Journal of the American Medical Informatics Association, 2024, 31(8), 1774–1784

https://doi.org/10.1093/jamia/ocae149
Advance access publication 27 June 2024
Review

Review
A primer for quantum computing and its applications to
healthcare and biomedical research
Thomas J.S. Durant , MD1,2, Elizabeth Knight, MS3, Brent Nelson, MD4,5, Sarah Dudgeon, MS6,
Seung J. Lee, PhD1,3, Dominic Walliman, PhD7, Hobart P. Young , PhD1,

Downloaded from https://academic.oup.com/jamia/article/31/8/1774/7700020 by guest on 22 October 2024

Lucila Ohno-Machado , MD, PhD2, Wade L. Schulz , MD, PhD�,1,2
1
Department of Laboratory Medicine, Yale School of Medicine, New Haven, CT 06520, United States, 2Biomedical Informatics and Data
Science, Yale School of Medicine, New Haven, CT 06510, United States, 3Yale School of Medicine, Yale University, New Haven, CT 06510,
United States, 4Newport Healthcare, Minneapolis, MN 55435, United States, 5Department of Psychiatry, University of Minnesota,
Minneapolis, MN 55454, United States, 6Computational Biology and Bioinformatics, Yale University, New Haven, CT 06510, United States,
7
Domain of Science LTD, Bristol, BS37 6AG, United Kingdom
�Corresponding author: Wade L. Schulz, MD, PhD, Department of Laboratory Medicine, Yale School of Medicine, 55 Park St, New Haven, CT 06510,
United States ([email protected])

Abstract
Objectives: To introduce quantum computing technologies as a tool for biomedical research and highlight future applications within healthcare,
focusing on its capabilities, benefits, and limitations.
Target Audience: Investigators seeking to explore quantum computing and create quantum-based applications for healthcare and biomedical
research.
Scope: Quantum computing requires specialized hardware, known as quantum processing units, that use quantum bits (qubits) instead of clas­
sical bits to perform computations. This article will cover (1) proposed applications where quantum computing offers advantages to classical
computing in biomedicine; (2) an introduction to how quantum computers operate, tailored for biomedical researchers; (3) recent progress that
has expanded access to quantum computing; and (4) challenges, opportunities, and proposed solutions to integrate quantum computing in bio­
medical applications.
Key words: quantum; quantum computing; biomedical research; healthcare; quantum annealing; universal gate-based quantum computing.

Introduction the need for specialized programming skills to effectively use

quantum computing resources.10–12
Quantum computing is an emerging field with the potential
The accessibility of quantum resources has rapidly
to accelerate computation in many areas of science and tech­
nology.1,2 Based on the principles of quantum mechanics, expanded with the general availability of commercially avail­
quantum computing uses a fundamentally different approach able hardware, high-level programming languages, and
than classical computing.3 Because of this, quantum com­ cloud-hosted resources, including central processing unit
puters (QCs) have the potential to perform complex calcula­ (CPU)-based simulators and quantum processing units
tions more efficiently than classical computers, and thus may (QPUs).13 Despite this progress, the practical application of
offer promising applications within healthcare and biomedi­ quantum computing to specific tasks remains a challenge,
cal research.4,5 with limited examples available for practitioners who may
Quantum computing has several possible benefits over not have extensive experience in low-level programming or
classical computing in biomedical problems that have high quantum concepts.11
computational complexity (Table 1).6 For example, QCs can By harnessing the power of quantum computing, certain
be used to optimize solutions for complex problems that have problems that were previously intractable or extremely time-
an inherently underlying quantum structure, such as simulat­ consuming may lead to new insights and discoveries.
ing molecular interactions and protein folding.5,7,8 Similarly, However, this may also work against many applications in
quantum algorithms, such as Grover’s algorithm, can search healthcare, particularly those that rely heavily on encryption
large, unstructured datasets more efficiently than their classi­ as a primary means to protect data. In this manuscript, we
cal counterparts, enabling more effective optimization of present an overview of quantum computing principles, specif­
complex problems in healthcare and biomedical research.4,9 ically tailored to biomedical researchers. We provide exam­
However, there are also notable drawbacks, including cur­ ples of how and when quantum computing can be applied to
rent limitations in error correction, hardware scalability, and biomedical and healthcare applications. Lastly, we provide a

Received: March 1, 2024; Revised: May 29, 2024; Editorial Decision: May 30, 2024; Accepted: June 3, 2024
# The Author(s) 2024. Published by Oxford University Press on behalf of the American Medical Informatics Association.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which
permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.
Journal of the American Medical Informatics Association, 2024, Vol. 31, No. 8 1775

Table 1. Areas of application in healthcare and biomedical research.

Area of application Description References

10,26–32,34–40,101,102
Artificial intelligence Quantum computing brings novel algorithms for machine learning applications,
and machine learning with recent publications demonstrating potential advantages for feature selection
and classification models.
25,41–47,103
Genomics Quantum computing may help with the analysis of large-scale genomic data, lead­
ing to improved understanding of genetic diseases and personalized medicine.
Quantum algorithms like Grover’s search algorithm and its variations may be
employed to search and analyze vast genomic databases more efficiently than
classical methods.
48–52
Protein folding Understanding protein folding is essential for the study of diseases related to protein
misfolding, such as Alzheimer’s and Parkinson’s. Quantum computing can poten­
tially improve the efficiency of predicting protein folding patterns. Algorithms
like the Quantum Approximate Optimization Algorithm (QAOA) can be used to
tackle optimization problems, including protein folding.

Downloaded from https://academic.oup.com/jamia/article/31/8/1774/7700020 by guest on 22 October 2024

5,53–61
Drug discovery Quantum computing could significantly accelerate the drug discovery process by
simulating molecular interactions and predicting the behavior of drug candidates.
Quantum algorithms like the Variational Quantum Eigensolver (VQE) and
Quantum Phase Estimation Algorithms (QPEA) can be used to determine molecu­
lar ground-state energies, which is crucial for understanding molecular properties
and reactions.
62–66
Network analysis Network analysis in systems biology, utilized extensively in bioinformatics, involves
tasks like molecular modeling and mapping complex biological pathways. Quan­
tum computing has been applied to enhance community detection within net­
works, offering approaches that are comparable to classical methods. This has
been described using quantum annealing and other quantum-based algorithms,
which show promise in effectively partitioning networks into communities, a crit­
ical step in understanding biological and computational systems more deeply.
67,68,70,73
Cryptography Quantum computing poses a significant challenge to current encryption methods
like RSA, as quantum algorithms, such as Shor’s algorithm, can efficiently factor
large prime numbers, potentially undermining these encryption schemes58. Con­
versely, it also enables the development of quantum-resistant encryption techni­
ques, such as Quantum Key Distribution (QKD), which may enhance modern
cybersecurity.

high-level overview of recent progress that has expanded

PH
access to quantum computing, as well as the challenges,
opportunities, and proposed solutions to integrate quantum
solutions into biomedical computing. NP-Problems

