Artificial Intelligence (AI) Model to Predict the Risk of COVID-19 ICU Severity: A Pandemic Success Story

Sara Alshaya; Maryam Sayed Jaffar; Aji Gopakumar; Sheik Abdullah Jamal Mohideen; Vibhor Mathur; Sudheer Kurakula; Badshah Mukherjee

Sara Alshaya Data and Statistics Department (DSD), Emirates Health Services (EHS), Dubai, United Arab, Emirates.
Maryam Sayed Jaffar Data and Statistics Department (DSD), Emirates Health Services (EHS), Dubai, United Arab, Emirates.
Aji Gopakumar Data and Statistics Department (DSD), Emirates Health Services (EHS), Dubai, United Arab, Emirates.
Sheik Abdullah Jamal Mohideen Data and Statistics Department (DSD), Emirates Health Services (EHS), Dubai, United Arab, Emirates.
Vibhor Mathur Data and Statistics Department (DSD), Emirates Health Services (EHS), Dubai, United Arab, Emirates.
Sudheer Kurakula Data and Statistics Department (DSD), Emirates Health Services (EHS), Dubai, United Arab, Emirates.
Badshah Mukherjee SAS Middle East FZ, Dubai, United Arab Emirates.

Keywords: COVID-19 Severity, Prediction Model, ICU Motility Risk, United Arab Emirates

Abstract

Background: WHO considers a crucial indicator of infectious disease severity to be its mortality rate. With multiple COVID-19 waves, variant emergence, and case surges, timely prediction of severity risks for fatal outcomes in the UAE was essential. Amid pandemic challenges, prioritising critical care for high-risk patients was vital. This research on critical care severity risks adds significant value to current knowledge.
Objective: The plan was for the development of an AI model with high accuracy and estimation of patients at risk of severity across various characteristics.
Method: A retrospective cohort study design was used to conduct the research. Correlation and causation analytics were conducted using exploratory data analysis. A statistical risk scoring mechanism was built and combined with the other data attributes to build the AI model, then determined all the key factors determining the fatal outcome.
Results: Retrospective data spanning five months, including 71 variables from an EHS facility in Sharjah, was analysed. Among 783 positive PCR cases, less than one-fourth were found severe in the ICU admissions. The optimal model, using the Gradient Boosting Algorithm, demonstrated high accuracy (90%), with training and validation accuracies of 94% and 91%, respectively. The key severity risk factors identified included elevated ferritin levels, ventilator usage, high MCHC, and hypotension.
Conclusion: A highly interpretable machine learning model predicts severity risk in emergency care, and can contribute to the revision of EHS’s procedural manual and enable resource mobilisation, and effective care strategy.

How to cite this article:
Alshaya S, Jaffar M S, Gopakumar A, Mohideen
S A J, Mathur V, Kurakula S, Mukherjee B.
Artificial Intelligence (AI) Model to Predict the
Risk of COVID-19 ICU Severity: A Pandemic
Success Story. J Commun Dis. 2024;56(2):6-14.

DOI: https://doi.org/10.24321/0019.5138.202426

References

World Health Organization [Internet]. Global excess deaths associated with COVID-19, January 2020 - December 2021; 2022 [cited 2024 Feb 24]. Available from: https://www.who.int/data/stories/global- excess-deaths-associated-with-covid-19-january-2020- december-2021

Chen Y, Liu Z, Li X, Zhao X, Zhao J, Wu D, Xiao M, Zhang D, Gao P, Tian R, Xie J, Fan S, Li B, Xu Y, Li Y, Zhang S. Risk factors for severity due to COVID-19 in intensive care units: a single-center study. Ann Transl Med. 2021;9(4):276. [PubMed] [Google Scholar]

Al Saleh M, Alotaibi N, Schrapp K, Alsaber A, Pan J, Almutairi F, Abdullah M, Aboelhassan W, AlNasrallah N, Al-Bader B, Malhas H, Ramadhan M, Hamza M, Abdelnaby H, Alroomi M. Risk factors for mortality in patients with COVID-19: the Kuwait experience. Med Princ Pract. 2022;31(2):180-6. [PubMed] [Google Scholar]

Jamshidi E, Asgary A, Tavakoli N, Zali A, Setareh S, Esmaily H, Jamaldini SH, Daaee A, Babajani A, Kashi MA, Jamshidi M, Rahi SJ, Mansouri N. Using machine learning to predict mortality for COVID-19 patients on day 0 in the ICU. Front Digit Health. 2022;3:681608. [PubMed] [Google Scholar]

Aslam N. Explainable artificial intelligence approach for the early prediction of ventilator support and mortality in COVID-19 patients. Computation [Internet]. 2022 [cited 2024 Feb 22];10(3):36. Available from: https:// doi.org/10.3390/computation10030036 [Google Scholar]

Almuqbil M, Almoteer AI, Suwayyid AM, Bakarman AH, Alrashed RF, Alrobish M, Alasalb F, Alhusaynan AA, Alnefaie MH, Altayar AS, Alobid SE, Almadani ME, Alshehri A, Alghamdi A, Asdaq SM. Characteristics of COVID-19 patients admitted to intensive care unit in multispecialty hospital of Riyadh, Saudi Arabia: a retrospective study. Healthcare (Basel). 2023;11(18):2500. [PubMed] [Google Scholar]

Huang S, Vieillard-Baron A, Evrard B, Prat G, Chew MS, Balik M, Clau-Terre F, Backer DD, Dessap AM, Orde S,

Morelli A, Sanfilippo F, Charron C, Vignon P; ECHO- COVID Study Group. Echocardiography phenotypes of right ventricular involvement in COVID-19 ARDS patients and ICU mortality: post-hoc (exploratory) analysis of repeated data from the ECHO-COVID study. Intensive Care Med. 2023;49(8):946-56. [PubMed] [Google Scholar]

Pezoulas VC, Kourou KD, Mylona E, Papaloukas C, Liontos A, Biros D, Milionis OI, Kyriakopoulos C, Kostikas K, Milionis H, Fotiadis DI. ICU admission and mortality classifiers for COVID-19 patients based on subgroups of dynamically associated profiles across multiple timepoints. Comput Biol Med. 2022;141:105176. [PubMed] [Google Scholar]