Subscribe Now

By entering these details you are signing up to receive our newsletter.

What is a flare in Sjögren’s? TREND Community’s innovative use of artificial intelligence to find answers for the community

SPONSORED CONTENT

During the first conference for Sjögren’s patients, in Athens in October 2023, TREND Community explained how its innovative social listening research model is helping to align patient and clinician perspectives of the disease

Written by Geoff Case, digital editor, RARE Revolution Magazine

Maria Picone
What is a flare in Sjögren’s?
Christopher DeFelice
Dr Chiara Baldini
What is a flare in Sjögren’s?
Coralie Peter Bouillot

Co-founder and
CEO of TREND Community

Co-founder and
CTO of TREND Community

Rheumatologist and assistant professor at the University of Pisa, Italy

General secretary of Sjögren Europe

TREND Community and its bi-directional research model

Maria Picone and Christopher DeFelice are well versed in the many challenges faced by people living with rare diseases. Their daughter was born with Prader-Willi syndrome (PWS), a rare genetic disorder that causes a wide range of physical symptoms, learning difficulties and behavioural challenges. When Maria and Christopher joined a Facebook community for caregivers of children living with PWS, they noticed discrepancies between how doctors and researchers described the disorder in the medical literature and how patients described it in their online conversations.

This discovery inspired their founding of TREND Community, a digital health analytics company. TREND uses artificial intelligence (AI) tools to analyse conversations taking place on social media within communities affected by rare and chronic disease communities, in a process called social listening. Its rigorous research, overseen by an institutional review board, discovers trends, identifies key concepts and delivers actionable insights

TREND’s analytic engine recognises relationships among different words and phrases in social media conversations, providing valuable insights into the diverse discussions happening within disease support groups as they share their experiences. TREND works with datasets in the public domain, such as social media conversations on Reddit, and—at a community’s request—with private datasets, such as the conversations within closed groups on Facebook.

TREND’s Sjögren’s workshop

In October 2023 at the first conference for Sjögren’s patients, organised by Sjögren Europe, TREND community hosted a workshop to explain how its bi-directional research model is helping to align patient and clinician perspectives of the disease.

The workshop was led by Maria, who is the company’s chief executive officer, and Christopher, its chief technology officer. Dr Chiara Baldini and Coralie Bouillot were on the workshop’s panel of experts to share their reactions to TREND’s findings. Chiara is an expert rheumatologist and an assistant professor at the University of Pisa. Coralie lives with Sjögren’s and is also the general secretary of Sjögren Europe (the European federation of national associations representing Sjögren’s patients).

Maria began the workshop by explaining how TREND turns data from social media conversations into insights through an iterative, bi-directional process: “To extract meaningful insights from the data we gather through conversations, we rely on community stakeholder input. This ongoing process of transforming data into real-world evidence and sharing it with the community fosters continuous learning and advancement.”

What is a flare in Sjögren’s?
This infinity loop describes TREND’s approach to research: insights from a community spark progress, the progress empowers the community, and the cycle repeats.

Maria went on to explain how TREND had attended EULAR in June 2023 to present its early data on Sjögren’s. (EULAR is the annual European congress of rheumatology). Hearing that “flare” was a trending concept within TREND’s dataset, Chiara had asked whether the data could be analysed more deeply.

Chiara explained why she had felt this deeper analysis was important: “In Sjögren’s some of the unmet needs probably arise from the discrepancy between the medical definitions of flares, symptoms and other words and the community’s definitions.” She added that “a definition of flare is important to empower patients with their disease management, so that they know when they need to visit their doctor. It’s going to be really helpful to the population to have that exact definition to know the difference between existing symptoms worsening and new symptoms occurring.”

“When we use patient-reported outcomes (PROs) as an endpoint in clinical trials, the drug may fail if we have not agreed on definitions.” – Chiara

Coralie agreed that in Sjögren’s disease there was a problematic communication gap that needed to be closed. She said: “As patients, we were not speaking the same language as clinicians.”1

Christopher explained how, following the conversation with Chiara at EULAR, TREND had analysed its data more deeply. Its research had focused on three questions: How much does the community talk about flares? What are flares to the community? What are the mental and emotional impacts of flares?

TREND’s engine analysed a dataset of more than 59,000 posts and comments within social media conversations on a subreddit for Sjögren’s.

1: How much does the community talk about flares?

When TREND’s engine analysed the dataset to see how much the community talks about flares, the data revealed instances of “flares” co-occurring with symptoms, but it also indicated that they were mentioned in isolation. Christopher described another notable finding: “When the engine’s clinical entity recognition (CER) model identified ‘flare’ and labelled it as a SYMPTOM, it faced challenges in establishing connections with the terminology within the engine’s internal Lexical Knowledge Base.” He explained that this difficulty arose because the concept of a “flare” is typically not categorised as a symptom, and references to it in other medical terminologies, which form the core of the engine’s Lexical Knowledge Base, exhibit inconsistencies.