NPC
Impact of quantum computing on healthcare
and biomedical research
The strength of quantum computing resides in its capacity to
simultaneously represent multiple possibilities through quan­
tum parallelism, which stems from the quantum properties of P
superposition, entanglement, and interference, described fur­
ther in the following section.6 Computational complexity
theory serves as a framework for understanding the bounda­
ries and conditions in which quantum computing resources
Figure 1. Computational complexity theory has traditionally included 3
can demonstrate a proven advantage (Figure 1).14 However, major complexity classes: polynomial-time (P), non-deterministic
it remains an open and crucial question as to whether QCs polynomial-time (NP), and NP-complete (NPC), all of which are contained
can more efficiently solve problems that are intractable for in polynomial hierarchy (PH) space. Bounded-error quantum polynomial
classical computers.15 Classical computational complexity time (BQP) is a complexity class that represents problems that should be
efficiently solved by a quantum computer. Problems that fall within this
theory revolves around the P (polynomial time) and NP (non­
space and the bounds of the BQP space itself remain an open question,
deterministic polynomial time) classes. Problems in P can be with recent literature indicating that it may not be constrained by the PH-
solved quickly (in polynomial time) by classical computers, space. Depending on the bounds of the BQP-space, there may be
whereas NP problems have solutions that can be verified problems that will only be efficiently solved by quantum algorithms.
quickly but may not be solvable quickly. QCs introduce new Figure adapted from Nielsen and Chuang6.
complexity classes, like BQP (bounded-error quantum poly­
nomial time).14 This class is significant because some prob­ quantum solutions may lead to new efficiencies in genomic
lems that are believed to be outside of P (and potentially in data analysis, molecular simulation, or encryption technology
NP) for classical computers, like integer factorization, fall (or, potentially, breaking current encryption algorithms).16,17
into BQP, suggesting that QCs may be able to solve them effi­ Quantum algorithms have also shown promise in machine
ciently.14 These problems are of practical note since efficient learning (ML); however, while some problems in ML can be
1776 Journal of the American Medical Informatics Association, 2024, Vol. 31, No. 8

analyzed in terms of computational complexity theory, ML bits are used by traditional computing hardware to store data
as a whole is not defined by these classes. and are manipulated through a series of mathematical and
Despite our limited understanding of how and when to use logical operations to perform computational tasks. The most
QCs to tackle real world problems, recent literature has only notable feature of bits in this context is that bits can only be
just started to identify near-term areas where quantum com­ in 1 of 2 states (0 or 1), and each bit exists in a state that is
puting applications may achieve quantum advantage.4 In the independent from others such that the state of one bit does
domain of biomedicine, there exists a wide range of computa­ not directly influence the state of another bit.6 In contrast, a
tionally intractable problems that intuitively lend themselves qubit, or quantum bit, is the fundamental unit of information
to quantum computing, with algorithms that appear to be used in QCs (Figure 2B). Qubits are minimalistic, physical
well-suited for these challenges (Table 1). As quantum com­ systems where the 0 and 1 states are encoded onto a 2-level
puting resources continue to mature, applied research in this quantum state that displays the unique properties of quantum
domain is expected to grow. We anticipate that, mirroring mechanics, including superposition, entanglement, and
the trends in ML, inter-disciplinary collaboration will speed interference.1
the development and use of quantum algorithms.

Downloaded from https://academic.oup.com/jamia/article/31/8/1774/7700020 by guest on 22 October 2024

How quantum computing works Superposition and measurement
Quantum advantage Superposition refers to the ability of a quantum object to
simultaneously exist in a mixture of states.20 In contrast to
Superposition and entanglement are widely recognized as
critical features in quantum computing, which may afford it classical bits, this means that qubits can exist in a continuum
significant advantages over classical computing.6 In addition, of states between 0 and 1. However, if a measurement is per­
several other quantum phenomena, such as coherence, con­ formed, the probabilistic nature of states will cease to exist,
textuality, interference, and superdense coding, may also play and the state of the qubit will resolve to a base state of 0 or 1
pivotal roles.18 For the interested reader, further information (Figure 2B).6 In the context of quantum computing, measure­
on additional quantum mechanical properties and how they ment refers to the process of extracting classical information
may be leveraged to accelerate computing can be found in from a quantum system, such as a qubit, by observing its
recent literature.18,19 In addition, it is important to note that state. Measurement plays a crucial role in quantum comput­
the following sections will focus on principles that are pri­ ing, as it bridges the gap between the quantum realm and the
marily employed in gate-based QCs. classical world.6,20 Importantly, the superposition state of a
qubit is unobservable, but the superposition state can be
Bits and qubits manipulated prior to measurement which results in experi­
In classical computing, a bit is the fundamental unit of infor­ mentally observed differences in the probability of
mation, represented by either a 0 or a 1 (Figure 2A).20 These outcomes.6,20

1
Quantum Bit (qubit) y

Classical Binary System (bit)

y
1 1 x z

65%

x M
z
35%
y

0 0 0
x z

0
A B

Figure 2. (A) Classical bits, the fundamental units of information in classical computing, represent binary data as either a 0 or a 1, typically by utilizing voltage
levels within an electronic system. By flipping the voltage between 2 predefined levels, such as a high voltage to signify a 1 and a low voltage for a 0, the
state of a bit can be easily switched, stored, and transmitted. The use of voltage levels in this manner allows for a straightforward, reliable method of
encoding and processing digital information. (B) Quantum bits, or qubits, are the fundamental units of quantum computing, and unlike classical bits, they can
exist in a superposition of both 0 and 1 states simultaneously. This unique property is made possible by the principles of quantum mechanics, particularly
wave-particle duality and superposition. The actual state of a qubit can be described by a complex probability amplitude that defines the likelihood of
observing either a 0 or a 1 upon measurement. When a qubit is measured, its superposition collapses to one of the two classical states (0 or 1).
Journal of the American Medical Informatics Association, 2024, Vol. 31, No. 8 1777

Entanglement referred to as quantum logic gates.6 QCs that follow this

The study of quantum systems has shown that, under specific paradigm are known as universal gate-based QCs.23 Classical
conditions, when 2 particles are prepared in close physical and quantum logic gates are similar in that both take a set of
proximity to each other, the superposition state of particle A inputs and produce an output. Some logic gates are analo­
and particle B will strongly correlate with each other, even gous between classical and quantum systems. For example,
after the particles are physically separated over large distan­ the classical NOT gate (Figure 3A) and quantum Pauli X gate
ces, which is known as entanglement.2 Experts in quantum (Figure 3B) and their corresponding truth tables demonstrate
computing have shown that entanglement can be leveraged as that the logical output of these gates is equivalent.6
a key feature that may allow QCs to gain a computational While quantum and classical computing gates are concep­
advantage over classical hardware for certain computational tually analogous, there are also key differences. A significant
problems.20 difference is the absence of conventional data flow through a
In quantum computing, a system of n qubits can be pre­ quantum gate. As quantum gates are applied to a register of
pared in an entangled state such that their individual states qubits, the state probabilities are changed, influencing the
are no longer mathematically separable. When the 2 qubits eventual outcomes when these qubits are measured.20 The

