Health, Illness and Medicine – Ethical Principles versus Algorithms and AI Medical Biases in Pandemics
Ana María López Narbona, University of Málaga, Spain
The most unethical thing you can do is ration when you could have prevented it if you had better sharing and cooperation between institutions.
Arthur L. Caplan, Head of the division of medical ethics at NYU School of Medicine in New York City, which runs Bellevue Hospital (Los Angeles Times, March 19th, 2020).
Since the beginning of the pandemic, I was interested in knowing what would happen in the case of scarce medical resources. This is not something new as, every day, medical authorities need to decide who, among the many people waiting for an organ transplant, is the best receiver of an organ.
Furthermore, these kinds of decisions are not new or rare in less developed countries. Indeed, in Africa, these kinds of decisions are more present because these countries face pandemics, hungers, wars, and other humanitarian crises which mean that, together with their reduced systemic (health and economic) capacity, doctors and other medical authorities have to decide about such matters almost every day. Scholars, doctors, and intellectuals in these countries can teach us many things.
H1N1 and other pandemics, were just warnings of what was going to happen from November 2019 onwards. No country took the proper measures to cope with such a terrible crisis, either because such a crisis could not be foreseen, or because taking proper prevention measures would have been too expensive or too un- or anti-economical.
During our seclusion, the Covid-19 pandemic is providing us with a precious amount of time to rethink our society. Will we dare? The disastrous process of destruction of nature and of our own society seems to be in stand-by.
We should begin first of all by rethinking the nature of sharing and cooperation. Important as it is, this issue is not within the scope of this article. Instead, the main objective is to analyse ethical principles applicable to medical protocols (especially Intensive Care Unit - ICU - protocols) when resources are scarce.
A very important issue we need to address is the controversy between ethical principles versus algorithms in decisions made when resources are scarce. Many scholars are warning about biases in algorithms and Artificial Intelligence when applied to medical decisions of any kind.
This essay reflects on the biases of algorithms and the need to be more rigorous with the application of ethical principles in decision making in ICU during a pandemic.
Ethical Principles versus Algorithms and Artificial Intelligence
One important point to consider when working on a protocol of this kind is the assertion that the disease, which has caused the pandemic, is potentially reversible. However, the aim of reversibility may not be possible in a situation of resource scarcity, and, consequently, the allocation of resources becomes compulsory.
Having made this important statement, there are two main issues which condition the Covid-19 pandemic; first, the need of ventilators by those infected with Covid-19, and the amount of days or weeks that people will need ventilators. Together with these two first conditions, highly skilled human resources will also be in high demand.
The outbreak of a pandemic brings many other life-and-death decisions. Many other ethical questions arise on the basis of scarcity in pandemics. Life-sustaining treatments being withdrawn may also become an imperative issue. Ethical and clinical principles must be applied. One of the main clinical principles is the greater chance of therapeutic success.
However, there are also issues of civil rights. In some protocols, age and disability cannot be claimed as tie-breakers, as the law protects the equal dignity of every human being. In other protocols, an age-limit for admission to the ICU may ultimately need to be set, in order to maximise the benefits for the largest number of people. In any case, appropriate palliative care must always be provided to patients in cases where they cannot have access to ventilators.
Transcendental decisions must be taken when there is an outbreak and resources are scarce. Ethical and clinical principles and algorithms work as tie-break rules to determine to whom to apply life-sustaining resources.
These rules should always be updated and already should have been debated in society (in times of social peace). Indeed, ethical experts, physicians, politicians and the public must be consulted in advance and their opinions must be taken into account in the elaboration procedure. However, the most recent protocols have been drafted during the Covid-19 pandemic.
Journals showing how people feel medical decisions are unfair or alien to their cultural values and religious beliefs indicate how important it is to reach an agreement on the main ethical and clinical principles and even on the algorithms applicable in these transcendental circumstances.
But the compromise or consensus must be reached before bad times come, because society needs a calm environment to discuss important decisions, which will rule in times of a health crisis.
In this regard, ethical and clinical principles seem to be more transparent and, in general, they are the reflection of society’s values. In any case, all of us know that society’s values change. Inglehart [1, 2, 3] has prolifically analysed these changes in values.
Algorithms are more opaque when assigning priority scores to patients. In principle, they are based on underlying pathologies and age. But there are many biases, which scholars and human rights advocates have denounced. In the table below, the most used algorithms are the so-called SOFA (Sequential Organ Failure Assessment) and MSOFA (Modified Sequential Organ Failure Assessment) ones.
