How this company wants to use AI to help therapists navigate clawbacks

As the CEO and co-founder of Eleos Health, Alon Joffe has been working to try to address a reimbursement problem dogging therapists across the US that has contributed to some practitioners in states like Illinois and Colorado forgoing insurance altogether.

Clawbacks, or retroactive coverage denials, are a pervasive problem in healthcare. It occurs when Medicare, Medicaid, or private health plans recoup payments that have already been made.

Many mental health providers already struggle with low reimbursement rates, and clawbacks can negatively affect their ability to keep offering services.

Eleos’s platform aims to help mental health providers better document services like talk therapy through artificial intelligence (AI), potentially reducing the likelihood that the insurer can find cause to “claw back” a payment. (The insurance trade organization AHIP did not return a request for comment to discuss concerns raised about clawbacks.)

Joffe, who spoke with Healthcare Brew from his home office in Tel Aviv, Israel, declined to disclose the cost of Eleos’s product and support services, but said it has more than 25 customers across 15 states.

Mental health providers are turning to AI for help with everything from diagnosing patients to reducing their administrative workloads. Other companies—like Seattle-based Lyssn, which employs AI to assess the use of evidence-based practices such as cognitive behavioral therapy—are also working to fill the need.

Joffe spoke about what his company is doing to help therapists tackle the sticky problem of clawbacks.

This interview has been edited and condensed for clarity.

How does Eleos’s platform work?

The technology we use can help the clinician produce and write the notes in a way that is compliant with insurer guidelines. AI can help you understand what happened in a conversation, produce the baseline for the note, and make sure that it’s compliant because it’s generated from their specific conversation.

How does the technology stay in the background—so it’s not in any way interfering with a normal session—but still pick up on all the important elements of a conversation between a therapist and client?

So let’s differentiate between telehealth and in person. If it’s telehealth, we’re on a Zoom call like we are right now, and it’s transcribing the audio in real time. Then it summarizes what happened in the session. When you’re doing that in person, you can use any standard device, whether it’s a smartphone or laptop, based on the guidelines and the procedures of that organization.

So your technology transcribes then summarizes the therapy session?

Yes. Summarization is a very, by the way, complex technological challenge. How do you take 45 minutes of unstructured conversation and summarize it with the elements that are actually important? How do you know what’s important and what’s not important?

That’s where the AI comes in?

Exactly. We take the transcription and then summarize it and rewrite it in a way as a clinician would write it. Then, within a minute or two of the session ending, that summary appears in the patient’s electronic health record.

Is that what’s submitted to insurance?

The clinician reviews the summary and can add additional feedback or insights from the conversation. Then they submit that for insurance purposes with the correct CPT code, which is used by insurers to process payments.

You have quality documentation, you have timely documentation, and you have the right coding.