Downloaded from https://academic.oup.com/jamia/article/31/8/1774/7700020 by guest on 22 October 2024

are measured, the possible results 00, 01, 10, or 11 will be measurement process translates the quantum information
observed according to a probability distribution defined by into a classical form, which is read out to a classical register,
the entangled quantum state.1,6,20 This represents a novel providing results that can be used as a computing resource.6
approach to computing, since the individual state of a classi­ Another key difference in quantum computing is that quan­
cal bit does not directly influence other bits. If we consider a tum gates are reversible, meaning that for a given output
system of n qubits, the quantum state is defined by 2n ampli­ there is a known input.6 This follows a fundamental theorem
tudes.1,20 In practice, the number of possible states in a sys­ in quantum mechanics that information cannot be lost or
tem of 300 qubits would be larger than the number of atoms destroyed. This constraint is not true for all classical logic
in the observable universe.6 While a similar principle applies gates, such as the AND gate.
to classical computers, quantum computing can exploit this Despite these differences, both classical and quantum logic
characteristic without the need for the user to concurrently gates are similarly combined into circuits. For the interested
manage all state permutations, which constitutes a distinct reader, Nielsen and Chuang provides a comprehensive review
advantage over classical computers.1,20 of the details associated with these quantum operations.6
Quantum circuits can take complex forms, but a simple cir­
Quantum computation and matrix representation cuit can demonstrate a computation analogous to a classical
Mathematically, qubits and their associated quantum proper­ bit-flip (Figure 3C, D). Quantum circuits are read from left-
ties can be described using linear algebra. Accordingly, the to-right, with the top line representing a quantum register
2-level superposition state of a qubit can be described containing information associated with one qubit (Q). The
mathematically as the following: bottom-most line (C) represents a classical register consisting
of 1 bit for storing the qubit measurement (M). In this circuit,
jψi ¼ αj0i þ βj1i: qubit Q is initialized in a base state of j0i, flipped to a state of
j1i, and measured (M). The output of this quantum circuit
The j and i is called Dirac notation and it is used as a would yield a state of 1 on the classical register 100% of the
reminder that the enclosed variable is a vector.6,20,21 In this time (Figure 3C). If, after initialized, the Hadamard (H) gate
case, the following qubit states can be written as:
" # " # NOT Gate
1 0
A A Q0
j0i ¼ and j1i ¼ : M

0 1 A A 1
C
A C
Here, α and β are complex numbers and are referred to as
Pauli X Gate
amplitudes. The square of these amplitudes represents the
A A Q0
probability of a qubit being in either of the corresponding M

A A 50:50

states. The exact values of α and β cannot be precisely deter­ C

1
0

mined because, when a qubit is measured, the superposition B D

state of the qubit will collapse to either 0 with a probability
of jαj2 or 1 with a probability of jβj2 . The probabilities sum Figure 3. (A) The classical NOT gate, which inverts the input (A) to
� indicated in the adjacent truth table indicating
produce the output (A)
to 1 and therefore jαj2 þ jβj2 ¼ 1. classical register output. (B) The quantum Pauli X gate, which operates on
�
a quantum bit (qubit) in state jAi and yields the inverted state jAi,
Universal gate-based QCs analogous to the classical NOT operation. Both gates demonstrate the
Classical computers operate on traditional bits through oper­ inversion principle where an input of 0 yields a classical register output of
ations called logic gates, such as AND, XOR, and NOT.22 1, and vice versa. (C) Quantum circuit diagram illustrating the initialization,
manipulation, and measurement of a single qubit. The qubit Q starts in
Logic gates are a hardware abstraction layer that represent the ground state j0i, followed by the application of a Pauli-X gate (X)
the fundamental logic upon which classical computer algo­ which inverts the state to j1i. A measurement operation (M) is then
rithms are built.22 QCs can offer a similar layer of abstrac­ performed, the result of which is recorded in the classical register C. The
tion to leverage quantum states as a general-purpose register output is indicated as “1,” representing the classical bit obtained post-
measurement. (D) Quantum circuit diagram with the Hadamard (H) gate,
(small storage areas inside a processing unit that temporarily which encodes the input data into quantum superposition, resulting in a
hold data for rapid access during processing).20 Computa­ probability of results post-measurement based on the inputs and circuit
tions on quantum states are facilitated by linear operators composition.
1778 Journal of the American Medical Informatics Association, 2024, Vol. 31, No. 8

is used to encode the data into quantum superposition, the studies in this area may help identify datasets that better lev­
classical register from the circuit would yield a state of 0 erage QML algorithms and identify optimization opportuni­
approximately 50% of the time and 1 approximately 50% of ties for QML algorithms.34,37
the time (Figure 3D). There has also been research into reducing model complex­
Similar to how qubits are represented in vector form, quan­ ity and improving interpretability through quantum-based
tum gates and circuits can also be represented with linear feature selection.32,38,39 This work predominantly employs
algebraic matrices. The X-gate (ie, Pauli X Gate) can be writ­ quantum annealing (QA) for feature selection. For this
ten as: approach, correlations such as mutual information or Pear­
son correlation matrices are converted into quadratic uncon­
" # strained binary optimization (QUBO) problems, which
0 1
Pauli X � : represent the total energy of the system.40 QA can then be
1 0 leveraged to minimize this energy, where the optimal features
correspond to the system’s ground state.32,38,39 But much
Taken together with the qubit-vector representation, the like QML, the benefits of quantum feature selection in terms

Downloaded from https://academic.oup.com/jamia/article/31/8/1774/7700020 by guest on 22 October 2024

circuit described in Figure 3C would be written as: of performance, computational scalability, and feature stabil­
" #" # " # ity are yet to be fully established.32
0 1 1 0
j0i ¼ ¼ ¼ j1i: Genomics
1 0 0 1 As the use of high-throughput sequencing continues to
increase, so does the need for robust and scalable data ana­
More detailed examples of the mathematical representa­ lytics solutions for genomic data.41 Given the raw size and
tion of quantum gates and circuits can be found in several complexity of genomic data, it is an ideal candidate for
references and example circuits with their resulting output exploring QC-based solutions which may offer better accu­
simulated with software tools.6,24 racy and performance across many bioinformatics tasks.17
To this end, recent publications have demonstrated the suc­
Applications of quantum computing in cessful implementation of quantum-based algorithms for de
healthcare and biomedical research novo and reference-based DNA sequence assembly.41–44
Others have also implemented QC-based solutions for var­
Recent publications have increasingly described the applica­ iant detection,25 reconstruction of phylogenetic trees,45,46
tion of QCs in specialized areas of biomedical informatics, and the detection of epistatic interactions between genomic
such as ML, protein folding, and drug discovery.12,17 This single nucleotide polymorphisms.47 In these studies, the
section offers a practical overview of promising research ave­ authors demonstrate that QC-based methods are feasible and
nues, open questions, and quantum-based computational accurate but note the limited ability to scale due to the limited
methods in biomedical informatics. However, it is important size and lack of error correction in current QCs.41
to note, recent applications of quantum computing bioinfor­
matics are constrained by scale, as current QCs are limited in Protein folding
size and error correction capabilities, preventing them from Predicting 3-dimensional structures of proteins is a widely
handling full-sized datasets.25 Therefore, developing larger researched problem. Recent studies that have employed QC-
and more fault-tolerant QCs remains essential for achieving based algorithms typically frame this challenge as an optimi­
practical quantum advantage in these fields. zation problem. The objective of this problem is to identify
the ground state energy representing the most stable protein
Artificial intelligence and machine learning conformation based on a model of the protein’s interactions
Quantum computing offers potential benefits in predictive and conformations.17,48,49 Numerous publications have eval­
performance and computational scalability for artificial intel­ uated this approach and investigated the use of adiabatic
ligence (AI) and ML applications, particularly for tasks such algorithms, such as QA, and circuit-based algorithms, the lat­
as classification, regression, clustering, and feature selec­ ter of which often use variational quantum algorithms
tion.10,26,27 However, because classical ML algorithms can­ (VQAs).48–50 Collectively, these works have demonstrated
not be directly implemented on quantum hardware, the that subsampled protein folding problems can be successfully
development of quantum machine learning (QML) algo­ integrated and executed on modern QCs. These results are
rithms has become an area of focus in this emerging field.10 encouraging as they suggest potential speed enhancements
Over the past decade, quantum adaptations of classical algo­ over traditional optimization algorithms and are expected to
rithms have been described, such as quantum support vector motivate further research into scaling these methods to larger
machines (QSVMs),28 quantum neural networks (QNNs),29 protein models as advancements in quantum computing tech­
variational quantum classifiers (VQCs),30 and quantum ker­ nology occur.17,48,51,52
nel estimators,30 along with other QC-based algorithms with­
out clear classical analogs.31 Drug discovery
Recent applications of QML in biomedical and healthcare Computer-aided drug discovery (CADD) plays an integral
research include evaluating brain MRI images,32 predicting role in the drug discovery and development process, encom­
patient responses to cancer drugs,33 and detecting the pres­ passing tasks such as predicting drug-target interactions, per­
ence or absence of diseases.34–36 In studies that have com­ forming energy calculations, and conducting virtual
pared classical ML and QML, the performance of QML screenings.53,54 Given the fundamental properties of quan­
models are often similar or underperform, relative to classical tum mechanics, QC-based methods are seen as a promising
counterparts.32,34 It is suggested that future comparative avenue of investigation for enhancing currently available
Journal of the American Medical Informatics Association, 2024, Vol. 31, No. 8 1779

