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 use, distribution, and reproduction in any medium, provided the original work, first published in JMIR AI, is properly cited. The complete bibliographic information, a link to the original publication on https://www.ai.jmir.org/, as well as this copyright and license information must be included.
Imbalanced health care resource distribution has been central to unequal health outcomes and political tension around the world. Artificial intelligence (AI) has emerged as a promising tool for facilitating resource distribution, especially during emergencies. However, no comprehensive review exists on the use and ethics of AI in health care resource distribution.
This study aims to conduct a scoping review of the application of AI in health care resource distribution, and explore the ethical and political issues in such situations.
A scoping review was conducted following the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews). A comprehensive search of relevant literature was conducted in MEDLINE (Ovid), PubMed, Web of Science, and Embase from inception to February 2022. The review included qualitative and quantitative studies investigating the application of AI in health care resource allocation.
The review involved 22 articles, including 9 on model development and 13 on theoretical discussions, qualitative studies, or review studies. Of the 9 on model development and validation, 5 were conducted in emerging economies, 3 in developed countries, and 1 in a global context. In terms of content, 4 focused on resource distribution at the health system level and 5 focused on resource allocation at the hospital level. Of the 13 qualitative studies, 8 were discussions on the COVID-19 pandemic and the rest were on hospital resources, outbreaks, screening, human resources, and digitalization.
This scoping review synthesized evidence on AI in health resource distribution, focusing on the COVID-19 pandemic. The results suggest that the application of AI has the potential to improve efficacy in resource distribution, especially during emergencies. Efficient data sharing and collecting structures are needed to make reliable and evidence-based decisions. Health inequality, distributive justice, and transparency must be considered when deploying AI models in real-world situations.
Global responses to COVID-19 are converging with the use of digital health and algorithms based on artificial intelligence (AI), impacting health care systems around the world [
Luengo-Oroz et al proposed that the application of AI during the COVID-19 pandemic can be broken down into 3 scales: molecular, clinical, and societal [
The application of AI to decision-making processes in health care systems significantly precedes the COVID-19 pandemic [
In the health care system, resource distribution is an essential issue for policymakers, as resources are always scarce [
During the COVID-19 pandemic, imbalanced health care resource distribution has been one of the central issues causing unequal health outcomes and political tension [
Health care resource distribution is determined by the supply-demand relationship, logistics, and governance structure [
Currently, there are no comprehensive reviews to provide an overall picture of the literature on the application of AI in resource distribution in health care settings, particularly with regard to societal and ethical aspects. This study aims to conduct a scoping review on the application of AI in health care resource distribution, particularly during the COVID-19 pandemic and to explore the ethics and implications of AI in health policymaking with regard to resource distribution.
This scoping review follows the framework proposed by Arksey and O’Malley [
Searches were conducted in MEDLINE (Ovid), PubMed, Web of Science, and Embase from inception to February 2022. The search featured 2 key terms: (1)
Inclusion and exclusion criteria were defined a priori. This scoping review includes qualitative and quantitative studies investigating the application of AI in health care resource allocation. Studies that are not relevant to AI or health care resource allocation were excluded, as were duplicate studies. The inclusion and exclusion criteria are summarized in
Selection was conducted in 2 steps. First, titles and abstracts were screened for topic relevance and study design. Second, full texts of the remaining studies were screened to check for eligibility. All of the study selection processes were conducted in EndNote X9 (Clarivate).
Inclusion and exclusion criteria.
Criterion | Inclusion | Exclusion |
Type of study | Qualitative, quantitative, mixed method, and review studies in peer-reviewed journals | Letters, comments, conference abstracts, editorials, and theses |
Language | English | All other languages |
Study variables | Includes (1) artificial intelligence/machine learning and relevant terms and (2) allocation of health care resources | Does not include (1) artificial intelligence/machine learning and relevant terms or (2) allocation of health care resources |
Study context | Health care resource allocation at either the population level or hospital level | All other resource allocation scenarios |
Selected studies were input into NVivo 12 (QSR International) for labeling and coding. Authors coded data of interest from the articles in NVivo 12 and extracted information regarding study author, study design, location, context, aim, main result, AI method under study, resource allocation situation, and policymaking relevance into a standardized Excel (Microsoft Corp) form.
