Researchers presented a new method of embedding electronic health records onto a biomedical knowledge graph to obtain high-dimensional health status profiles that can be used to identify people at risk of developing a chronic disease such as multiple sclerosis (MS), as published in the Journal of the American Medical Informatics Association.

Using their new method, the researchers were able to detect MS up to 5 years before the documented diagnosis of the disease in the clinic. The model has “broad applicability to other chronic medical conditions where early diagnosis can benefit patients,” they said.

The early identification of chronic diseases such as MS is of great importance because it can improve outcomes, reduce disease burden, and lower healthcare costs. Here, the team of researchers embedded millions of deidentified electronic health records into a biomedical knowledge graph, which resulted in knowledge-guided patient health signatures.


Continue Reading

Read more about MS prognosis

The researchers then used these health signatures as features in a random forest environment with the aim to identify patients who were at risk of developing a chronic condition.

“Our model predicted disease status of 5752 subjects 3 years before being diagnosed with [MS],” they wrote. This was better than predictions that used electronic health records only.

MS is a chronic disease with an estimated prevalence of 35.9 per 100,000 people worldwide. It is a complex disease affected by multiple factors including genetics and the environment. The disease can lead to symptoms such as fatigue and depression in its early stages.

However, these symptoms are nonspecific and not sufficient for patients to be referred to a neurologist at the prodromal stage of the disease. The early recognition of the disease could have major consequences in terms of disease trajectory for each patient.

Reference

Nelson CA, Bove R, Butte AJ, Baranzini SE. Embedding electronic health records onto a knowledge network recognizes prodromal features of multiple sclerosis and predicts diagnosis. J Am Med Inform Assoc. Published online December 16, 2021. doi:10.1093/jamia/ocab270