CADD approaches.55 A number of CADD tasks have cryptanalytically relevant QCs are unlikely to become avail­
recently been implemented on quantum-based hardware and able before 2030.70 Nonetheless, in response to these poten­
simulators.12 Lau et al investigated the use of QNNs for pre­ tial threats, governmental agencies have begun to develop
dicting mutational effects on drug binding properties, but and release 4 quantum-resistant cryptographic algorithms,
showed equivocal performance compared to classical neural with mandates to migrate federal systems to post-quantum
networks.5 In addition, variational quantum eigensolvers cryptographic methods by 2035.71,72 Despite these proactive
(VQEs) were used to compute protein-ligand activation ener­ steps, there is a theoretical risk that data, if encrypted and
gies; however, this was done using QC-simulators under stored today, could later be decrypted by more advanced,
noiseless conditions.56 Similarly, Kirsopp et al, also employed fault-tolerant QCs in future “harvest-decrypt attacks”.73
VQEs for calculating protein-ligand binding energies and suc­
cessfully implemented their approach on IBM and Quantini­
uum QC hardware, demonstrating comparable results with Quantum hardware and software: current
classical approaches.57 Additional studies have examined the state
use of quantum computing for predicting molecular docking The field of quantum computing has undergone remarkable

Downloaded from https://academic.oup.com/jamia/article/31/8/1774/7700020 by guest on 22 October 2024

configurations,58,59 QNNs for predicting molecular force advancement over the last decade, most notably evidenced by
fields,60 and hybrid quantum generative adversarial networks the development of universal gate-based QCs by multiple
(qGANs) and quantum variational autoencoders (qVAEs) to vendors including Microsoft, Honeywell, IBM, Google, and
generate small and large drug molecules, respectively.61 As IonQ.74–77 In recent years, these systems have been made
with many of the other recent applications of quantum com­ accessible via cloud-hosted platforms, which has enabled
puting, the work in this area remains exploratory and is often developers to deploy quantum algorithms on a diverse assort­
implemented in simplified systems to highlight prototypical ment of quantum hardware. The provision of cloud-based
use cases that work on available QC hardware.57 quantum computing resources presents a novel avenue for
exploration within biomedical research. Given the acceler­
Network analysis ated evolution of quantum computing, it will be essential for
Network analysis, a common framework used in systems researchers in their respective domains to identify potential
biology, is extensively leveraged in bioinformatics for pur­ applications and assess where it confers an advantage over
poses such as molecular modeling, mapping intracellular bio­ classical computing. Accordingly, this section provides a
logical pathways, and representing neuronal pathways.62–64 high-level overview of the hardware and software develop­
Detecting relationships among the constituents of complex ment kits (SDKs) currently available to the research commun­
systems has evolved into a dedicated field of study, known as ity to facilitate the acquisition of practical, hands-on
community detection.65 This area focuses on the accurate experience with QCs.
and efficient partitioning of networks into communities, lead­ In terms of quantum hardware, the current landscape of
ing to the development of graph partitioning algorithms to quantum computing can be divided into several paradigms
address this NP-hard problem.63,65 To date, several publica­ which sometimes overlap in their implementation: gate-based
tions have shown proof-of-concept implementations of quantum computing, QA, adiabatic quantum computing,
quantum-based algorithms for community detection. quantum simulation (with both gate-based and non-gate-
Recently, Negre et al, described the implementation of QA based variants), topological quantum computing, and
for community detection across various benchmark network measurement-based quantum computing.78 Gate-based
datasets, demonstrating that the quantum-based method was quantum computing implements operations via quantum
comparable to classical methods.63 In biological applications, gates, drawing parallels to how classical computing operates
Wierzbi�nski et al explored the use of D-Wave’s quantum with bits.6 QA and adiabatic quantum computing direct
annealers for community detection within brain connec­ quantum systems to evolve naturally under specified energy
tomes. They compared this approach to the Louvain Com­ configurations, circumventing the necessity for gates.79
munity Detection Algorithm and demonstrated superior Quantum simulation uses one quantum system to mimic
performance with QA when using higher modularity as a another and can be implemented with either gate-based or
proxy for cluster quality.66 non-gate-based approaches.19 Topological quantum comput­
ing leverages particular quasi-particles and their movement
Cryptography patterns, providing inherent resistance to errors owing to
In the United States, protected health information (PHI) is their unique properties.80 Conversely, measurement-based
defined and regulated primarily by the Health Insurance Port­ quantum computing conducts computations by performing a
ability and Accountability Act of 1996 (HIPAA) and the sequence of measurements on a pre-prepared, highly
Security Rule of 2003. These regulations mandate that PHI entangled state, also known as a cluster state.81 Despite the
be handled with care and confidentiality, often necessitating diversity of these paradigms, the close alignment of gate-
the use of cryptography. Accordingly, the development of based systems with classical computing principles has led to
quantum algorithms such as Shor’s and Grover’s, which can this being the predominant method employed and explored
solve mathematical problems foundational to many modern by many quantum computing companies and algorithm
public key cryptography algorithms, carries significant impli­ developers.74–76
cations for the security of healthcare data.12 Recent publica­ The availability of quantum computing resources varies
tions have explored the use of both gate-based QCs and QA across vendors, but universal gate-based QPUs are most com­
for integer factorization, which is a core mathematical princi­ mon. At the time of this writing, these QPUs typically possess
ple of Rivest–Shamir–Adleman (RSA)-based cryptographic between 50 and 100 qubits, with some implementations
methods.67–69 However, currently available QCs do not yet offering up to 400þ qubits (IBM Osprey).76 However, access
pose a meaningful threat to modern cryptography, and for most non-paid accounts is usually confined to QPUs with
1780 Journal of the American Medical Informatics Association, 2024, Vol. 31, No. 8

Classical Quantum
Application Common Classical Algorithms: Common Q-Algorithms: Shor's
Kruskal's Algorithm, Dijkstra's algorithm, Grover's algorithm,
Algorithm, Binary Search Algorithm, Variational Quantum Eigensolver
Software Algorithm Quicksort Algorithm

High-level Library/SDK Common Python libraries: Pandas, QISKit (IBM), Forest (Rigetti), Cirq
(Google), Quantum Development Kit
PyTorch, SciPy. (Microsoft), Ocean (D-Wave).
Low-Level Assembly
Increasing Abstraction

Machine Code x86 Assembly, ARM Assembly, RISC-V Quantum Assembly Language
Assembly, Microprocessor without (QASM), OpenQASM, Quantum
Interlocked Pipelined Stages. Instruction Language, Quantum
Instruction Set Architecture Intermediate Representation

Downloaded from https://academic.oup.com/jamia/article/31/8/1774/7700020 by guest on 22 October 2024

Micro Architecture Transistors are electrical components Qubits can be implemented using
used to implement bits - e.g., various devise, such as superconduct-
Hardware

metal-oxide-semiconductor ing circuits, trapped ions, or photonic

Logic Gates / Registers systems.

