Sjögren’s disease is an under-diagnosed, long-term autoimmune disease that affects the body’s fluid-producing glands. It often takes a long time before these patients are referred to a specialist and receive a final diagnosis. As a result, they cannot be treated early, something that is of great importance for these patients and ensures a better quality of life. Machine learning, a method for finding patterns in large amounts of data using algorithms, contributes to early recognition of possible patients with Sjögren’s disease. This is a first step towards decision support software to support general practitioners in recognizing Sjögren’s patients.
Testing algorithms to identify patients
The algorithms use anonymized data from electronic patient records (EHRs) of general practitioners. The algorithms filter patients who may have Sjögren’s disease based on the available data, such as medical history, age, number of consultations, medication use and gender. We tested which algorithm was best able to recognize potential patients.
Train and validate algorithm for clinical application
This algorithm could eventually be applied in clinical practice. Before this happens, however, it must first be optimized and this can be done by training and validating it with other datasets. In addition, it is of great importance that GPs are involved in further optimizing the algorithm and developing possible decision support software so that it fits well with their wishes.
By: National Care Guide
Source: Nivel, https://www.nivel.nl/nl/nieuws/early-recognition-of-possible-patients-met-disease-van-sjogren-dankzij-machine-learning, consulted 15-08-2022