What is a flare in Sjögren’s?
TREND’s Clinical Entity Recognition model identifies clinical entities in input statements and links them with concepts within the engine’s Lexical Knowledge Base, which is based upon established medical terminologies.

Coralie commented on how that finding aligns with how she understands patients’ lived experience with Sjögren’s. “As patients, if we are feeling bad, we won’t say ‘Today I’ve got a headache, joint pain…’ We say, ‘I think I’m in a flare.’ ”

2: What are flares to the community?

Christopher explained how TREND’s engine had analysed the data to see what other concepts authors were discussing when they talked about flares, measuring the strength of association between these concepts. Discussion of flares was associated with discussion of symptoms and of specific body parts, but it was not associated with discussion of tests.

Chiara commented on this apparent disconnect. She said that if worsening in patients’ symptoms is not being properly captured through blood and urine tests, our understanding of the effectiveness of available drugs—and novel drugs—may be incomplete.

What is a flare in Sjögren’s?
The bar chart shows how symptoms and body parts are talked about significantly more in posts and comments about flares; tests are talked about less.

Christopher described how surprised he had been to see “fatigue” so strongly associated with flares. Based on his reading of the medical literature, he would have expected dry eyes and dry mouth (xerostomia) to have greater prominence than fatigue, arthralgia (joint pain) and stress, which all ranked more highly.

What is a flare in Sjögren’s?
The symptoms associated with flares within TREND’s dataset on Sjögren’s. “PMI” means pointwise mutual information, a statistical measure of the strength of association between concepts within statements.

3: What are the mental and emotional impacts of flares?

Christopher explained how TREND’s engine analysed statements about flares to see whether they were associated with concepts related to mental health: feeling down, stress, anxiety or depression—the engine can detect the emotional underpinnings of a statement with high confidence. He said that TREND’s analysis had found that mental health was more likely to be discussed in conversations about flares than in conversations about other topics (a difference of approximately 10%).

The pie chart shows that mental health was more likely to be discussed in conversations about flares.

Reddit is a public data source so Christopher was able to share some of the statements about flares that are also related to mental health. The first of these referred to “flipping a switch”. Christopher said that one of TREND’s AI tools proposed that phrase as a general label for similar comments, which emphasises the suddenness with which a flare may begin.

Christopher highlighted how TREND’s analysis pointed to the frustration patients felt during flares, arising from the misalignment between their experience and their test results, and this was apparent from the comments he shared.

“It’s like someone flips a switch. The switch flips on and the flare starts. Then the switch flips again and it’s gone.”

“The waves of symptoms you’re experiencing are called flares. From experience, you can be flaring your ass off and your bloods will look ‘fine’. It’s frustrating. Bloodwork never tells the whole story.”

“I’m in one of my flares. I’ve had multiple I had ANA [antibody] tests in the past, which will come back normal. Is it better to be tested during a flare? Or doesn’t it matter? I don’t know what is wrong with me. All of my bones feel swollen. My entire body seems inflamed.”

Commenting on the data, Coralie said she considered these statements “unsurprising”:

“This is the story of Sjögren’s patients’ lives. We visit our doctors, the test results are normal, but we feel awful.”

The value of TREND’s research into Sjögren’s

In her concluding remarks at the end of the workshop, Chiara emphasised the importance of understanding flares better in rheumatological diseases, in which symptoms may “grumble” permanently or come and go (relapse and remit).

Chiara described the TREND model as an “impressive” way of approaching research that can supplement ongoing observational studies. It gives researchers an “immediate snapshot” into patients’ lived experience, she says, and enables deeper analysis to explore, for example, the differences in how patients and clinicians are using language, or the relationship between experiencing symptoms and having been prescribed specific drugs.

Coralie, too, is hopeful that TREND’s ongoing research will help to deliver further insights into patient’s lived experiences, to help align the perspectives of patients and clinicians.

“We need clear definitions. When we have them, we can work on tools to evaluate flares for clinical trials and for the daily management of patients… If we have clear definitions, patients and clinicians can speak the same language.” — Coralie

A manuscript reporting TREND’s insights into Sjögren’s will be released at a later date. If you would like to learn more about this research, or if you have research questions that you would like to answer within your own disease area, Maria and Christopher would love to hear from you: interested@trend.community

References

[1] C. Bouillot et al. AB1732-PARE “Flare, did you say flare?” Flares in Sjögren’s disease: the patient perspective. Ann Rheum Dis 82, 2103 (2023).

Further reading

Find out more about TREND Community at: trend.community



Skip to content