Integrated EHR Algorithm May Flag Chronic Kidney Disease

An automated and integrated EHR algorithm could diagnose chronic kidney disease before it could cause irreversible damage, according to research published in NPJ Digital Medicine.

Columbia University Vagelos College of Physicians and Surgeons researchers leveraged machine learning techniques to develop an automated EHR algorithm that dissects a patient’s EHR for blood and urine test results, which would alert the physician during the early stages of chronic kidney disease.

Roughly one in every eight US adults has chronic kidney disease, but only 10 percent are aware of having the condition during the early stages, according to the study authors. Furthermore, only 40 percent are currently mindful of having a severely reduced kidney function.

“Chronic kidney disease can cause multiple serious problems, including heart disease, anemia, or bone disease, and can lead to an early death, but its early stages are frequently under-recognized and undertreated,” said Krzysztof Kiryluk, MD, lead study author and associate professor of medicine at Columbia University Vagelos College of Physicians and Surgeons.

Health organizations currently face two main obstacles that lead to undertreatment and late diagnosis.

First, many individuals in the early stage of kidney disease are asymptomatic and it is typical for providers to prioritize symptomatic patients or patients with more urgent complaints. Next, health systems need to have a test that measures a kidney-filtered metabolite in the blood and another test that measures protein leakage in urine to detect early-stage, asymptomatic kidney disease.

“The interpretation of these tests is not always straightforward,” Kiryluk explained. “Many patient characteristics, including age, sex, body mass, or nutritional status, need to be considered, and this is frequently under-appreciated by primary care physicians.”

Researchers developed the algorithm to defeat those two obstacles. It automatically scans the EHR for test results, calculates the kidney damage, categorizes the disease, and alerts the physician of the diagnosis.

According to the study authors, the algorithm is on par with experienced nephrologists.

Researchers tested 451 diagnosed patients and correctly diagnosed kidney disease in 95 percent of the identified patients that two experienced nephrologists identified. It also ruled out kidney disease in 97 percent of healthy patients.

Researchers could integrate the algorithm into several different EHR systems. It could also be integrated into a clinical decision support system to help suggest appropriate stage-specific medications.

Health IT specialists or researchers could also optimize the algorithm if kidney disease diagnosis standards changed in the future. The algorithm is also available for other health systems and institutions, the study authors noted.

Columbia researchers have also integrated the algorithm into the Columbia EHR system to uncover unrecognized links between chronic kidney disease and other conditions.

Research revealed alcohol abuse, depression, and other psychiatric conditions were more common among patients with kidney disease than those without kidney disease.

“Our analysis also confirmed that a mild degree of kidney dysfunction is often present in blood relatives of patients with kidney disease,” says Ning Shang, PhD, associate Columbia University research scientist. “These findings support strong genetic determination of kidney disease, even in its mildest form.”

Looking forward, researchers could optimize the algorithm to enhance understanding of chronic kidney disease risk because it uncovers genetic analysis about new kidney genes, Kiryluk said. Researchers can also apply it to EHR datasets with an unlimited number of patients to identify individuals with chronic kidney disease, not only those diagnosed with the disease.

Furthermore, the algorithm could also improve chronic kidney disease research studies.

“Identifying kidney disease early is of paramount importance because we have treatments that can slow disease progression before the damage becomes irreversible,” concluded Kiryluk.