We employed an inductive approach to summarize the results from the included studies. First, the selected papers were grouped into 2 types: (1) studies of model development and validation of AI-based algorithms applied to health care resource distribution, and (2) qualitative studies, theoretical discussions, and review studies of the application of AI in health care resource distribution. For studies of model development and validation, we extracted the study objectives, resource distribution situations, AI model input variables, and policy relevance. For studies in the second category, objectives, resource distribution situations, discussed topics, and policy relevance were extracted. We further divided the input variables of the studies of model development and validation into 2 predefined categories: (1) ecological variables or variables at the group level, which included variables depicting characteristics at the population level, such as infant mortality in a region, local economic development, or disease prevalence and incidence; and (2) individual variables, which included variables that define individual characteristics such as diagnosis and age.
In total, 298 studies were identified in 4 databases after removing duplicates. After 1 round of screening for titles and abstracts, 255 studies were excluded due to irrelevant topics and unsuitable study designs. This left 43 studies for full-text screening. Of these, 2 were excluded because they were not directly relevant to health care, 8 because they were not related to resource distribution, 7 because they did not feature applications of AI, and 4 because of an inappropriate study design. In the end, 22 studies remained for qualitative synthesis. The PRISMA flow diagram for study selection is presented in
PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flow diagram of the study. AI: artificial intelligence.
The characteristics of the included studies on model development are summarized in
Of the 9 studies, 4 focused on resource distribution at the health system level, including financial resources for public health in Brazil [
Characteristics of the included studies on model development and validation.
Reference | Objectives | Resource allocation situation | Input variables |
Rosas et al (2013) [ |
To construct a financial resource allocation model using an artificial neural network | Financial resources for public health in Brazil | Mortality characteristics, proportion of teenage mothers, proportion of inadequate prenatal care, fertility rate, Gini index, proportion of elderly people in the population, literacy rate, financing capacity per capita, percentage of people with income below half minimum wage, percentage of urban households with basic sanitation, and proportion of urban households served by garbage collection |
Belciug & Gorunescu (2015) [ |
To propose a bed allocation and financial resource utilization strategy through queuing modeling and evolutionary computation | Bed allocation and financial resource utilization in the geriatric department of a London hospital | Bed inventory, arrival rate, mean service time, patient flow parameters, and holding and penalty cost and other cost considerations |
Gartner & Padman (2015) [ |
To evaluate how early determination of diagnosis-related groups can be used for better allocation of scarce hospital resources | Hospital resources, including day resources and overnight resources (beds), validated in a mid-sized hospital near Munich, Germany | Primary and secondary diagnoses, clinical procedures, age, gender, and weight in newborns |
Velez et al (2016) [ |
To present an artificial intelligence–based health planning model based on data from geospatial systems | Health care resource distribution in health planning in Ecuador | Geospatial variables based on the social determinants of health and geospatial patterns of territorial distribution in the allocation of equipment, supplies, and health services in relation to the availability, accessibility, and need of the population |
Xu et al (2018) [ |
To propose a health resource allocation model based on mass customization to maximize revenue and customization | Allocation of doctors and other medical resources in a public hospital system in China | Distribution of medical stations, professional level of doctors (salary and seniority), patient preferences and illness severity, medical cost, and revenue |
Yousefi et al (2018) [ |
To present a model based on agent-based simulation, machine learning, and a genetic algorithm for allocation of medical resources in emergency departments | Medical resource allocation in a teaching hospital in the capital of State of Minas Gerais in Brazil | Number of receptionists in the reception area; number of triage nurses in the triage room; number of laboratory technicians in the laboratory and X-ray room; and number of doctors, nurses, and nurse technicians in the suturing yellow zone, orthopedics department, surgical department, and clinical emergency area. |
Zhang et al (2018) [ |
To propose a framework introducing a novel approach to multi-attribute decision-making problems in the picture fuzzy context | Medical resource allocation in the hierarchical medical treatment system in China | Patient diagnostic characteristics and hospital tiers |
McRae et al (2020) [ |
To present a clinical decision-support system and mobile app to assist in COVID severity assessment, management, and care | Resource allocation during COVID in New York, with validation data sets from Wuhan, China | Outpatient score (age, gender, diabetes, cardiovascular comorbidities, and systolic blood pressure) and biomarker score (C-reactive protein, procalcitonin, and age) |
Bednarski et al (2021) [ |
To study how reinforcement learning and deep-learning models can facilitate the redistribution of medical equipment during pandemics | Pandemics in the context of COVID | COVID risk factors by region, COVID mortality by region, and current demand for medical equipment |
The characteristics of studies involving reviews and theoretical discussions are summarized in
Characteristics of the included studies involving theoretical discussions, qualitative studies, or review studies.
