How algorithmic bias created a mental health crisis [PODCAST]

Listen and followApple PodcastsSpotifyYouTubeMore apps · Browse all episodes

Health care executive Ronke Lawal discusses her article, “Digital mental health’s $20 billion blind spot.” Ronke explains how the booming digital mental health industry is systematically failing 40 percent of the U.S. population (racial and ethnic minorities), ignoring a $20 billion market opportunity. She argues that the “one-size-fits-all” model, based on Western-centric care, is a product failure that leads to algorithmic bias, misdiagnosis in diverse patients, and culturally incompetent artificial intelligence (AI) chatbots. This failure in digital mental health doesn’t just alienate users; it creates real financial consequences for health systems, including higher emergency department use and readmission rates. Ronke makes the ironclad business case for embedding “cultural intelligence” into technology, arguing it’s the only way to fix systemic bias and build effective digital mental health tools.

Our presenting sponsor is Microsoft Dragon Copilot.

Want to streamline your clinical documentation and take advantage of customizations that put you in control? What about the ability to surface information right at the point of care or automate tasks with just a click? Now, you can.

Microsoft Dragon Copilot, your AI assistant for clinical workflow, is transforming how clinicians work. Offering an extensible AI workspace and a single, integrated platform, Dragon Copilot can help you unlock new levels of efficiency. Plus, it’s backed by a proven track record and decades of clinical expertise, and it’s part of Microsoft Cloud for Healthcare, built on a foundation of trust.

Ease your administrative burdens and stay focused on what matters most with Dragon Copilot, your AI assistant for clinical workflow.

VISIT SPONSOR → https://aka.ms/kevinmd

SUBSCRIBE TO THE PODCAST → https://www.kevinmd.com/podcast

RECOMMENDED BY KEVINMD → https://www.kevinmd.com/recommended

Transcript

Kevin Pho: Hi, and welcome to the show. Subscribe at KevinMD.com/podcast. Today we welcome Ronke Lawal. She’s a health care executive. Today’s KevinMD article is “Digital mental health’s $20 billion blind spot.” Ronke, welcome to the show.

Ronke Lawal: Thank you, Dr. Pho.

Kevin Pho: All right, let’s start by briefly sharing your story, then we’ll talk about your KevinMD article.

Ronke Lawal: OK, so hi everyone and our audience. My name is Ronke Lawal and I’m the founder of Wolve. So Wolve is currently being built within the intersection of neuroscience and artificial intelligence, and it became something I had to pursue due to my concern for the global mental health crisis.

So currently about 1 billion people globally are living with a mental health disorder, and two-thirds of them do not have access to mental health care. It’s either too expensive, there aren’t enough professionals, or stigma simply holds them back. In certain regions, especially developing countries, people rely heavily on traditional and religious healing practices, so they never consider seeking mental health care. So Wolve is meant to provide adequate, accessible, and affordable mental health care for them globally.

Kevin Pho: All right. So your KevinMD article is “Digital mental health’s $20 billion blind spot.” For those that didn’t get a chance to read your article, tell us what this one is about.

Ronke Lawal: Yes. So my article is about a $20 billion market opportunity that’s currently being ignored by players in the United States digital mental health market. So currently the market is worth $8 billion, which seems quite large, but it could be bigger if they considered 40 percent of the United States population that’s currently being ignored.

And these are people within certain communities, such as African-American, Asian-American, and also the Hispanic community. So these people actually face a lot of cultural barriers when it comes to seeking mental health care. And my article addresses that closing this gap by providing inclusive health care isn’t just a moral goal, but it’s also the single biggest opportunity for the industry.

Kevin Pho: All right. You mentioned some cultural barriers to mental health care in these demographics. So talk more about that. Give us some examples of what that may look like.

Ronke Lawal: OK. So generally the cultural barriers occur traditionally and also in the digital mental health space. So traditionally, let me talk about that. This is what that looks like.

In the traditional space, currently African-Americans who are suffering from affective disorders such as depression, bipolar disorder, dysthymia are often being misdiagnosed as having schizophrenia. While African-Americans and also Hispanics are less likely to receive guideline recommended care within the health care system in comparison to their white counterparts. And this is really unfair, and usually stems from lack of representation quite often, and also sometimes just lack of care.

And for the digital mental health care space, the problem is usually more of a product failure, whereby the product that’s currently in the market isn’t compatible for users within these communities.

Kevin Pho: So you’re saying that a lot of the training models that these digital mental health models are used, they don’t necessarily include a lot of these other cultural demographics.

Ronke Lawal: Yes. And there’s a lot of reason behind that. So for these models, the data they actually collect to build models are from traditional mental health care. And what does that mean? Now we already addressed how there’s an underrepresentation when it comes to care.

And yet, so within this we also have situations whereby even Black children and even Hispanic children who are suffering from depression are often being misdiagnosed as having conduct disorder. So with all this data that currently exists accumulated and now used to train models, and something that artificial intelligence is really good at is magnifying things, it expands it. So we are scaling the same problem that actually exists within traditional mental health care.

Kevin Pho: And in your article you talk about this chatbot example, Tessa, right? So tell us about what happened in that particular instance.

Ronke Lawal: So for Tessa, it was them creating an application that was meant to help people that are suffering from eating disorder, but it was very uncompatible, and it became something that was affecting patients negatively, users, and it had to be taken down.

