A new study published in JAMIA Open looked at whether natural language processing (NLP) could help identify symptoms of systemic mastocytosis (SM) in electronic health records.
NLPs are a type of artificial intelligence that analyzes written text. Because many symptoms of SM are documented in free-text clinical notes rather than structured fields, they can be difficult to track, compare or use for research. The study authors aimed to see whether this new technology could reliably extract this information.
Researchers built an NLP tool trained to detect 23 symptoms commonly seen in SM across several categories, including skin reactions, gastrointestinal problems, neurological symptoms, musculoskeletal pain and systemic reactions. They tested the tool on clinical records from 135 people diagnosed with SM and compared the results to two other patient groups: those with chronic spontaneous urticaria (chronic hives and itching) and those with neither condition.
Learn more about SM signs and symptoms
During testing, the researchers noted that the tool’s performance was strong. Most symptoms were correctly identified more than 90% of the time. A few symptoms, like abdominal bloating and swelling, were harder to interpret from clinical notes, but overall accuracy remained high.
When applied to more than 118,000 clinical notes, the NLP system found clear patterns. Notes for people with SM included more gastrointestinal and systemic symptoms compared to the other groups, while skin symptoms appeared more often in patients with urticaria. This mirrors what many people with SM already experience in real life: Symptoms show up across many systems, and patterns aren’t always obvious without careful review.
The study suggests that automated tools can help make sense of real-world symptom patterns that otherwise stay hidden in medical records. For a disease like SM that often involves diagnostic delays and a wide range of symptoms, this type of technology may one day support earlier recognition, better tracking of symptom burden and more informed care.
The researchers note that while this is still an early step, the approach has potential. “These findings support the broader utility of NLP for characterizing rare disease symptoms, informing early recognition, and enhancing data-driven care strategies,” they concluded.
Sign up here to get the latest news, perspectives, and information about SM sent directly to your inbox. Registration is free and only takes a minute.
