Finding each other: Peer recognition as a clinical intervention in chronic illness

By Morgan Rose, chief science officer, Ema and Erlyn Macarayan, PhD, vice president, data science at PatientsLikeMe

May is Mental Health Awareness Month.

Most of the conversation this month treats depression and anxiety as standalone conditions, things people experience independently of their physical health.

For the patients we serve through the PLM platform, that framing leaves out something important.

Mental health in chronic illness functions as an overlay on a condition that does not end. It runs alongside the disease for as long as the disease lasts.

We have been analysing anonymised, aggregated conversational data from Ema, our agentic AI for patient support, and PLM users.

One pattern stands out across the dataset: what people are reaching for when they reach out.

The single most common behavior in the data, appearing in nearly 500 unique conversations, is the search for someone with the same diagnosis.

“Can you connect me with an MS group?” “Are there other people here with fibromyalgia?” “Does anyone in my city have what I have?” “I am looking for people like me.”

That request has a name in the research literature. It is peer support, and in the context of chronic illness, it functions as a mental health intervention.

The clinical case for peer recognition

The literature on peer support in chronic illness is well-developed.

Connection with others who share your diagnosis is associated with reduced depression severity, better treatment adherence, and a measurable drop in perceived isolation.

The mechanism behind those outcomes is recognition.

Someone else has lived inside the same symptom, navigated the same medication side effect, sat with the same diagnostic delay, and that recognition closes a specific gap that conventional therapy alone often cannot reach.

Therapists matter.

So do the people who know what an MS fatigue day actually feels like, what a fibro flare does to a person’s sense of self, what it is to be 34 and on a chemo regimen your friends cannot picture.

In clinical terms, that community is part of the care infrastructure for chronic illness, alongside medications, specialists, and labs.

PLM was built around this insight.

What Ema adds is a conversational layer that can route someone toward that community at the moment of need, before they have finished learning a platform.

“The PatientsLikeMe community has made living with MS manageable and in some bizarre way, even enjoyable sometimes because I’ve garnered these friendships and I am no longer afraid because all these other people are doing it with me.”

• PLM member living with MS

Why the burden is hard to address inside a clinical visit

There is a structural reason the mental health weight of chronic illness routinely goes undertreated. The visit is consumed by the physical condition.

The provider’s task list is long, the slot is short, and there is rarely a person in the room whose role is to ask how the patient is actually doing.

Some of that weight is also biological.

Depression in MS, for example, is roughly twice as common as in the general population, and is frequently undiagnosed because fatigue, cognitive change, and social withdrawal can be read as MS symptoms.

Similar overlap exists across cancer, autoimmune disease, and chronic pain.

Two systems run in parallel, shaped in part by the same underlying biology, yet routinely treated as separate.

That gap is where unguarded conversation tends to appear, and where the PLM data gets revealing.

The disclosure pattern

When mental health is mentioned in the PLM dataset, it rarely appears at the start of a conversation.

It surfaces sideways, after trust has been established by a clinical or logistical question.

One thread opens with questions about gabapentin and how PatientsLikeMe works.

A few exchanges later, the same user asks whether Ema has crisis resources. The conversation moves to feeling anxious, then depressed, then “I don’t know how I feel.”

Another thread spends several turns on MS management, medication questions, and which groups exist on PLM. Then the user writes, “My MS is making me feel overwhelmed and like everything is just too much. I’m not sure how to go on.”

Ema’s response in moments like that one matters.

She receives what was actually said, validates its weight, offers concrete steps for support, and connects the user back to the PLM community for the kind of isolation a clinical encounter cannot address.

The conversation pivoted because the user needed it to, and Ema followed.

That arc, the one that begins with a logistical question and ends in a disclosure about feeling unable to go on, is one of the clearest pictures we have of what the untreated chronic illness mental health burden looks like from the inside.

Crisis in the middle of an ordinary conversation

In 25 separate conversations in the PLM dataset, the mental health weight rose to the level of crisis. Users disclosed suicidal ideation directly.

One wrote, “I’m thinking of suicide.” Another asked what to do “if having a crisis and feeling suicidal.”

These conversations were happening on a patient platform, amid otherwise routine exchanges about a chronic condition.

The disclosures came in mid-thread, with no triage process to queue them.

Ema’s response was immediate and grounded. Hotline numbers, emergency services, an acknowledgment of the seriousness, and a reminder that the person is not alone.

The infrastructure to capture a moment like that at any hour, with no wait time, is something the conventional care system struggles to provide at scale. People are reaching for something in those moments.

Ema is built to be the thing they reach toward, and to hand them off to the human resources they need next.

What the data points toward

Pulling the patterns together, a coherent picture emerges.

People living with chronic illness carry a real and persistent mental health burden, and the burden tends to surface in the same conversations where they are managing medications, asking about treatment, and looking for others who share their diagnosis.

The most common request across the dataset is the request for community.

For PLM, that is the platform’s foundational thesis turning up in every dataset.

The platform was built on the premise that finding others with your condition is therapeutic. The conversation data confirms this, with patients explicitly asking for the connection.

For Ema, the implication is a design constraint.

We need to recognise when a question about gabapentin is an entry point into a question about feeling overwhelmed.

The route to peer recognition has to be as accessible as the route to clinical information.

And a moment of disclosure, whenever it arrives, has to land somewhere it can be received and responded to with care.

For Mental Health Awareness Month, the implication for chronic illness patients is direct.

Mental health in this population does not require a separate appointment that most patients will not make. It requires the people who already share the diagnosis to be part of the conversation, and it requires the platform to make that connection fast.

That is the work. It is what the data is asking us to build.

About Morgan Rose

Morgan Rose is chief science officer at Ema, an AI platform for patient health engagement.

Ema partners with health platforms and life sciences organisations to deliver clinically grounded, emotionally intelligent AI support where patients already are.

Learn more about Ema at emahealth.ai