AI Ethics in Mental Healthcare

Users bring their hopes and fears to their experiences with AI.

What if this takes my job away?

What does it do with all this information that I’m sending it anyway?

How does this technology even work?

But like with any fear, taking a courageous posture and moving forward with curiosity tends to provide relief.

To this end, the potential solution for these fears might be surprising: ethics. Ethics and morality – everyone’s favorite subject. Right? For many people, ethics was that class you sleepily strolled through before moving on to more “practical” courses.

But with the advent of artificial intelligence (AI), the field of ethics is taking on a more pronounced and practical relevance in the daily lives of its users. Paralleling the concerns that exist with social media use, people are more-and-more interested in how to use these technologies responsibly, so that we can avoid potential unforeseen negative outcomes.

So, our first question: what does ethical use of AI look like?

The European Commission’s 2019 publication of Ethics Guidelines for Trustworthy Artificial Intelligence gave us working criteria for “trustworthy” AI: (a) lawful, (b) ethical, and (c) robust. This is strengthened by Hanna et al.’s (2024) scholarship, which highlights how modern ethical principles (e.g., beneficence, nonmaleficence, etc.) can be specifically applied to the development of Medical AI systems. Taken together, this creates a vision for what ethical Medical AI might look like: transparent systems that promote the common good, reduce potential risk, and support human autonomy in how they are used.

As you continue to have interactions with AI, here are four ethical principles to help guide your interactions.

Principle 1 - Bias & Fairness

Question: To what extent is AI perpetuating or eliminating existing biases?

Humans have biases – and AI models are only as effective and unbiased as the humans that are developing them. The limitations that can be present within AI systems often reflect the societal biases/limitations that human developers/users bring to their interactions with AI (Hanna et al., 2024). These limitations are then carried forward into how the AI systems operate, can cause them to deviate from their intended function.

Principle 2 - Transparency & Explainability

Question: To what extent must we understand AI’s reasoning?

Explainability: Machine/deep learning often presents surprising or unintuitive conclusions. It’s important that individuals (particularly medical providers) seek specialty training in how best to interpret AI outputs. This parallels the “Digital Skills Divide” that emerged during the advent of the Internet, smartphones, and social media – where people varied widely in their comfort with using these technologies.

Harm-Reduction: Particularly relevant to clinical care is that AI systems clearly demonstrate their methods for harm-reduction, such as detecting suicidal behaviors and de-escalating crises.

Disclosure: Financial disclosures for individuals and companies manufacturing AI technologies are critical. This includes outlining the system development team, funding/business model, AI training methodologies, and the primary beneficiaries of the model.

Principle 3 - Privacy & Data Security

Question: How does AI protect the data input once it enters the system?

How would you feel if ANY data that went into your medical record was analyzed by data scientists doing work on AI? People will have different reactions to this, but the reality is: this work is already being done, at least in some organizations. It is essential that mental healthcare data is adequately secured across its lifecycle (including collection, storage, processing, sharing). Users are also cautioned about what information they share with AI systems, since it could become training data.

Principle 4 - Human Autonomy & Control

Question: To what extent can AI make autonomous decisions without human input?

Because of the speed and convenience at which AI answers prompts, we need to guard against becoming overly confident in the results we are given. Just because the AI’s results may look impressive, that does not necessarily mean that it’s accurate. As AI technologies continue to become integrated into clinical practice, we must continue to support patient decision-making authority and protect clinician agency regarding treatment recommendations. Organizations should consider how to make explicit, public-facing policies on how they maintain human oversight within their use of AI systems.

Are these perfect solutions? No – these systems are only as good as the people that develop and use them. But hopefully, this post gives you some hope – that we can have a say with how we interact with AI and move towards an ethical future.

Be sure to join our session on the ethics of AI at the International OCD Foundation's upcoming virtual training, Strengthen Your OCD Practice: Ethics, AI, and Options for Care, on Saturday, Dec. 13, from 11:00am–6:00pm ET.

This blog was a collaborative effort by members of the IOCDF and ADAA Artificial Intelligence/OCD Special Interest Groups.

If you are interested in joining the International OCD Foundation Artificial Intelligence / OCD SIG, please complete this interest form to receive meeting information and updates. This is a unique opportunity to have a decision-making voice within AI applications in OCD research & treatment. More information about Special Interest Groups is available here.

If you are interested in joining ADAA's Artificial Intelligence (AI) Special Interest Group, ADAA members can email sigs@adaa.org to sign-up. Not an ADAA member? Learn more about member benefits and join today!

Most helpful to everyone:

Pulling Back the Curtain: An Introduction to Artificial Intelligence from the AI Special Interest Group

“Ethics Guidelines for Trustworthy AI” by the European Commission: https://digital-strategy.ec.europa.eu/en/library/ethics-guidelines-trustworthy-ai

Most helpful to researchers:

Hanna, M. G., Pantanowitz, L., Jackson, B., Palmer, O., Visweswaran, S., Pantanowitz, J., Deebajah, M., & Rashidi, H. H. (20254). Ethical and bias considerations in artificial intelligence/machine learning. Modern Pathology, 38(3), https://doi.org/10.1016/j.modpat.2024.100686

Search Blogs

Filter by:

What are the causes of depression?

There are several possible causes of depression, and they often interplay with each other. Generally, biological or genetic, psychological and/or environmental factors, or a combination of these, are thought to underlie depression.
Depression can be a result of a primary psychiatric condition, or it can be secondary to a medical condition. Depression may be caused by use of certain substances, such as alcohol.
Some of the medical conditions which can cause depression, are hypothyroidism, certain types of cancer such as pancreatic cancer, Parkinson’s disease, and others.