Devices Two-state quantum mechanical

low) across an electrical component. system (e.g., spin up/down of an
The abstraction of this is referred to as electron). The abstraction of this is
Physics a ‘bit’. referred to as a ‘qubit’.

Figure 4. Hierarchical layers of abstraction in computer systems, both classical and quantum. Each layer represents a step up from the physical
components that execute basic operations to the sophisticated applications that interact with users. It illustrates how quantum computing parallels and
diverges from classical systems, providing a visual comparison of their distinct computing models at each abstraction level. Figure adapted from Fu et al.89

fewer qubits, for instance 5-10. While paid accounts are approach to interact with the various gate-based systems. To
available, their cost is prohibitive for many academic endeav­ this end, QSDKs are designed to work both with specific
ors. In response to this, cloud providers extend a variety of quantum hardware and across platforms; however, the
academic subscriptions that allow researchers to secure dedi­ authors have observed that cross-platform compatibility and
cated time on smaller QPUs, in addition to enabling quantum performance is not universally consistent. There is hope that,
computer simulations on classical computers. These simula­ as QSDKs and specialized quantum libraries for methods
tors widen the accessibility of quantum development as they such as QML mature, this interoperability will continue to
typically have shorter queue times and are generally offered improve.
free of charge. Quantum simulators can be employed to test Overall, the current landscape of quantum computing is
and optimize algorithms iteratively using a selected quantum characterized by a diverse array of quantum hardware and
software development kits (QSDKs) prior to submitting jobs QSDKs. This diversity of resources, marked by differences in
to actual quantum hardware, considering that queue times design and computational paradigms, offers unique capabil­
for these shared and limited resources vary. ities and constraints that researchers can leverage. QA-based
Some vendors also provide access to non-gate-based QCs, QCs, for example, can possess a substantial number of
such as the QA platform offered by D-Wave (Advantage/ qubits, reaching up to 5000. However, not all computational
LEAP).82 Diverging from gate-based QCs, the QA systems do problems are readily adaptable to the format required by
not typically rely on circuits or logic gates and are tailored these non-gate-based systems. Gate-based QCs, conversely,
towards resolving optimization problems.83 Accordingly, are seen as offering a more adaptable computing framework
problems submitted to these quantum systems are repre­ that can function independently or in conjunction with an
sented as a specific mathematical function, such as a QUBO analog-like structure. Nonetheless, these gate-based systems
problem, that can subsequently be mapped onto the QPU. grapple with issues related to scalability and are currently
This mathematical function is then optimized through itera­ limited in the number of qubits they can support in compari­
tive sampling of the quantum system toward a target energy son to their non-gate-based counterparts. Ultimately, the
state. QA forms part of a subset of a larger quantum comput­ choice between gate-based and non-gate-based QCs hinges
ing paradigm known as adiabatic quantum computing.84 on the specific requirements of the algorithm or task in ques­
To interact with quantum hardware and build quantum tion, making the variety of available quantum computing
applications, QSDK and other layers of abstraction have resources a substantial asset to the field. The continued evolu­
been developed (Figure 4). These SDKs offer abstractions and tion of these systems and development of their accompanying
tools for interfacing with quantum hardware, enabling devel­ software tools promise to further expand the possibilities of
opers to manipulate quantum algorithms, gates, and circuits quantum computing research and applications.
in higher level languages, such as Python and .NET. Exam­
ples of QSDKs include IBM’s Qiskit, Microsoft’s Quantum
Development Kit (QDK), and Google’s Cirq.74,85–88 While Integrating quantum computing with today’s
the physical realization of qubits varies among vendors (eg, data science platforms
superconductor and trapped ion), the application of quantum Historically, CPUs fulfilled the role of executing general-
logic gates remains consistent, facilitating a universal purpose computing tasks, including those related to ML.
Journal of the American Medical Informatics Association, 2024, Vol. 31, No. 8 1781

However, as ML models evolved in complexity and conse­ addressing well-defined tasks like factorization and unstruc­
quently demanded increased computational capacity, the lim­ tured search.9,96 Recent studies have shown proof of concept
itations of relying solely on CPUs became apparent. In the applications that have theoretically achieved quantum
mid-2000s, ML researchers adapted graphics processing units advantage over classical computing approaches.19,97,98 How­
(GPUs) to substantially expedite the training process of deep ever, bridging the gap between theory and real-world use
neural networks, which had been a considerable impediment remains challenging.
in the advancement of ML models.90 In response to this, Lastly, quantum computing is an inherently multidiscipli­
manufacturers released specialized GPUs and tensor process­ nary field that necessitates collaboration among experts from
ing units (TPUs) tailored for ML applications.91 These inno­ computer science, mathematics, physics, and target applica­
vations provided more cost-efficient solutions for ML tasks, tion areas, including biomedical research and healthcare. The
enabling a greater capacity for model complexity and proliferation of cloud based QPUs and the provision of asso­
performance. ciated development environments hold the potential to foster
Just as the accessibility of GPUs for ML empowered organ­ collaborative efforts and information sharing across these
izations to train more complex models, the availability of diverse fields. However, widespread adoption of quantum

Downloaded from https://academic.oup.com/jamia/article/31/8/1774/7700020 by guest on 22 October 2024

QPUs will likely foster the development of more advanced computing hinges on numerous contributing factors, includ­
quantum algorithms, thereby catalyzing new applications ing the maturation of dedicated software tools, a deeper
and use cases. Since not every computation or algorithm will understanding of where and how quantum advantage can be
achieve higher speed or performance on quantum hardware, achieved, and the establishment of easily accessible quantum
QPUs should be viewed as an accelerator to CPU-based com­ computing infrastructures integrated with existing platforms.
puting and considered as complementary hardware, rather
than as a replacement for conventional computing systems.
Therefore, incorporating this emergent technology into Conclusion
increasingly modular system architectures will enable devel­ Quantum computing holds significant promise in addressing
opers to select the most suitable hardware based on the com­ certain challenges encountered by classical computing, partic­
putational tasks being devised. ularly concerning complex analyses and optimization.99 Tra­
ditionally, biomedical researchers have grappled with scaling
Challenges and opportunities complex algorithms, such as those simulating chemical reac­
tions or biological structures. The computational complexity
One of the primary impediments to the progress of quantum
of these analyses often overwhelms classical computing. The
computing lies in achieving and maintaining quantum coher­
sheer number of variables involved in these processes leads to
ence, or the quantum state, of qubits for extended periods.92
an exponential increase in the complexity of comprehensive
Physical qubits exist in a delicate state and are sensitive to
and accurate models.4,100 Classical simulation then becomes
environmental interference, which can introduce noise into
computationally intractable, often necessitating the adoption
the system.6 Factors like imperfections in the physical qubit,
of simplified models and/or approximations. In these scenar­
temperature fluctuations, and electromagnetic interference
can disrupt their state, leading to a phenomenon known as ios, quantum computing offers a theoretical advantage in
decoherence.6,92 Efficient error correction mechanisms are that it can potentially solve certain problems that are compu­
vital to the functionality of QCs. However, current error cor­ tationally intractable for classical computers.5
rection methods are resource-intensive, wherein some Despite the theoretical advantages of quantum computing
approaches require a significant number of physical qubits to compared to its classical counterparts, tangible demonstra­
protect a single logical qubit. Finally, building scalable quan­ tion of these benefits for real-world, previously intractable
tum systems that can effectively leverage quantum entangle­ problems remains limited. Nevertheless, the emergence of
ment and superposition to perform complex computations quantum computing resources presents an opportunity to
remains a substantial technical hurdle, limiting the number of bolster collaborative endeavors across varied domains of
qubits available in individual systems. expertise. By emulating the interdisciplinary approach
Given these challenges, there is a need for robust evalua­ adopted within AI and ML, there exists the potential to
tion methods and benchmarking tools to assess quantum broaden and deepen the application of quantum technology.
computing performance.93 The DiVincenzo criteria act as In parallel to the ongoing technological advancements,
foundational guidelines, outlining the essential requirements exploratory initiatives within quantum algorithm and appli­
for a practical and reliable quantum computer.94 These crite­ cation development may precipitate significant break­
ria emphasize the importance of coherence, scalability, error throughs in biomedical research and healthcare, potentially
rates, and the functionality of quantum gates, among other effecting profound changes in these fields.
aspects. Related to these criteria, benchmark metrics such as
quantum volume offer a comprehensive approach to provide
a unified measure of a device’s overall capability.95 Such met­
Author contributions
rics can be used to compare quantum computing systems by Thomas J.S. Durant and Wade L. Schulz conducted the litera­
normalizing the variations due to system design and func­ ture review and drafted the manuscript. Elizabeth Knight,
tional heterogeneities. Brent Nelson, Seung J. Lee, Dominic Walliman, Hobart P.
In addition to the current limitations of quantum hard­ Young, and Lucila Ohno-Machado, added critical discussion
ware, direct applications of quantum-based algorithms to points and edited the manuscript. Wade L. Schulz was the
real-world problems remain limited. Notable quantum algo­ principal investigator of this study; he provided the original
rithms, such as Shor’s and Grover’s algorithms, demonstrate idea, overall supervision of this study, and critical editing of
theoretical benefits over their classical counterparts by the manuscript.
1782 Journal of the American Medical Informatics Association, 2024, Vol. 31, No. 8

