Algorithm drastically cuts diagnostic time for Gaucher disease
A highly efficient machine learning algorithm capable of identifying patients with Gaucher disease (GD) from electronic health records has been developed, according to a study. This algorithm has the potential to significantly shorten diagnostic delays for individuals with GD, allowing for timely intervention and treatment before significant symptoms arise.
The study demonstrates that the machine learning algorithms are 10-20 times more efficient at identifying GD patients compared to the clinical diagnostic algorithm.
In addition, the combined use of 2 distinct algorithms, one age-based and one prevalence-based, provides a more comprehensive approach to capturing the heterogeneity of GD.
Results demonstrated that splenomegaly emerged as the most significant predictor for diagnosed GD using both algorithms. Additionally, geographical location, particularly in the northeast USA, played a pivotal role. Other key indicators included thrombocytopenia, osteonecrosis, bone density disorders, and bone pain.
Reference
Wilson A, Chiorean A, Aguiar M, et al. Development of a rare disease algorithm to identify persons at risk of Gaucher disease using electronic health records in the United States. Orphanet J Rare Dis. 2023;18(1):280. doi: 10.1186/s13023-023-02868-2. PMID: 37689674.