Reference | Objective | Resource allocation situation | Reviewed/discussed methods for the application of AIa during the COVID-19 pandemic |
Rajkomar et al (2018) [ |
To explore how model design, biases in data, and interactions of model predictions with clinicians and patients exacerbate health inequalities | Intensive care unit and in-hospital stay length |
Suggested that future AI models for health care resource distribution should include principles of distributive justice. |
Laudanski et al (2020) [ |
To analyze the applications of AI during COVID using the WHOb framework of pandemic evolution | Global COVID-19 pandemic |
Reviewed cases in Italy where AI was used in studying computed tomography scans for COVID prognosis, and suggested that AI-driven scans can help predict prognosis and therefore allow better resource distribution. Discussed AI-driven triage based on patient characteristics and AI-supported health resource allocation and ethics. |
Adly et al (2020) [ |
To discuss the potential of using AI to prevent and control COVID | Global COVID-19 pandemic |
Suggested that the application of AI was valuable in medical resource distribution that included the parameters of patients and the pandemic. |
Bernardo et al (2020) [ |
To present approaches for using technology to facilitate resource distribution in disasters and outbreaks | Disasters and disease outbreaks |
Found that data collected from crowdsourcing and the human-technology interface could be used as data sources. |
Neves et al (2020) [ |
To discuss the basic principles of medical resource allocation choices during COVID | Global COVID-19 pandemic |
Discussed rationalization of care, medical and team conflict, modeling of the pandemic, and application of AI. Explored the use of AI as a support tool to streamline inventory control and standardize resource distribution. |
Xie et al (2020) [ |
To present an overview of the application of AI technology in ophthalmology, with a focus on deep-learning systems | Diabetic retinopathy screening |
Reviewed empirical considerations behind the formation of successful screening programs. Examined potential methods for health economics and safety analyses that can assess concerns regarding AI-based screening. |
Zou et al (2020) [ |
To present the COVID response of Shenzhen, China and discuss the potential of a successful model for COVID prevention and control | COVID-19 pandemic in Shenzhen, China |
Reviewed methods applied by Shenzhen, including early action and centralized response, care for vulnerable persons, community response teams, and technology. Discussed the integration of information technology in Shenzhen’s response, including mobile technology, big data, and AI. |
Basit et al (2021) [ |
To discuss the data sharing and collection process and the ethical considerations around pandemic data | Global COVID-19 pandemic |
Discussed the required data, failures and challenges in obtaining pandemic data, success in data access, model creation using data, and ethical challenges associated with data access during the COVID-19 pandemic. Discussed the application of AI in the allocation of intensive care resources and ventilators. |
Huang et al (2021) [ |
To investigate China’s health informatization, especially during the COVID-19 pandemic | COVID-19 pandemic in China |
Discussed the development of China’s health informatization from 5 perspectives: health information infrastructure, information technology applications, financial and intellectual investment, health resource allocation, and the standard system. |
Jain et al (2021) [ |
To discuss the implications of AI for employability by analyzing issues in the health care sector | Human resources in health systems |
Displayed hierarchical relationships between employability and a range of characteristics. Discussed measures that could potentially enhance employability in the health care sector through AI. |
Lu et al (2021) [ |
To establish barriers that affect medical information digitalization innovation and development through interviews and a literature review | Medical information digitalization |
Applied the importance-resistance analysis model and identified the resistant factors, including data sharing, infrastructure, regulation, and operations in the context of data privacy. Proposed several ways to overcome these limitations, including transparency regulation and infrastructure building. |
Pereira et al (2021) [ |
To present interindividual variability and the roles it plays in the variability of COVID presentation and susceptibility. | Global COVID-19 pandemic |
Reviewed the biological differences that contribute to variability in COVID manifestation. Reviewed efforts to use AI to integrate digital data to enable the identification of high-risk COVID-19 patients. |
Röösli et al (2021) [ |
To discuss possible bias in the application of AI during the COVID-19 pandemic | Global COVID-19 pandemic |
Discussed how COVID exacerbated racial and socioeconomic disparities. Explored how an AI-informed resource allocation strategy can be influenced by biases. |
aAI: artificial intelligence.