Funding New York, NY: Association for Computing Machinery;

2010:141–50.
This work was supported by National Institutes of Health 015. Au-Yeung R, Chancellor N, Halffmann P. NP-hard but no longer
(NIH) grant number 1OT2OD032742-01. hard to solve? Using quantum computing to tackle optimization
problems. Front Quantum Sci Technol. 2023;2:1128576. https://
doi.org/10.3389/frqst.2023.1128576
Conflicts of interest 016. Rasool RU, Ahmad HF, Rafique W, et al. Quantum computing
T.J.S.D. was a consultant for Roche, a diagnostics company for healthcare: a review. Future Internet. 2023;15(3):94. https://
(fees). W.L.S. was a technical consultant to HugoHealth, a doi.org/10.3390/fi15030094
personal health information platform (equity, fees); is a 017. Pal S, Bhattacharya M, Lee S-S, et al. Quantum computing in the
cofounder of Refactor Health, an AI-augmented data man­ next-generation computational biology landscape: from protein
folding to molecular dynamics. Mol Biotechnol. 2024;66
agement platform for healthcare (equity); is a consultant for
(2):163-178.
Detect, a point-of-care diagnostics company (equity, fees). B.
018. Howard M, Wallman J, Veitch V, et al. Contextuality supplies
N. is a consultant for Refactor Health, an AI-augmented data the ‘magic’ for quantum computation. Nature. 2014;510
management platform for healthcare (equity). H.P.Y. is a

Downloaded from https://academic.oup.com/jamia/article/31/8/1774/7700020 by guest on 22 October 2024

(7505):351-355.
consultant for Refactor Health, an AI-augmented data man­ 019. Daley AJ, Bloch I, Kokail C, et al. Practical quantum advantage
agement platform for healthcare (fees). S.D. consults for Roy­ in quantum simulation. Nature. 2022;607(7920):667-676.
alty Pharma (fees). 020. Hughes C, Isaacson J, Perry A, et al. Quantum Computing for
the Quantum Curious. Springer International Publishing; 2021.
https://doi.org/10.1007/978-3-030-61601-4
Data availability 021. Dirac PAM. A new notation for quantum mechanics. Math Proc
No new data were generated or analyzed in support of this Camb Phil Soc. 1939;35(3):416-418.
022. Mano MM, Ciletti MD. Digital Design (4th Edition). Prentice-
research.
Hall, Inc. 2006.
023. Singh S, Chawla P, Sarkar A, et al. Universal quantum computing
using single-particle discrete-time quantum walk. Sci Rep.
References 2021;11(1):11551.
001. Ladd TD, Jelezko F, Laflamme R, et al. Quantum computers. 024. IBM Quantum Learning. Accessed January 25, 2024. https://
Nature. 2010;464(7285):45-53. https://doi.org/10.1038/ learning.quantum.ibm.com/
nature08812 025. K€osoglu-Kind B, Loredo R, Grossi M, et al. A biological sequence
002. Alexeev Y, Bacon D, Brown KR, et al. Quantum computer sys­ comparison algorithm using quantum computers. Sci Rep.
tems for scientific discovery. PRX Quantum. 2021;2(1):017001. 2023;13(1):14552.
003. Feynman RP. Simulating physics with computers. Int J Theor 026. Brooks M. Towards quantum machine learning. Nature. Pub­
Phys. 1982;21(6-7):467-488. lished Online First: May 24, 2023. https://doi.org/10.1038/
004. Emani PS, Warrell J, Anticevic A, et al. Quantum computing at d41586-023-01718-2
the frontiers of biological sciences. Nat Methods. 2021;18 027. Houssein EH, Abohashima Z, Elhoseny M, et al. Machine learn­
(7):701-709. ing in the quantum realm: the state-of-the-art, challenges, and
005. Lau B, Emani PS, Chapman J, et al. Insights from incorporating future vision. Expert Syst Appl. 2022;194:116512. https://doi.
quantum computing into drug design workflows. Bioinformatics. org/10.1016/j.eswa.2022.116512
2023;39(1):btac789. 028. Rebentrost P, Mohseni M, Lloyd S. Quantum support vector
006. Nielsen MA, Chuang IL. Quantum Computation and Quantum machine for big data classification. Phys Rev Lett. 2014;113
Information. Cambridge University Press; 2010. (13):130503.
007. Cao Y, Romero J, Olson JP, et al. Quantum chemistry in the age 029. Schuld M, Sweke R, Meyer JJ. Effect of data encoding on the
of quantum computing. Chem Rev. 2019;119(19):10856-10915. expressive power of variational quantum-machine-learning mod­
008. Brooks M. Quantum computers: what are they good for? Nature. els. Phys Rev A. 2021;103(3):032430.
2023;617(7962):S1-S3. 030. Havl�ı�cek V, C�orcoles AD, Temme K, et al. Supervised learning
009. Grover LK. A fast quantum mechanical algorithm for database with quantum-enhanced feature spaces. Nature. 2019;567
search. In: Proceedings of the Twenty-Eighth Annual ACM Sym­ (7747):209-212.
posium on Theory of Computing—STOC ’96. ACM Press; 031. Sergioli G, Giuntini R, Freytes H. A new quantum approach to
1996:212–219. https://doi.org/10.1145/237814.237866 binary classification. PLoS One. 2019;14(5):e0216224.
010. Rani S, Kumar Pareek P, Kaur J, et al. Quantum machine learning 032. Felefly T, Roukoz C, Fares G, et al. An explainable MRI-
in healthcare: developments and challenges. In: 2023 IEEE Inter­ radiomic quantum neural network to differentiate between large
national Conference on Integrated Circuits and Communication brain metastases and high-grade glioma using quantum annealing
Systems (ICICACS). IEEE; 2023:1–7. https://doi.org/10.1109/ for feature selection. J Digit Imaging. 2023;36(6):2335-2346.
ICICACS57338.2023.10100075 033. Sagingalieva A, Kordzanganeh M, Kenbayev N, et al. Hybrid
011. Hoefler T, Haener T, Troyer M. Disentangling hype from practi­ quantum neural network for drug response prediction. Cancers
cality: on realistically achieving quantum advantage. 2023. (Basel). 2023;15(10):2705.
Accessed August 13, 2023. http://arxiv.org/abs/2307.00523 034. Gupta H, Varshney H, Sharma TK, et al. Comparative perform­
012. Fl€
other FF, Griffin PF. The state of quantum computing applica­ ance analysis of quantum machine learning with deep learning
tions in health and medicine. Res Dir Quantum Technol. 2023;1: for diabetes prediction. Complex Intell Syst. 2022;8:3073-3087.
e10. https://doi.org/10.1017/qut.2023.4 https://doi.org/10.1007/s40747-021-00398-7
013. Ravi GS, Smith KN, Gokhale P, et al. Quantum computing in the 035. Maheshwari D, Ullah U, Marulanda PAO, et al. Quantum
cloud: analyzing job and machine characteristics. In: 2021 IEEE machine learning applied to electronic healthcare records for
International Symposium on Workload Characterization ischemic heart disease classification. Hum-Centric Comput Inf
(IISWC). IEEE; 2021:39–50. https://doi.org/10.1109/ Sci. 2023;13:1-15. https://doi.org/10.22967/HCIS.2023.13.006
IISWC53511.2021.00015 036. Abdulsalam G, Meshoul S, Shaiba H. Explainable heart disease
014. Aaronson S. BQP and the polynomial hierarchy. In: Proceedings prediction using ensemble-quantum machine learning approach.
of the Forty-Second ACM Symposium on Theory of Computing. Intell Autom Soft Comput. 2023;36(1):761-779.
Journal of the American Medical Informatics Association, 2024, Vol. 31, No. 8 1783

