Digital transformation to mitigate emergency situations: increasing opioid overdose survival rates through explainable artificial intelligence
| DOI | https://doi.org/10.1108/IMDS-04-2021-0248 |
| Published date | 12 October 2021 |
| Date | 12 October 2021 |
| Pages | 324-344 |
| Subject Matter | Information & knowledge management,Information systems,Data management systems,Knowledge management,Knowledge sharing,Management science & operations,Supply chain management,Supply chain information systems,Logistics,Quality management/systems |
| Author | Marina Johnson,Abdullah Albizri,Antoine Harfouche,Salih Tutun |
Digital transformation to mitigate
emergency situations: increasing
opioid overdose survival rates
through explainable
artificial intelligence
Marina Johnson and Abdullah Albizri
Feliciano School of Business, Montclair State University, Montclair, New Jersey, USA
Antoine Harfouche
Paris-Nanterre University, Nanterre, France, and
Salih Tutun
John M. Olin Business School, Washington University in St. Louis, St. Louis, Missouri, USA
Abstract
Purpose –The global health crisis represents an unprecedented opportunity for the development of artificial
intelligence (AI) solutions. This paper aims to integrate explainable AI into the decision-making process in
emergency scenarios to help mitigate the high levels of complexity and uncertainty associated with these
situations. An AI solution is designed to extract insights into opioid overdose (OD) that can help government
agencies to improve their medical emergency response and reduce opioid-related deaths.
Design/methodology/approach –This paper employs the design science research paradigm as an
overarching framework. Open-access digital data and AI, two essential components within the digital
transformation domain, are used to accurately predict OD survival rates.
Findings –The proposed AI solution has two primary implications for the advancement of informed
emergency management. Results show that it can help not only local agencies plan their resources for timely
response to OD incidents, thus improving survival rates, but also governments to identify geographical areas
with lower survival rates and their primary contributing factor; hence, they can plan and allocate long-term
resources to increase survival rates and help in developing effective emergency-related policies.
Originality/value –This paper illustrates that digital transformation, particularly open-access digital data
and AI, can improve the emergency management framework (EMF). It also demonstrates that the AI models
developed in this study can identify opioid OD trends and determine the significant factors improving
survival rates.
Keywords Digital transformation, Explainable artificial intelligence, Machine learning, Emergency
management framework, Opioid OD, Survival prediction
Paper type Research paper
1. Introduction
Digital transformation refers to the integration of digital technology into all areas of a
business or an organization, fundamentally changing their processes and operations (Vial,
2019). Digital transformat ion, artificial intelligen ce (AI) and machine learning (ML)
revolutionize various industry domains (Kapletia et al., 2019;Roscoe et al., 2019;Wamba
and Queiroz, 2020). Such disruptive technologies are garnering significant attention due to
their many advantages, including maturity speed, limitless possibilities and power of
transforming organizations (Bohr and Memarzadeh, 2020;Delmolino and Whitehouse, 2018).
Following the private sector, governments at all levels have started employing digital
transformation, AI and ML in order to deliver services and programs more productively,
transparently and cost-effectively (Luna-Reyes and Gil-Garcia, 2014). Many government
agencies collect and store data as part of their digital transformation efforts and encourage
IMDS
123,1
324
The current issue and full text archive of this journal is available on Emerald Insight at:
https://www.emerald.com/insight/0263-5577.htm
Received 21 April 2021
Revised 14 September 2021
Accepted 25 September 2021
Industrial Management & Data
Systems
Vol. 123 No. 1, 2023
pp. 324-344
© Emerald Publishing Limited
0263-5577
DOI 10.1108/IMDS-04-2021-0248
researchers to extract insights and create applications for data-driven decision-making
through AI (Luna-Reyes and Gil-Garcia, 2014).
The outbreak of the COVID-19 pandemic has given rise to a dramatic increase in the opioid
crisis, a disruptive situation that has significantly increased the number of weekly visits to
hospital emergency departments. Compared to prepandemic rates, opioid overdose (OD) has
increased by 45% in this context (Nissen, 2021). In pre-COVID times, the opioid epidemic was
already ranked by government agencies and health organizations as one of the worst public
health crises in a generation with far-reaching social and economic effects (Centers for
Medicare and Medicaid Services, 2018). Opioids are a class of prescription drugs used for pain
relievers, including not only oxycodone, hydrocodone, fentanyl and tramadol, but also the
illegal drug heroin (NIH, 2020). Opioid addiction is defined by a strong and compulsive urge to
use opioid drugs due to their ability to produce feelings of pleasure, even when they are no
longer required medically (MedlinePlus –US National Library of Medicine, 2020). Opioid OD
refers to taking high doses of opioids (WHO, 2020). Opioid OD is signaled by pinpoint pupils,
unconsciousness and breathing troubles and can cause death when the brain function of
regulating breathing is seriously damaged (WHO, 2020).
The death toll of opioid OD in the USA alone stood around 450,000 between 1999 and 2018
(CDC, 2020). Moreover, opioid OD deaths were four-folds more in 2018 vis-
a-vis 1999,
indicating an accelerating trend of the opioid epidemic (CDC, 2020). Globally, 115,000
individuals died of opioid OD in 2017 only, while the number of nonfatal OD cases was several
times higher than the fatal OD cases (WHO, 2020). The annual economic burden and societal
costs (e.g. healthcare and loss in wages) of the opioid epidemic are estimated to be as high as
$1.02 trillion in 2017 in the USA alone (Florence et al., 2021). While guidelines and
recommendations to address the opioid epidemic are available, the gap between suggested
courses of action and reality is significant (WHO, 2020). Therefore, government agencies,
private organizations and emergency response teams need to mitigate, plan and direct their
resources to reduce the impacts of the opioid epidemic.
There have been growing investments in AI interventions to combat the opioid-driven OD
epidemic plaguing the entire world (Ti et al., 2021). We aim to examine the use of digital
transformation –particularly open-source digital government data –and AI in helping
government agencies and organizations to address opioid OD incidents, which account for an
emergency crisis. More specifically, this study’s primary research objective is to investigate
the role of AI in emergency mitigation and preparedness efforts concerning opioid OD and
provide government agencies with the adequate relevant insights. Additionally, we address
the following research questions throughout this study: (1) What are the important factors
increasing the survival rates of victims after an OD incident? (2) How do these factors
contribute to OD survival rates?
To attain the research objective and answer the aforementioned research questions, we
constructan informationtechnology (IT) artifactusing the design scienceresearch paradigmas
an overarchingframework (Abbasi et al.,2012;Hevner et al.,2004). Based on the design science
research paradigm, this paper presentsan IT artifact (i.e. an AI-based solution)utilizing state-
of-the-art AI algorithms. We employ the emergency management framework (EMF)
(McLoughlin, 1985) as the kernel theory guiding the design of the IT artifact. To build this
AI-basedsolution, we curatea dataset that includes the OD incidentsof the past two years,from
the open dataportal of the US government. We then createan explainable AI framework thatis
consistentlycapable of predicting OD survivalrates and identifying importantvariables. This
research has two primary implications in medical emergency situations: (1) it can help local
agencies plan their resources for a timely response to OD incidents, thus improving survival
rates. (2) It can beharnessed by state and federal governments to identify geographical areas
with lower survival rates and their primary contributing variables while planning and
allocating long-term resources to increase survival rates.
Artificial
intelligence for
emergency
325
To continue reading
Request your trial