bWHO: World Health Organization.
The policy implications of studies on model development are relevant on 2 levels: (1) health system level [
Policy relevance of the included studies on model development and validation.
Reference | Policy relevance |
Rosas et al (2013) [ |
Divided municipalities in Brazil into quartiles of health care financial needs. Proposed that the selection of input variables should consider the vulnerability of the population, the true representation of the factors of need, political choice, and the availability of reliable data. |
Belciug & Gorunescu (2015) [ |
Provided tools to estimate the appropriate parameters for optimal resource utilization. Enabled the hospital manager to simulate scenarios to make the near-best decision. |
Gartner & Padman (2015) [ |
Provided decision-makers with information on admission and scheduling decisions. Offered an approach to integrate and analyze the financial objectives of health care delivery. |
Velez et al (2016) [ |
Facilitated the management of multidisciplinary information with the entire range of determinants of a specific context. Provided enough flexibility to allow the exploration of different complex circumstances in health planning. |
Xu et al (2018) [ |
Reduced costs by making doctors mobile. Addressed personal preferences, such as treatment time and the professional level of doctors. |
Yousefi et al (2018) [ |
Decreased the average length of stay in this emergency department case study by 14%. Provided a framework to efficiently combine simulation and metamodels in the health care industry. |
Zhang et al (2018) [ |
Facilitated decision-making to divide patients under different conditions into different levels of hospitals in the hierarchical medical treatment system. |
McRae et al (2020) [ |
Supported the validity of a clinical decision support system and mobile app Provided tools to be deployed to community clinics and sites for decision support. |
Bednarski et al (2021) [ |
Facilitated officials managing future public health crises. Improved algorithm performance for future applications. |
Policy relevance of the included studies involving theoretical discussions, qualitative studies, or review studies.
Reference | Policy relevance |
Rajkomar et al (2018) [ |
Proposed that the principles of distributive justice be incorporated into model design, deployment, and evaluation. |
Laudanski et al (2020) [ |
Suggested that AIa can couple outbreak data with measures of potential demand and direct supplies more efficiently. |
Adly et al (2020) [ |
Found that no study had been published on the application of AI in medical resource distribution during the COVID-19 pandemic as of 2020 and that such studies are required to inform policy decisions. |
Bernardo et al (2020) [ |
Suggested that automation by AI and machine learning can further our abilities in predictive analytics. |
Neves et al (2020) [ |
Emphasized that the ethical values for the rationing of health resources in an epidemic should converge with basic ethical values and that transparency is essential to ensure public trust. |
Xie et al (2020) [ |
Proposed that technical feasibility and patient acceptability must be assessed for AI to be deployed in real-world settings, and that health professionals’ acceptance and interpretability of AI-based screening strategies must also be assessed. |
Zou et al (2020) [ |
Proposed that the model adopted in Shenzhen, including multisectoral coordination, proactive contact tracing and testing, timely isolation and treatment, hospital infection control, effective community management, and prompt information dissemination, could be a potential model for other cities around the world for containing the pandemic. |
Basit et al (2021) [ |
Proposed that informaticians globally should continue collecting, recording, and analyzing data with the intent of gathering new knowledge and translating it into a better, faster, and more successful response to the next pandemic. Suggested that professionals must come together to develop ways to collect, standardize, and disseminate the data needed to make necessary decisions. |
Huang et al (2021) [ |
Suggested that China’s health informatization needs to strengthen top-level design, increase investment and training, upgrade health infrastructure and information technology applications, and improve internet-based health care services. |
Jain et al (2021) [ |
Proposed that an AI intervention could impact the employability of the workforce through operational and training changes, and therefore impact human resource distribution in health. |
Lu et al (2021) [ |
Provided a basis for the future development directions of medical information digitalization and its impacts on health care and health systems. |
Pereira et al (2021) [ |
Suggested that predicting which COVID-19 patients will develop progressive diseases that require hospitalization has important implications for clinical trials targeting outpatients. |
Röösli et al (2021) [ |
Proposed that transparency in reporting of AI algorithms is necessary to understand intended predictions, target populations, hidden biases, and class imbalance problems. |
aAI: artificial intelligence.