And this is a problem that also occurs even with usual mental health care, whereby a physician could refer a patient to use an application, and the patient could actually log in, and let’s assume this patient is a Spanish speaking patient, and then they log into the application and they realize that there is something missing. And this is the fact that the application might be offering them an intervention that is more individualistic, meanwhile this patient comes from a background whereby culture and family is very valid, a collectivist culture.

So for the patient this is a product failure, meanwhile on the side of physicians who might be unaware, it seems like non-compliance to them, because they could see this as a patient that is unwilling to change. So it’s a kind of situation that affects everybody in different ways.

Kevin Pho: Now what are the reasons why a lot of these AI models, they don’t include a lot of culturally diverse demographics? Is it simply due to cost? Is it difficult to incorporate some of these cultural nuances?

Ronke Lawal: So I would say a lot of it stems from underrepresentation. There are about 37,000 mental health professionals and only 10 percent are people of color. And we have about 300 million people within the United States. Now 40 percent of these people are represented by just 10 percent. And it’s even more nuanced than that, because you have to consider we have biracial individuals, also people within interracial marriages and all of that. So they don’t even have enough people to cater to them.

So when artificial intelligence models are being trained, everybody needs data. And the data you’re going to get would most likely be something that isn’t really beneficial for a certain demographic. It’s a homogeneous user based data. It’s a data that only benefits a certain group. So when you use it, you’re simply amplifying biases, and it really wouldn’t be able to build something very intelligent.

In addition to that, another problem is that when people are building these applications, quite often they really do not see diversity or cultural competency as something that should be built within the system.

They never consider it as something that should be part of the DNA, or a very critical logical part of the system, right? They simply see it as something they need to add. And for diversity and inclusion, you can’t retrofit those things. You can’t create an app with a particular kind of data and expect several people across the world to use it, or for it to actually benefit several people across the world. That’s something that would hardly ever work.

Yes. So that’s part of the problems that people actually have, and I think a lot of the time they just never consider what the issue could actually be, or what bottlenecks they could experience along the way.

Kevin Pho: So you’re saying it’s not a problem with the apps themselves, but it’s just the fact that we don’t have a lot of cultural representation when it comes to mental health care, and a lot of that data that apps draw from comes from that, right?

Ronke Lawal: Yes, you’re right.

Kevin Pho: So where does that leave patients with culturally diverse demographics? Tell us about some of the effects that it has on them. Do they simply just not receive the culturally appropriate mental health care?

Ronke Lawal: Yeah. So mental health care is really critical, and untreated mental health conditions would lead to severe consequences. You know, studies continue to show how untreated mental health issues often increase the risk of disabilities, heart diseases, and also nervous system dysregulation and other forms of chronic illnesses.

So this would definitely place a financial burden on individuals and society as a whole. And then it increases poverty levels and reliance on government support. And we need to realize that once there’s an increase in poverty levels, it means that there’s going to be a health care system problem.

So the people within this demographic we are speaking about right now, who don’t have access to good mental health care, there are people who don’t get help until they’re in crisis, which would lead to higher emergency department utilization. That means they would always show up at the emergency room, and it’s always when they dial 988, 911, or when a primary caregiver tells them show up at the emergency room right now. That’s what happened. And this costs a lot of money and increases readmission rates, and hospitals who operate under value based care models would lose money on readmissions every time. It’s a vicious cycle.

Kevin Pho: In your article, you call for cultural intelligence to be embedded into the technology. What exactly do you mean by that, and what would that look like?

Ronke Lawal: Yeah. So for cultural intelligence, it’s an adaptive therapeutic framework, right? So whereby currently we have a one-size-fits-all application, that shouldn’t be what should be the reality.

Many patients usually narrate that they’re having some affective disorder or some issues at home and things like that. They should be able to relate to people across various diverse backgrounds. You know, person A might be someone you could actually tell, oh, you can meditate or do some grounding exercises whereby you count certain numbers. While person B, you might actually need to ask them about their family. Ask them how they interact at home. How do you solve problems at home? What would you like to see? Or would you like for a family member to join you in this interaction that we are having, that kind of thing. So it makes them feel like, OK, you understand their cultural framework, and it makes them feel considered and respected. So that’s part of it.

For us at Wolve, one of the things that we have considered is to actually create an ontology system that maps culturally specific expressions to therapeutic concepts and adjusts interpretations based on cultural markers. Like it adjusts your interpretation no matter what a person says. You know how sometimes when someone says it’s fine, it varies across different backgrounds. The interpretation varies.

So adjusting interpretation based on cultural markers is very key. And we also try to incorporate cultural healing frameworks with evidence-based therapy. So all of this shows that cultural competency is something that, with the importance, it’s something that can actually help everybody and not just a particular group. It benefits everybody across the globe, across the country.

Kevin Pho: We’re talking to Ronke Lawal, who is a health care executive. Today’s KevinMD article is “Digital mental health’s $20 billion blind spot.” Ronke, let’s have some take-home messages that you want to leave with the KevinMD audience.

Ronke Lawal: Yeah. So my take-home message would be that cultural competence is not a feature. It is the foundation. And we must stop asking how to add diversity and start asking if a tool was designed for diversity from day one. You know, the consideration must be intentional and not arbitrary.

And for clinicians, I think that it is very appropriate for them to hold their vendors accountable and ask them questions. You can demand to see results broken down by racial groups, by ethnic groups.

Kevin Pho: Ronke, thank you so much for sharing your perspective and insight. Thanks again for coming on the show.

Ronke Lawal: Thank you so much, Dr. Pho. So good to meet you.

Listen and followApple PodcastsSpotifyYouTubeMore apps · Browse all episodes