Some preliminary research is showing important differences in death rates depending on race, gender, and age. Beyond living conditions, what part of higher death rates are attributable to ethical (ethnical?) principles and biased algorithms? How to take into account the most excluded groups of society, like prostitutes, inmates in jails, and homeless people?
For Felländer-Tsai , bias and confounders must be managed because the origin of data on which the algorithms are based can lead to erroneous interpretations and potential damage. According to Oh et al. , the main issue to tackle is that randomised trials estimate average treatment effects for a trial population, but participants in clinical trials often are not representative of the patient population that ultimately receives the treatment with respect to race and gender. Johnson  warns that drugs and interventions are not tailored to historically mistreated groups; for example, women, minority groups, and obese patients tend to have generally poorer treatment options and longitudinal health outcomes. Artificial Intelligence shows disparities in health care. Chen et al.  showed statistically significant differences in error rates in ICU mortality for race, gender, and insurance type. ‘Once algorithmic bias is uncovered, clinicians and AI must work together to identify the sources of algorithmic bias and improve models through better data collection and model improvements’ (Miller ).
Crisis moments are not the proper time to address issues like the fairness of ethical principles or algorithms, because as Gordijn & Ten Have  suggest, in a public health emergency, you shift from a focus on individual patients to how society as a whole may benefit, and that is a big change from usual practice of care.
There remain some important issues to analyse once this health crisis is over. Are algorithms more efficient than ethical principles in taking decisions in moments of crisis? Are algorithms and ethical principles taking into account changing values in societies?
Regarding issues of ethical principles at an international scale, Africa is confronting a huge health crisis. These countries have not the same capacity as Western countries. An important economic contribution to make these countries able to cope with pandemics is still pending.
Furthermore, there are still imperialist biases in Western countries regarding Africa and African people. A deep controversy has arisen recently because French doctors proposed the testing of preliminary vaccines on African people, based on the argument that these countries have no masks, treatments, or ventilators.
Using less developed countries as laboratories for the most developed ones is not ethical. International ethical principles should also be discussed within international fora.
Ethical protocols for medical decisions in pandemics when resources are scarce must be updated and discussed involving all strata of society. Health crisis are not the proper moment to discuss ethical principles or algorithms.
Ethical principles should avoid becoming ethnical principles because societies are continually changing.
On the other hand, ethical principles must take into account the most excluded and vulnerable groups in society, such as minorities (‘racial’ and ethnic minorities, immigrants), prostitutes, inmates in jails, homeless people, and the elderly. These groups are more prone to be excluded from scarce resources in times of pandemics (or health crises) because of social gaps and inequalities.
At the international level, countries must also conform to agreed ethical principles for the sake of dignity and respect to everyone.
 Inglehart, R. F. (2008). ‘Changing values among western publics from 1970 to 2006’. West European politics, 31(1-2), 130-146.
 Inglehart, R. F. (2017). ‘Evolutionary Modernization Theory: Why People’s Motivations are Changing’. Changing Societies & Personalities. 1(2): 136-151.
 Inglehart, R. F. (2018). Cultural Evolution: People's motivations are changing, and reshaping the world. Cambridge: Cambridge University Press.
 Felländer-Tsai, L. (2020). ‘AI ethics, accountability, and sustainability: revisiting the Hippocratic oath’. Acta Orthopaedica, 91(1).
 Oh, S. S., Galanter, J., Thakur, N., Pino-Yanes, M., Barcelo, N. E., White, M. J., ... & Borrell, L. N. (2015). ‘Diversity in Clinical and Biomedical Research: a promise yet to be fulfilled’. PLoS Medicine, 12(12). e1001918.
 Johnson, K. S. (2013). ‘Racial and Ethnic Disparities in Palliative Care’. Journal of Palliative Medicine, 16(11): 1329-1334.
 Chen, I. Y., Szolovits, P., & Ghassemi, M. (2019). ‘Can AI Help Reduce Disparities in General Medical and Mental Health Care?’, AMA Journal of Ethics, 21(2): 167-179.
 Miller A. P. (2018). ‘Want Less-Biased Decisions? Use Algorithms’. Harvard Business Review. July 26.
 Gordijn, B., & Ten Have, H. (2015). ‘Disaster Ethics’. Medicine, Health Care and Philosophy, 18(1): 1-2.
Comment on this article – log in with your ESA username and password: a comment field will appear.