AI-powered mental health screener Aiberry raises $8M

The company uses a conversational chatbot to diagnose and monitor symptoms of depression

Unlike physical illnesses, diagnosing and tracking mental illness can be difficult; while there are some companies that are using biomarkers to diagnose and treat mental illness, and other neurological disorders, that's a pretty nascent space. The standard way to find out if someone had depression or anxiety, and to determine if they've improved, is still through self-reported tests, such as the PHQ-9 and the GAD-7.

There are some big problems and limitations that come with these tests, though, specifically around engagement and accuracy, which is why Linda Chung and Johan Bjorklund founded Aiberry (pronounced “I” + “berry”), an AI-powered mental health screener that launched out of beta on Wednesday, along with an $8 million seed funding round.

"It's hard to fill out those digital forms: patients don't always know how to fill them out, even when they genuinely want to respond the right way and then, when they do respond, we don't really get the directionality of the symptoms. If I'm not sleeping well, does that mean I'm sleeping too much, or too little? We don't get that from the PHQ-9," explained Chung.

"Also, a lot of the people who are seen by a behavioral health provider have to typically do them every visit, so it's the same form every seven days, or every 14 days. And so, people start to either fill out all on the none at all side, or they'll fill it out the way that they filled it out last time, because you think that you haven't really changed. They’re also not reading it out loud, they’re not talking to anyone, so it's really easy to just mark responses in the form."

Aiberry gets around this by using a chatbot to more effectively engage, and measure, symptoms of mental illness. The assistant can conduct a conversation to detect mental health disorders by analyzing what is being said, the speech patterns being used, and even subtle changes in facial expressions.

Not only that, the chatbot actually talks to the patient, conducting a two-way conversation so it isn't just them filling out the forms all by themselves.

"She greets you and gets into the conversation. It's a little different each time, but that first question is, 'How have you been feeling lately?' It isn't, 'Do you feel well one day out of the 14 days? Half the time? Most of the time?' It really is an open ended response and you'll see that the more depressed somebody is the longer the response is; they really think about their answer. And so, we let them get all of that off their chest," Chung explained. 

"One thing that our customers and their patients love about our product is that not only do you get an objective measure of those symptoms, but also, what's the context around it? Did I not sleep for the last four nights, and I've been tossing and turning, because I haven't been feeling well? Do I fall asleep, and then wake up and have trouble going back to bed again? These are things that our tool can tell you, whereas the PHQ-9 or the GAD-7 can't necessarily."

Patients can access the chatbot in numerous ways, including with in clinic with their clinician, or they can access it in the waiting room and fill them out before they come in. Patients are also allowed to screen on their own from their own device.

Once the conversation is over, Aiberry is able to use data from the call to help diagnose mental health conditions and provides users with a risk score to show them if they have depression, as well as if there's a risk of suicide, along with insights about mood, energy, concentration, rate of speech, and cognitive bias, which gives clinicians extra information to better treat their patients. 

"We're taking input, not only what you say, but also how you say it through your voice, and also your facial expressions, through the video. So, we're not just looking at one modality, we have three different modalities, which somewhat helps with that," said Bjorklund.

"If somebody blatantly lied and said that they're not suicidal, we're not going to pick up on that, so we need people to tell the truth there. But we have had people, when we tested this out, that have been quite depressed, but they've been trying to mask that by being very positive, overly cheery, and AiBerry has picked up on that and has come back with a score that's much closer to what the reality was." 

In addition to being good for the patient, conducting these screeners via a chatbot is good for the clinician as well, as it allows them to save time: they get a transcript of the call, and they get access to a histogram, showing them how they patient are trending by comparing their previous risk scores. 

On top of that, it also saves them money they'd have to spend on staff to guide patients through the process.

"Wouldn’t we love to have more and more mental health providers out there; we have a shortage, they're getting burnt out, we're losing them, and they're not coming back to the field. And so, we have an issue of access to care. In an ideal situation, you'd have somebody pick up the phone and call you, greet you and talk to you before you got to see your psychiatrists, but with the lack of access, and the staffing issues, those things are really big barriers to people getting care," said Chung.

Currently, the company has only been used by a handful of customers, and it has been methodical about that, Bjorklund said, working with customers that have deep expertise in behavioral health to get feedback both around usability and around accuracy. That also means that, it doesn't have any specific ROI numbers from any of its clients.

"What we do find, what we hear from them, is that they're getting measurable outcomes now that they didn't have before. So, it's really helping them with their workflow, and it's helping them get insights that they might not have had before," he said. 

The new funding, which were led by Confluence Capital Group, with participation from Ascension AI, bringing total funding to $10M, will be used to accelerate adoption of the Aiberry platform, which means building out its sales team. It will also be used to build out more product features, including more languages, which will allow it to expand to more geographies.

"We're not quite at that point yet because, if you look at the overall software market for behavioral health, it's about $3.7 billion right now and it's scheduled to grow and is predicted to grow double that over the next five years, but roughly half of that is actually spent in the US. The US is such a large market. So we have a lot of work ahead of ourselves in this market. But the next couple of languages to add would be Spanish and French, for sure, which is really focused on North America then and satisfying this marketplace first," said Bjorklund.

The company also wants to expand who can use the product; currently, AiBerry can be used by anyone between by the ages of 13 and 79, and part of roadmap is to include the younger age group as well. 

Ultimately, though, the goal is not to replace the PHQ-9 and GAD-7 tests, but to expand access, giving everyone to the ability to screen themselves periodically and keep track of their own mental health and doing so in a fun and objective way so that they have a feel for what their baseline mental health looks like.

"We believe that by doing that, a lot more people that will get with early detection versus letting it go too far before they seek help. So, our ultimate goal, really, is to partner up with different companies that can help us in that quest. So, we're really addressing the front end with screening here, but then we're also partnering with providers and want to partner with a slew of different companies that can help us proliferate this out and make it available to as many people as possible," Bjorklund explained.