success of a medicine. The ability to predict the interaction of a drug with a receptor or protein is essential to understanding its efficacy and effectiveness. It also allows the repurposing of drugs and prevents poly-pharmacology. In this space, multiple approaches, frameworks, and models have been successfully used. However, the end goal of using AI, ML, and deep learning in drug development is to overcome efficacy, safety, time, and money challenges and bring the best drugs to market, faster.



AI/ML IN CLINICAL DEVELOPMENT IS MORE LIMITED


AI/ML have been used less often in clinical trial development, where about 80% of trials do not meet enrollment timelines. Based on recent estimates, among Phase 3 trials with novel therapeutics, 54% failed in clinical development, with 57% of those failures due to inadequate efficacy. A major contributing factor is failure to identify the appropriate target patient population with the right dose regimen, including the right dose levels and combination partners. The industry is also still seeking solutions to incorporate underrepresented populations in studies, but progress using AI/ML has been limited in this critical cost-driven development phase.



One of the disease areas that can greatly benefit from AI-based technology in drug development is Alzheimer’s disease. Over the past 20 years there has been a lack of success in Alzheimer’s drug development, leaving many among the rapidly growing aging population in many developed countries without a treatment option.


Deep learning technology has been explored to predict therapeutic use of less than a handful of drugs but has not yet been deployed for use in clinical development.


In a recent study, a 3D simulation of tumor growth and its dynamic microvascular network was used to investigate the efficacy of three different antiangiogenic drug treatments that were approved by the FDA and are currently available to patients in the U.S. This simulation is important as it demonstrated that AI methods can be used to predict the efficacy of different drug treatments before they are started as a treatment.


Between 2015 and 2020, the FDA and the European Commission approved 222 and 240 AI devices, respectively, for clinical use, often under the “Software as a Medical Device” or similar designation.


In the field of AI-linked devices, the FDA has already approved 71 devices that can be used by clinical practitioners as diagnostic tools. The majority of these devices are for cancer radiology. The specific cancer types that now are experiencing the greatest advantages from the use of AI-based devices in clinical practice are breast cancer, lung cancer, and prostate cancer.



SIGNIFICANT BARRIERS TO ADOPTION OF AI REMAIN


Collaborative partnerships have been established by some leading pharmaceutical companies and AI organizations to assist in fast-tracking their drug discovery and development. The adoption of AI on a full scale for clinical development, however, poses a number of challenges, such as, for example, a lack of substantial data to train the system, a lack of skilled employees to operate AI-based technology, and the investments needed by companies to bring this technology in-house.


In conclusion, despite growing utilization of these technologies, especially in drug discovery, no published literature is available to show that these strategies result in more cost-effective investment and faster clinical development to bring new drugs to patients.

Comentários


bottom of page