There are multiple areas in the healthcare industry where advances in quantum computing could directly benefit pathologists, clinicians, and experimentalists. The downstream effects of these benefits would likely result not only in life-saving new drugs, but also in making it possible for patients to gain quicker and easier access to the healthcare they need. At this moment in history, that speed element is especially important, and a major reason why quantum’s applications in health are an increasingly critical research realm. The World Health Organization is predicting a deficit of almost 10 million nurses, doctors, and other health providers by 2030. At the same time, rising life expectancies worldwide paired with more and more advanced and specialized approaches to medical treatment means that our population will be older than ever, and a shrinking medical workforce will be required to serve a growing number of patients with increasingly complex needs.

Quantum computing uses quantum physics to run algorithms in ways that are not currently possible on classical (i.e., normal) computers. This nascent field is producing “prototype” quantum computers which are growing in size and computational power. Quantum algorithm developers are working with the computers that are available today to identify where quantum may have the greatest impact. Current focuses of their work include removing bottlenecks caused by high data volumes, exploiting quantum tools to study quantum systems, and further optimizing existing quantum methods.

At Booz Allen, we understand that as the power of quantum computers grows over the next decade, we must be prepared to rapidly apply those methods to computational biology and vaccinology, biostatistics, and other health research areas. To ensure that our company—and our clients—are ready for the quantum future, Booz Allen is actively engaged in cross-cutting studies at the intersection of health and quantum computing to identify where these novel computers could make a significant impact. 

One of the key health areas that quantum technology will likely disrupt is the study of protein folding. The characteristics of any given protein are reliant upon the properties of the amino acids which form it, which interact with one another to form unique folds. These help proteins bind to one another and mark them as friend or foe to the body’s immune system. In recent years, computational biologists have written impressive algorithms that can model the shape of proteins. But even the best methods cannot reach the level of precision required to achieve desired breakthroughs in personalized medicine and targeted therapeutics—something that the quantum computing community hopes to be able to achieve with future devices.

Using a quantum algorithm called a Variational Quantum Eigensolver, it's possible to produce a statistical representation of possible folds in an amino acid chain, with the result being a prediction for the protein configuration we would observe in nature. The size of today’s quantum computers limits the overall size of the protein that can be modeled, but with larger, more powerful machines, future generations will be able to capture the dynamics of larger and larger proteins. The ability to determine the configuration of a protein computationally, rather than experimentally, could save time and cost when trying to determine the behavior of new bacteria and viruses. 

Another promising intersection of health sciences and quantum computing is in small molecule simulation. Small molecules—which are around 20 atoms long and made of carbon, hydrogen, oxygen, and/or nitrogen—can easily pass through cell membranes, making them crucial to the metabolism of living organisms and very useful for pharmaceuticals. Despite their limited size, there are more possible combinations of those four elements than there are atoms in the known universe. Experimentalists will never be able to test that many combinations, so theorists use computational models to provide them with a “good enough guess” to narrow the search.

Quantum computers have been used in combination with state-of-the-art neural networks to help produce candidates for small molecules. In one method, called a Quantum Generative Adversarial Network, a neural network implemented on a quantum computer—the “generator”—produces a candidate after being trained on a dataset of known small molecules with useful properties. Simultaneously, a second neural network—the “discriminator”— assigns a likelihood that the generated molecule is realistic. The cycle continues until the generator is solely producing molecules that are acceptable to the discriminator. Today’s results, though small-scale, demonstrate that a more sophisticated quantum computer will be able to produce viable candidates for experimentalists to synthesize and explore. Shifting more of the design cycle to in silico experimentation will decrease the overall cost and time of the research and development lifecycle, allowing biologists to focus their work on molecules that have been vetted for useful—and potentially life-saving—properties. 

Medical diagnostics—including imaging, pathology, and genetic sequencing—is considered to be one of the most promising healthcare applications of artificial intelligence and machine learning, and it’s one that quantum computers could significantly enhance. Quantum computers may be able to calculate distances between data exponentially faster than classical computers, creating marked speed-ups for a variety of machine learning algorithms. Booz Allen scientists simulated a quantum computer executing one such algorithm, called k-nearest neighbors, on a dataset of patients with heart disease indicators. We were able to predict whether a given patient had heart disease with 84% accuracy (correctly determining if a patient had heart disease or not) and 90% specificity (correctly determining when a patient did not have heart disease). As the amount and quality of data to train these quantum algorithms increases, a future quantum computer could help doctors evaluate diagnostics more efficiently, leaving them more time to spend with patients. 

The examples above represent just a few of the many areas in the health sciences where incremental gains in quantum algorithms for machine learning can be applied. Others include the detection of fraud in health spending, automatic coding of medical records, and genomic analysis. Eventually, researchers hope to not only model the folding of proteins, but entire systems of proteins to provide robust and accurate models for personalized therapeutics. The healthcare industry is no stranger to making the investments necessary to seize this advantage—magnetic resonance imagers (MRI), for example, rely on quantum-based sensors. Leaders in the industry must remain aware of the impact quantum computing can have on existing plans for technology adoption and integration and be prepared to seize the quantum advantage as it appears.