037. Liu J, Liu M, Liu J-P, et al. Towards provably efficient quantum 058. Hamilton CS, Kruse R, Sansoni L, et al. Gaussian Boson sam­
algorithms for large-scale machine-learning models. Nat Com­ pling. Phys Rev Lett. 2017;119(17):170501.
mun. 2024;15(1):434. 059. Banchi L, Fingerhuth M, Babej T, et al. Molecular docking with
038. Choi J, Kim K, Park SH, et al. Investigation of factors regarding Gaussian Boson sampling. Sci Adv. 2020;6(23):eaax1950.
the effects of COVID-19 pandemic on college students’ depres­ 060. Kiss O, Tacchino F, Vallecorsa S, et al. Quantum neural networks
sion by quantum annealer. Sci Rep. 2024;14(1):4684. force fields generation. Mach Learn Sci Technol. 2022;3:035004.
039. Vlasic A, Grant H, Certo S. An advantage using feature selection https://doi.org/10.1088/2632-2153/ac7d3c
with a quantum annealer. QIC. 2023;23(13&14):1091-1104. 061. Li J, Alam M, Sha CM, et al. Invited: drug discovery approaches
040. Li RY, Di Felice R, Rohs R, et al. Quantum annealing versus clas­ using quantum machine learning. In: 2021 58th ACM/IEEE
sical machine learning applied to a simplified computational biol­ Design Automation Conference (DAC). 2021:1356–1359.
ogy problem. NPJ Quantum Inf. 2018;4:14. https://doi.org/10. https://doi.org/10.1109/DAC18074.2021.9586268
1038/s41534-018-0060-8 062. Ma’ayan A. Introduction to network analysis in systems biology.
041. Varsamis GD, Karafyllidis IG, Gilkes KM, et al. Quantum gate Sci Signal. 2011;4(190):tr5.
algorithm for reference-guided DNA sequence alignment. Com­ 063. Negre CFA, Ushijima-Mwesigwa H, Mniszewski SM. Detecting
put Biol Chem. 2023;107:107959. https://doi.org/10.1016/j. multiple communities using quantum annealing on the D-Wave

Downloaded from https://academic.oup.com/jamia/article/31/8/1774/7700020 by guest on 22 October 2024

compbiolchem.2023.107959 system. PLoS One. 2020;15(2):e0227538.
042. NałeR cz-Charkiewicz K, Nowak RM. Algorithm for DNA 064. Weidner FM, Schwab JD, W€ olk S, et al. Leveraging quantum
sequence assembly by quantum annealing. BMC Bioinformatics. computing for dynamic analyses of logical networks in systems
2022;23(1):122. biology. Patterns (NY). 2023;4(3):100705.
043. Boev AS, Rakitko AS, Usmanov SR, et al. Genome assembly 065. Fortunato S. Community detection in graphs. Phys Rep.
using quantum and quantum-inspired annealing. Sci Rep. 2010;486(3-5):75-174.
2021;11(1):13183. 066. Wierzbi� nski M, Falc�o-Roget J, Crimi A. Community detection in
044. Sarkar A, Al-Ars Z, Bertels K. QuASeR: quantum accelerated de brain connectomes with hybrid quantum computing. Sci Rep.
novo DNA sequence reconstruction. PLoS One. 2021;16(4): 2023;13(1):3446.
067. Wang B, Hu F, Yao H, et al. Prime factorization algorithm based
e0249850.
045. Ellinas D, Jarvis P. Quantum simulation of phylogenetic trees. on parameter optimization of Ising model. Sci Rep. 2020;10
(1):7106.
arXiv. 2011. https://doi.org/10.48550/arXiv.1105.1582
068. Bhatia V, Ramkumar KR. An efficient quantum computing tech­
046. Onodera W, Hara N, Aoki S, et al. Phylogenetic tree reconstruc­
nique for cracking RSA using Shor’s algorithm. In: 2020 IEEE
tion via graph cut presented using a quantum-inspired computer.
5th International Conference on Computing Communication
Mol Phylogenet Evol. 2023;178:107636. https://doi.org/10.
and Automation (ICCCA). 2020:89–94. https://doi.org/10.1109/
1016/j.ympev.2022.107636
ICCCA49541.2020.9250806
047. Hoffmann M, Poschenrieder JM, Incudini M, et al. Network
069. Rivest RL, Shamir A, Adleman L. A method for obtaining digital
medicine-based epistasis detection in complex diseases: ready for
signatures and public-key cryptosystems. Commun ACM.
quantum computing. medRxiv. 2023. http://doi.org/10.1101/
1978;21(2):120-126. https://doi.org/10.1145/359340.359342
2023.11.07.23298205
070. National Academies of Sciences, Engineering, and Medicine,
048. Robert A, Barkoutsos P, Woerner S, et al. Resource-efficient
Division on Engineering and Physical Sciences, Intelligence Com­
quantum algorithm for protein folding. NPJ Quantum Inf.
munity Studies Board, et al. Quantum Computing: Progress and
2021;7(1):38.
Prospects. National Academies Press; 2019.
049. Perdomo-Ortiz A, Dickson N, Drew-Brook M, et al. Finding
071. Young SD. National Security Memorandum on Promoting
low-energy conformations of lattice protein models by quantum
United States Leadership in Quantum Computing While Mitigat­
annealing. Sci Rep. 2012;2:571. https://doi.org/10.1038/ ing Risks to Vulnerable Cryptographic Systems. The White
srep00571 House. 2022. Accessed April 23, 2024. https://www.whitehouse.
050. Chandarana P, Hegade NN, Montalban I, et al. Digitized coun­ gov/briefing-room/statements-releases/2022/05/04/national-
terdiabatic quantum algorithm for protein folding. Phys Rev security-memorandum-on-promoting-united-states-leadership-
Appl. 2023;20(1):014024. in-quantum-computing-while-mitigating-risks-to-vulnerable-
051. Wong R, Chang W-L. Fast quantum algorithm for protein struc­ cryptographic-systems/
ture prediction in hydrophobic-hydrophilic model. J Parallel Dis­ 072. NIST Announces First Four Quantum-Resistant Cryptographic
trib Comput. 2022;164:178-190. https://doi.org/10.1016/j.jpdc. Algorithms. NIST. Published Online First: July 5, 2022.
2022.03.011 073. Ali A. A pragmatic analysis of pre- and post-quantum cyber
052. Fox DM, MacDermaid CM, Schreij AMA, et al. RNA folding security scenarios. In: 2021 International Bhurban Conference
using quantum computers. PLoS Comput Biol. 2022;18(4): on Applied Sciences and Technologies (IBCAST). 2021:686–692.
e1010032. https://doi.org/10.1109/IBCAST51254.2021.9393278
053. Bajorath J. Computer-aided drug discovery. F1000Res. 2015;4: 074. Google Quantum AI. Accessed February 10, 2024. https://quan­
F1000 Faculty Rev-630. tumai.google/
054. Sadybekov AV, Katritch V. Computational approaches stream­ 075. Azure Quantum j Azure Quantum Homepage. Accessed Febru­
lining drug discovery. Nature. 2023;616(7958):673-685. ary 10, 2024. https://quantum.microsoft.com/
055. Santagati R, Aspuru-Guzik A, Babbush R, et al. Drug design on 076. IBM Quantum Computing. Accessed February 10, 2024. https://
quantum computers. Nat Phys. 2024;20:549-557. https://doi. www.ibm.com/www.ibm.com/quantum
org/10.1038/s41567-024-02411-5 077. Quantum Honeywell. Accessed February 10, 2024. https://www.
056. Mensa S, Sahin E, Tacchino F, et al. Quantum machine learning honeywell.com/us/en/company/quantum
framework for virtual screening in drug discovery: a prospective 078. Gyongyosi L, Imre S. A survey on quantum computing technol­
quantum advantage. Mach Learn Sci Technol. 2023;4:015023. ogy. Comput Sci Rev. 2019;31:51-71. https://doi.org/10.1016/j.
https://doi.org/10.1088/2632-2153/acb900 cosrev.2018.11.002
057. Kirsopp JJM, Di Paola C, Manrique DZ, et al. Quantum compu­ 079. Jain S. Solving the traveling salesman problem on the D-wave
tational quantification of protein–ligand interactions. Int J Quan­ quantum computer. Front Phys. 2021;9. https://doi.org/10.3389/
tum Chem. 2022;122(22):e26975. fphy.2021.760783
1784 Journal of the American Medical Informatics Association, 2024, Vol. 31, No. 8