China and Brazil are both developing countries with a similar per capita gross domestic product (China: US $10,435 and Brazil: US $6797) [
Rosas et al [
Zhang et al [
Overall, the allocation of medical resources based on the models from the 3 studies demonstrated that the key considerations proposed by studies from China were the hospital tier system, the professional level of doctors, the geographical distribution of medical resources, and cost-effectiveness [
In this review, we compiled evidence on the application of AI in health resource distribution, especially regarding COVID-related policy. After synthesizing 22 articles, we found that AI-based models were proposed at both hospital (secondary care in inpatient settings) and health system (public health) levels and that theoretical discussions and reviews focused on the potential for AI to improve the efficacy of resource distribution and on the ethics of applying AI in health resource distribution. Two major themes emerged from the review. First, we found that AI-informed resource distribution strategies are impactful for health access and equality. Second, the approaches can be categorized ideologically into revisionist and conservative groups.
AI and machine learning have considerable potential to improve efficacy in resource distribution, especially during emergencies, such as the COVID-19 pandemic, where quick decisions are required based on evolving situations [
However, very large amounts of data are necessary for AI algorithms to make reliable and evidence-based decisions [
The included articles highlighted the importance of distributive justice and transparency in AI model design. The analysis conducted by Rajkomar et al emphasized that machine learning systems should be used proactively to advance health equality [
The build-up of AI models and implementation plans can be broadly categorized into revisionist and conservative approaches. In revisionist approaches, the models aim to revise the disparity in resource distribution by actively correcting the biases in previous decision-making processes. For example, the models proposed by Rosas et al [
However, despite the revisionist model proposed by Brazilian academics [
This is one of the first reviews to incorporate all available evidence qualitatively and provide a comprehensive picture of the model development and theoretical discussion on AI in medical resource distribution. Our results contribute to the ongoing discussion of applying AI in medical resource distribution and add novel insights into the social and ethical implications. Nonetheless, this study has several limitations. First, due to the scope of the study, we focused on published journal articles but did not examine policy documents or grey literature. This could have led to incompleteness in the collected information. Further studies could examine policy statements and grey literature to better understand intercountry differences. Second, we included only articles published in English and therefore might have overlooked publications in other languages. Third, there are potential sources of meaningful heterogeneity in this scoping review, including the diverse use of AI technologies, different study designs, and different locations. The analyses in this study could be affected by such heterogeneities. Fourth, this study is a qualitative overview of the general application of AI in health care resource distribution and is exploratory. We did not compare different levels of resource distribution and distinguish various machine learning methods in detail. Further studies are needed to explore and contrast different AI approaches at various resource distribution levels in detail. Lastly, due to the availability of evidence, we only compared studies from China and Brazil. We were only able to compare the differences between the 2 countries based on a few studies, which could not represent the real situation in either country. The comparison should be interpreted as exploratory and demonstrative.
This scoping review synthesized evidence on the application of AI in health resource distribution, particularly during the COVID pandemic. The included studies suggested that AI and machine learning have high potentials to improve efficacy in resource distribution, especially during sudden and evolving situations. A coordinated and continuous data sharing and collecting mechanism is needed for better data input so that AI can make reliable and evidence-based decisions. Various issues, including health inequality, distributive justice, and transparency, should be considered when deploying AI models. Such considerations are required for implementing revisionist AI models that can correct distribution inequality in actual policy processes.
artificial intelligence
None declared.