Research News

Clinical trials are critical to developing new treatments and drugs, but the processes that ensure safety, efficiency and recruitment can be lengthy and complicated. To overcome these obstacles, clinicians and researchers from the Cleveland Clinic and IBM-led International Healthcare and Life Sciences Working Group explored how quantum computing can be used to advance clinical trials. 

Researchers looked at each step of the clinical trial process to determine the potential quantum advantage. Because quantum computers function differently than classical computers, they can solve some problems faster or more efficiently. Defining where quantum computing has the potential to be most successful is the start of its application to complex healthcare challenges.  

“Clinical trials are costly, both in terms of the time it takes to design and perform them and the monetary cost – but they’re essential to advancing science,” says Daniel Blankenberg, PhD, senior author on the paper. “Within the large, complicated process of clinical trials, there are many small, complicated aspects where we think quantum computing can help.” 

The group shared their insights in Cell Press. 

Quantum computing and simulating clinical trials 

Researchers often run pretrial simulations on virtual patients to test the potential effects of treatments and drugs and gauge success or risk. Current methods rely on classical artificial intelligence (AI) and machine learning (ML) that often struggle with complex calculations.  

The working group proposes that a combination of physiology-based pharmacokinetics and pharmacodynamics (PBPK/PD) modeling and quantum machine learning (QML) can help to accurately predict the potential effects in a fraction of the time it takes classical computers. Through PBPK/PD modeling researchers can better understand the properties of drug molecules. QML can help improve the speed and accuracy of these PBPK/PD models by combining the computing power of a quantum computer with ML algorithms.  

“As we explore this method, we may be able to craft models that focus on reactions at the individual level, which will be a crucial step for developing personalized medicine,” says Dr. Blankenberg, who is Assistant Staff in Cleveland Clinic’s Center for Computational Life Sciences. 

Quantum computing and site selection optimization 

Clinical trial success depends on researchers selecting the right locations. A site needs to have adequate staff and resources for the trial, while also being in close proximity to the target patient population. To select a site, experts usually implement rule-based methods to identify patterns in data, which can be costly and time-consuming.  

The working group proposes that quantum optimization algorithms can help select sites for a fraction of the time and cost. Quantum optimization solves problems by using quantum computing principles to find the optimal solution from a set of possibilities. These algorithms need less training data to make analyses, and they are less prone to bias, which is a common problem in selecting sites. This could reduce delays and trial failures by making sure only viable sites are selected.   

Quantum computing and patient selection  

Researchers evaluate potential clinical trial participants based on several factors including medical history, age, sex and health status. Classical ML and AI can help screen and select patients based on these factors, but this involves long computation times and risks of statistical bias.  

The working group determined that quantum neural networks (QNN) have the potential to overcome these problems. QNN is a type of quantum computing that processes data like a human brain. By using this method, researchers can quickly analyze patient data to select the best patients for the trial.  

QNN not only has the potential to help the selection process, it can also reduce the number of participants. Clinical trials are often split into groups that receive treatment and placebo groups that do not receive treatment. This helps to determine if the treatment is working or if an outside factor is causing a change. QNN can help reduce the need for large placebo groups because it can create high-quality synthetic data. By producing synthetic data, QNN can help predict what would happen to a patient who does not receive a drug/treatment.  

“By decreasing the number of patients needed for the placebo groups, more patients can receive the drugs and treatments they need,” Dr. Blankenberg says. “It can also lower the costs of clinical trials making them more accessible to the patients who need them most.”