080. Stern A, Lindner NH. Topological quantum computation—from 093. Resch S, Karpuzcu UR. Benchmarking quantum computers and
basic concepts to first experiments. Science. 2013;339 the impact of quantum noise. ACM Comput Surv. 2022;54
(6124):1179-1184. (7):1-35.
081. Raussendorf R, Browne DE, Briegel HJ. Measurement-based 094. DiVincenzo DP. The physical implementation of quantum com­
quantum computation on cluster states. Phys Rev A. 2003;68 putation. Fortschr Phys. 2000;48(9-11):771-783.
(2):022312. 095. Cross AW, Bishop LS, Sheldon S, et al. Validating quantum com­
082. D-Wave Systems j The Practical Quantum Computing Company. puters using randomized model circuits. Phys Rev A. 2019;100
Accessed February 10, 2024. https://www.dwavesys.com/ (3):032328. https://doi.org/10.1103/PhysRevA.100.032328
083. Boixo S, Rønnow TF, Isakov SV, et al. Evidence for quantum 096. Shor PW. Polynomial-time algorithms for prime factorization
annealing with more than one hundred qubits. Nat Phys. and discrete logarithms on a quantum computer. SIAM Rev.
2014;10(3):218-224. https://doi.org/10.1038/nphys2900 1999;41(2):303-332.
084. Albash T, Lidar DA. Adiabatic quantum computation. Rev Mod 097. Arute F, Arya K, Babbush R, et al. Quantum supremacy using a
Phys. 2018;90(1):015002. programmable superconducting processor. Nature. 2019;574
085. OceanTM Developer Tools j D-Wave. Accessed February 10, (7779):505-510. https://doi.org/10.1038/s41586-019-1666-5
2024. https://www.dwavesys.com/solutions-and-products/ocean/ 098. Kim Y, Eddins A, Anand S, et al. Evidence for the utility of quan­

Downloaded from https://academic.oup.com/jamia/article/31/8/1774/7700020 by guest on 22 October 2024

086. Qiskit. GitHub. Accessed February 10, 2024. https://github.com/
tum computing before fault tolerance. Nature. 2023;618
Qiskit
(7965):500-505. https://doi.org/10.1038/s41586-023-06096-3
087. Bradben. Get started with the Quantum Development Kit—
099. Abbas A, Ambainis A, Augustino B, et al. Quantum optimiza­
Azure Quantum. 2024. Accessed February 10, 2024. https://
tion: potential, challenges, and the path forward. 2023. Accessed
learn.microsoft.com/en-us/azure/quantum/install-overview-qdk
February 10, 2024. http://arxiv.org/abs/2312.02279
088. PennyLane. Accessed February 10, 2024. https://pennylane.ai/
100. Cheng H-P, Deumens E, Freericks JK, et al. Application of quan­
089. Fu X, Rol MA, Bultink CC, et al. An experimental microarchitec­
tum computing to biochemical systems: a look to the future.
ture for a superconducting quantum processor. In: Proceedings
of the 50th Annual IEEE/ACM International Symposium on Front Chem. 2020;8:587143. https://doi.org/10.3389/fchem.
Microarchitecture. Association for Computing Machinery; 2017. 2020.587143
https://doi.org/10.1145/3123939.3123952 101. Sapova MD, Fedorov AK. Variational quantum eigensolver tech­
090. Witt S. How Jensen Huang’s Nvidia Is Powering the A.I. Revolu­ niques for simulating carbon monoxide oxidation. Commun
tion j The New Yorker. Accessed February 10, 2024. https:// Phys. 2022;5(1):199.
www.newyorker.com/magazine/2023/12/04/how-jensen- 102. Kim Y, Choi J, Kim B, et al. Association between Prefrontal
huangs-nvidia-is-powering-the-ai-revolution fNIRS signals during cognitive tasks and college scholastic ability
091. Wang YE, Wei G-Y, Brooks D. Benchmarking TPU, GPU, and test (CSAT) scores: analysis using a quantum annealing
CPU Platforms for Deep Learning. 2019. Accessed February 10, approach. arXiv. 2024. https://doi.org/10.48550/arXiv.2402.
2024. http://arxiv.org/abs/1907.10701 04287
092. Bravyi S, Cross AW, Gambetta JM, et al. High-threshold and 103. Fedorov AK, Gelfand MS. Towards practical applications in
low-overhead fault-tolerant quantum memory. 2023. Accessed quantum computational biology. Nat Comput Sci. 2021;1
February 10, 2024. http://arxiv.org/abs/2308.07915 (2):114-119.

# The Author(s) 2024. Published by Oxford University Press on behalf of the American Medical Informatics Association.

This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits
unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.
Journal of the American Medical Informatics Association, 2024, 31, 1774–1784
https://doi.org/10.1093/jamia/ocae149
Review