“Artificial intelligence & COVID-19: (bio)ethical aspects of end of life

Bioethics; Artificial Intelligence; Coronavirus Infections; Decision Making; Decision Support Systems, Clinical; Palliative Care

Dear Editor,

The disease caused by the SARS-CoV-2 virus (COVID-19) has brought significant challenges to clinical medicine and public health¹ since its appearance in China during the month of December 2019. In fact, this novel virus - whose transmission mechanisms and natural history are still under investigation - has disseminated and grown exponentially regarding the number of infected people² , on a global scale. In this scenario, there is, naturally, the need to make decisions when faced with questions such as: (1) defining which patients will be given priority in intensive care units³ , ⁴ and (2) whether or not to submit COVID-19 patients to mechanical ventilation⁵ . The answers to these questions - that are closely related to (bio)ethical reflections - can be supported by computational techniques, especially those based on artificial intelligence (AI).

A study by Ouyang et al. ⁴ used mathematical modeling to determine what type of screening policy could be useful in ICUs during the SARS-CoV-2 pandemic. The article - whose aim was to find a heuristic to minimize the average global lethality rate over time - analyzed the circumstances in which patients could be placed in line, for admission to a hypothetical ICU with limited beds, or transferred to a general ward as the condition changes. The proposed heuristic worked satisfactorily in the simulation.

Another possible way of applying these techniques concerns the use of AI in determining personalized sedation and analgesia in the case of mechanical ventilation and extubation. In this context, Prasad et al. ⁵ used a reinforced learning algorithm - known as Q-learning - that suggested better decisions than those proposed by specialists regarding extubation time. AI techniques, especially those belonging to the machine learning area, allow for the construction and extraction of behavioral patterns implicit in decision histories, regarding a problem situation.

Assuming the existence of a database of previous decisions that comprise, for example, the type of care to be used at the end of life or the therapy to be prescribed to a patient, the patterns underlying this data can assist in choosing the most appropriate conduct to be adopted in each situation, provided it is properly extracted. In the current COVID-19 pandemic scenario, in which these decisions become even more pressing, this type of support can be of great value for ethically, more responsible, and fair conducts.

REFERENCES

• ¹
  Vieira JM, Ricardo OMP, Hannas CM, Kanadani TCM, Prata TDS, Kanadani FN. What do we know about COVID-19? A review article. Rev Assoc Med Bras. 2020;66(4):534-40.
• ²
  Gouveia CC, Campos L. Coronavirus disease 2019: clinical review. Acta Med Port. 2020. doi: 10.20344/amp.13957.
• ³
  Siqueira-Batista R, Gomes AP, Braga LM, Costa AS, Thomé B, Schramm FR, et al. COVID-19 e o fim da vida: quem será admitido na Unidade de Terapia Intensiva? Observatório COVID-19 Fiocruz. Rio de Janeiro: Fiocruz; 2020. 6p. [cited 2020 Jul 2]. Available from: https://portal.fiocruz.br/sites/portal.fiocruz.br/files/documentos/COVID-19_e_o_fim_da_vida_-_quem_vai_para_uti_27-5_.pdf
  » https://portal.fiocruz.br/sites/portal.fiocruz.br/files/documentos/COVID-19_e_o_fim_da_vida_-_quem_vai_para_uti_27-5_.pdf
• ⁴
  Ouyang H, Argon NT, Ziya S. Allocation of intensive care unit beds in periods of high demand. Operations Research. 2020;68(2):591-608. [cited 2020 Jul 2]. Available from: http://ziya.web.unc.edu/files/2019/04/Allocation_of_ICU_Beds-FINAL.pdf
  » http://ziya.web.unc.edu/files/2019/04/Allocation_of_ICU_Beds-FINAL.pdf
• ⁵
  Prasad N, Cheng L-F, Chivers C, Draugelis M, Engelhardt BE. A reinforcement learning approach to weaning of mechanical ventilation in intensive care units. arXiv preprint. arXiv:1704.06300. [cited 2020 Jul 2]. Available from: https://arxiv.org/abs/1704.06300#:~:text=A%20Reinforcement%20Learning%20Approach%20to%20Weaning%20of%20Mechanical,care%20of%20patients%20admitted%20to%20intensive%20care%20units
  » https://arxiv.org/abs/1704.06300#:~:text=A%20Reinforcement%20Learning%20Approach%20to%20Weaning%20of%20Mechanical,care%20of%20patients%20admitted%20to%20intensive%20care%20units

• Funding
  This work was supported by CNPq (National Council for Scientific and Technological Development).

• Erratum

  Regarding the article “Bioethical aspects of artificial intelligence: COVID-19 & end of life” with DOI number: http://dx.doi.org/10.1590/1806-9282.66.S2.5, published in Journal of the Brazilian Medical Association, 2020;66(SUPPL 2):), page 5, title changed:
  From: Bioethical aspects of artificial intelligence: COVID-19 & end of life
  To: “Artificial intelligence & COVID-19: (bio)ethical aspects of end of life

Publication Dates

• Publication in this collection
  21 Sept 2020
• Date of issue
  2020

History

• Received
  09 July 2020
• Accepted
  11 July 2020