Ethics

What are the core guidelines of medical ethics?

• Beneficence — act with the best intentions of the patient
• Non-maleficence — protect the patient from harm
• Autonomy — help the patient understand what is happening to them and ensure they are involved in all decisions about their care
• Justice — ensure all patients of equal need are treated in the same way

How are these principles upheld in physical medicine?

In physical medicine, there are lots of rules that help doctors meet these ethical requirements. For example, when a patient is having an operation, they have to give informed consent (told what will happen, the risks and benefits), and the operation should be scheduled according to priority (patients with the greatest need get to the operating room first). This practice upholds all four medical ethics principles.

‍How do these principles apply to AI?

In the context of AI, it can sometimes be harder to interpret the principles. What does it mean, for example, to cause no harm to a person if nothing is physically happening to their body? This does not mean interpreting the "Do No Harm" principles is impossible, and abiding by ethical principles is still very important for all those involved in developing and using AI for healthcare.

Beneficence

To act with a person's best interests at heart means that all aspects of a person's life should be considered when a medical decision is made. Just because a drug has proved to be the most effective at treating a particular type of cancer, does not mean it is the best drug for that patient if the side effects are incompatible with that person's life. If an algorithm is used to advise doctors on how to treat a patient, it is important that a wide range of factors — such as their priorities, where they live, and what their home life is like — are taken into consideration as well.

More broadly, beneficence refers to the duty of healthcare providers to both prevent/remove harm and promote wellbeing. This involves more than "just" identifying a diagnosis that fits a list of quantifiable symptoms and matching it to an "effective" drug, or identifying potential risk factors. Beneficent care also involves seeing the person as a whole (taking into account their personal beliefs and values), shared decision-making, and providing care in an empathetic, compassionate, and trustworthy manner.

Whilst AI might be able to mimic human empathy, it cannot truly "understand" it and therefore might not be able to completely replicate its effects. Furthermore, there is growing concern that AI's reliance on "quantifiable" data might lead to the exclusion of other "data" about a patient's life from the decision-making process. For these reasons, it is important to see AI as a helpful aid, but not a replacement for clinicians who are capable of contextualising "evidence" and focusing on the more "holistic" aspects of care.

Non-maleficence

To protect patients from harm in the context of AI means to protect them from the harms of privacy infringement, overdiagnosis, or psychological harm.

• Privacy — Medical information is very sensitive. If it is leaked to the wrong person or organisation, significant harm could come to that patient — for example, they could be denied insurance or experience discrimination. It is critical that their privacy is protected at all times when any form of patient data is being used in the context of AI.
• Overdiagnosis — The process of diagnosing a patient can sometimes be tiring and painful. Patients should only be subject to tests (e.g. blood tests or biopsies) when it is absolutely necessary and it's known that a diagnosis will lead to better outcomes. Algorithms should not be used to "screen" patients for abnormalities or conditions that may cause no harm if left untreated, as the harm from the process of diagnosis might be worse.
• Psychological harm — A person can also experience psychological harm if, for instance, they are informed that they might one day develop a life-threatening disease, causing them to worry or feel excessively anxious. When algorithms are used to predict a patient's risk of developing diseases, the impact on their psychological wellbeing should be factored in. If a person is predicted to be at high risk of developing a disease and nothing can be done to reduce that risk, it might be best not to tell the patient and instead advise them to come for regular testing.

Non-maleficence is the principle most closely linked to the Hippocratic "Do No Harm" oath, and the concerns raised by AI mostly stem from its ability to do more harm than good — whether through privacy infringement, overdiagnosis, or enabling healthcare to be unethically manipulated by economic and market forces.

Autonomy

To ensure patients are able to exert control over their own healthcare, and to know what is happening to them, there should be transparency around who has had access to their data and for what purpose. This enables patients to give informed consent to any form of AI-powered diagnosis or treatment recommendation, and to be able to understand how an algorithm has reached a decision about their care.

Autonomy broadly refers to the ability of a person to make their own life choices, protected or harmed by their ability (or inability) to self-govern free from external control and undue interference. It is now widely seen as the "primary principle" in modern medicine, tied to the shift from "paternalistic care" toward "patient-centred care."

AI relies on large volumes of data, including data patients may collect themselves (e.g., from smartwatches or shopping records), and it can predict risk with the intention of encouraging preventative action. Without careful thought, this gives AI significant potential to nudge and police individuals into behaving in ways that do not necessarily align with their own personal values, in the name of pursuing "optimum health." The fact that much of this nudging happens within a black box, evaluated against often inscrutable baselines, amplifies its potential negative impact on autonomy.

Although AI can be empowering (such as learning from smartwatch use), a future of near-continuous, unobservable screening may impinge on a person's right to informed and meaningful consent, along with their right to "not know" — if they believe certain health information (such as future risk) might cause them psychological harm.

Justice

If the provision of healthcare is to be just, it needs to be fair. In the context of AI, this means being aware of data bias. If an algorithm is trained on biased data, it will likely be unfair because it may be more accurate for some patients than for others — which could cause problems if, for instance, one group of patients were regularly being misdiagnosed.

Justice is the most familiar of these principles in the public domain, given its close tie to bias. The concern is that issues with how medical training data is collected, curated, and interpreted may lead to biased algorithms which, over time, might lead to discrimination. While most public focus has been on bias tied to sex, gender, or race, there are lesser-known problems too: precision medicine (enabled by AI) has the potential to divide the population into "good patients" (those who respond well to treatment and act on preventive advice) and "bad patients" (those who don't) — a form of latent bias that develops over time, potentially amplifying the inverse care law, where those in greatest need of care are least able to access it.

If the widespread implementation of AI is to be successful, it will be necessary to question any assumption that algorithms are inherently more objective than humans, and to build mechanisms for identifying the sources and consequences of bias.

Is medicine only governed by rules, regulations, and standards?

No. Medicine, despite being one of the most highly regulated fields in existence, is not purely governed by rules, regulations, policies, and standards. It also has a long history of ethical governance — from the Hippocratic Oath to "do no harm," to the introduction of bioethics principles, to more recent medicine-adjacent ethical interventions such as the Bermuda Principles, intended to govern human genome sequencing.

It's necessary to ensure the introduction of AI is also subject to rigorous ethical analysis, alongside technical, regulatory, and sociocultural analysis. This can be done by applying the expanded list of bioethics principles — autonomy, beneficence, non-maleficence, justice, and explainability — to the analysis of AI ethics in healthcare, and by considering the broader value-based implications.