What is Major Depressive Disorder (MDD)?

The most commonly diagnosed form of depression is Major Depressive Disorder (MDD). In 2017, around 17.3 million aged 18 years or older in the U.S. had experienced at least one major depressive episode in the last year, which represented 6.7 percent of all American adults. Depression is the leading cause of disability in the United States among people ages 15-44. ²

Major depression is characterized by an overwhelming feeling of sadness or a loss of interest and pleasure in most usual activities.

The other symptoms that are associated with major depression include:

decrease or increase in appetite,
insomnia or hypersomnia,
psychomotor agitation or retardation,
constant fatigue,
feelings of worthlessness or excessive and inappropriate guilt,
recurrent thoughts of death and suicidal ideation with or without specific plans for committing suicide, and
cognitive difficulties, such as, diminished ability to think, concentrate and take decisions.

The symptoms must persist for two weeks or longer and represent a significant change from previous functioning. Social, occupational, educational, or other important functioning are impacted by major depressive disorder. For instance, the person may start missing work or school, or stop going to classes or their usual social activities. 

Watch ADAA's Video - What is Major Depressive Disorder (MDD)?

Who is at risk of depression?

Depression can affect anyone.
Females are more likely to suffer from depression than males, however, depression affects males as well.
Some of the risk factors that increase the likelihood of developing depression:
- Having a family member with depression
- Stressful life events (Katon 2011)
- Adverse childhood experiences (Katon 2011)
- Certain medical conditions may increase risk for depression, for example, stroke, multiple sclerosis, HIV (NIMH)

What is the difference between depression and sadness?

Experiencing sadness at one time or another is part of the normal human experience. However, depression is more than the occasional sadness that people experience. When an individual suffers from depression, the sadness tends to be more pervasive or long lasting, and can be accompanied by one or more of the following: decreased interest in pleasurable activities, feelings of hopelessness, low self-worth, excessive guilt, decrease in energy, concentration difficulties, appetite changes, sleep difficulty or excessive sleep, headaches, body aches and pains, and/or thoughts of suicide.

Can depression affect physical health?

Mental health and physical health are closely inter-related. Depression is linked with changes in the brain. Additionally, depression can increase risk for medical conditions such as cardiovascular disease, stroke, Type 2 diabetes, Alzheimer’s disease, osteoporosis, migraine (NIMH, Steffan et al 2020). Chronic depression can also increase the risk of obesity.
Studies show that major depressive disorder may increase risk for cardiovascular disease (Van der Kooy et al 2007). On the other hand, depression may emerge after a myocardial infarction, as a first episode or as a recurrent one (Spijkerman et al 2005).
Similarly, studies show that depression may increase risk for diabetes, and diabetes may increase the risk of depression (Mezuk et al 2008, Katon 2011, Oladeji and Gureje 2013). Depressive disorders in people who have diabetes, can lower adherence to diabetes treatment, and can increase the likelihood of ‘poor glycemic control’ and complications from diabetes (Oladeji and Gureje 2013).

Does depression affect older adults?

Depression can affect the elderly as well. In fact, it is often missed in the elderly. Elderly with chronic health conditions and related impairment in functioning may be at greater risk of depression. According to the CDC, depression among elderly ranges from 1% to 5% for those living in the community, and is about 11.5% & 13.5% among elderly who are hospitalized and elderly requiring home healthcare respectively (CDC).

Can depression affect children?

Just like adults, children can suffer from depression that significantly impairs their social, academic and/or other domains of functioning. For more information on childhood depression, check out this ADAA blog post

What is postpartum depression?

When feelings of sadness, low mood, or emptiness after the birth of the baby are significant and persists for 2 weeks or longer, postpartum depression may be present. It may be accompanied by hopelessness, crying spells, decrease in interest or motivation, worthlessness, decreased energy, thoughts of suicide, thoughts of hurting the baby, lack of interest in the baby, sleep difficulty or excessive sleep, appetite increase or decrease, and/or concentration, memory difficulties. Postpartum depression can affect both the mother and the baby in various, significant ways. Timely treatment is important.

Is depression treatable?

There are effective treatments for depression which have been established through research; many treatments for depression have been around for years. Many people ignore signs of depression despite experiencing significant suffering. This may be due to stigma, lack of awareness, and/or barriers to accessing care. Untreated depression confers several risks. Discussing your concerns about depression with your doctor can often be the first step in getting professional help.

What are the risks of untreated depression?

Untreated depression can increase risk of depressive episodes becoming more frequent, prolonged and/or severe over time and increase risk of suicide. It can significantly interfere with one’s functioning at work, school, and/or in interpersonal relationships. Depression can also increase risk of alcohol and/or substance use disorders. Additionally, depression can worsen the course and outcome of chronic medical conditions, such as diabetes, cardiovascular disease, and others.

What are some of the treatments for depression?

Psychotherapy and medications are the most commonly used treatments for depressive disorders. For mild to moderate depression, psychotherapy is generally thought to be the first line treatment. For moderate to severe depression, a combination of medications and psychotherapy is often considered. For severe depression that has not responded to multiple trials of medications and psychotherapy, electroconvulsive therapy (ECT), transcranial magnetic stimulation (TMS) may be some of the treatment options. The type of treatment hat is suitable for an individual depends on an individual’s specific clinical situation.

What are other psychiatric conditions that can co-exist with depression?

Anxiety disorders and substance use disorders are some of the psychiatric conditions commonly found co-existing with major depressive disorder (Hasin et al 2018, Steffan et al 2020). Other psychiatric conditions too, such as posttraumatic stress disorder (PTSD), eating disorders, can co-occur with depression.

#fafafa