AI Models Accelerate Rare Disease Diagnosis at Boston Children's and CHEO
核心洞察
AI initiatives at Boston Children's Hospital and Children's Hospital of Eastern Ontario are helping to diagnose rare diseases by analyzing clinical and genomic data.
CHEO's ThinkRare (搜索) algorithm scans electronic health records and has led to 21 new rare disease (搜索) diagnoses with a 70% success rate so far.
Boston Children's Hospital's reasoning model reviews unsolved cases and suggests disease candidates, significantly speeding up the diagnostic process.
Artificial intelligence is beginning to shorten the notoriously long "diagnostic odyssey" faced by patients with rare diseases, according to a new report from JMIR Publications. Science journalist and JMIR correspondent Simon Spichak reports on AI initiatives at Boston Children's Hospital (BCH) and Children's Hospital of Eastern Ontario (CHEO) that have begun to help diagnose rare diseases, as detailed in the article "How AI Is Speeding Up the Diagnostic Odyssey for Rare Diseases."
The two institutions have deployed complementary AI approaches to tackle a persistent clinical challenge: rare diseases are difficult to identify, and patients often wait years for an accurate diagnosis.
CHEO's ThinkRare Algorithm
At Children's Hospital of Eastern Ontario, the ThinkRare (搜索) algorithm scans electronic health records and flags possible rare diseases. The tool has already prompted genetic sequencing and follow-up in a handful of patients, leading to 21 new rare disease (搜索) diagnoses and a 70% success rate so far.
Boston Children's Hospital's Reasoning Model
Boston Children's Hospital has developed a reasoning model that works by reviewing clinical and genomic data in individual unsolved cases and suggesting possible disease candidates. This approach has been credited with speeding up the investigative process significantly.
Limitations and Future Directions
While the early results are promising, Spichak notes the risk of clinicians failing to catch errors produced by these models. Despite this caution, further research and model validation, as well as future initiatives to scale their use, have the potential to transform diagnosis and care for patients with rare diseases.