Credit: Unsplash / CC0 Public Domain
The diagnosis of chronic kidney disease, which often goes undetected until it causes irreversible damage, can soon be automated with a new algorithm that interprets data from electronic patient records.
Developed by researchers at Columbia University's Vagelos College of Physicians and Surgeons, the algorithm automatically searches a patient's electronic medical record for blood and urine test results, and can use a mix of established equations and machine learning to alert doctors to process the data Patients in the earliest stages of chronic kidney disease.
A study of the algorithm was published in April in the journal npj Digital Medicine.
"Detecting kidney disease early is of paramount importance as we have treatments that can slow the progression of the disease before the damage becomes irreversible," says study leader Krzysztof Kiryluk, associate professor of medicine at Columbia University's Vagelos College of Physicians and Surgeons . "Chronic kidney disease can cause several serious problems, including heart disease, anemia, or bone disease, and can lead to early death, but its early stages are often undetected and under-treated."
Chronic kidney disease progresses silently
Roughly one in eight American adults is thought to have chronic kidney disease, but only 10% of people with the early stages of the disease are aware of their condition. Of those who already have severely impaired kidney function, only 40% are aware of their diagnosis.
The reasons for an underdiagnosis are complex. People in the early stages of chronic kidney disease usually have no symptoms, and GPs can prioritize more immediate complaints from patients.
Two tests are also required, one to measure a kidney-filtered metabolite in the blood and one to measure protein loss in the urine, to determine if asymptomatic kidney disease is present.
"Interpretation of these tests is not always easy," says Kiryluk. "Many characteristics of the patient, including age, gender, body mass, or nutritional status, need to be considered and these are often underestimated by general practitioners."
The algorithm automates the diagnosis
The new algorithm overcomes these obstacles by automatically searching electronic patient records for test results, performing calculations that indicate kidney function and damage, orchestrating the patient's illness, and alerting doctors to the problems.
The algorithm works almost as well as experienced nephrologists. In tests with electronic health records in 451 patients, the algorithm correctly diagnosed kidney disease in 95% of kidney patients identified by two experienced nephrologists and correctly excluded kidney disease in 97% of healthy controls.
The algorithm can be used with various types of electronic health record systems, including those with millions of patients, and can be easily integrated into a clinical decision support system that assists physicians by suggesting appropriate stage-specific medications. The algorithm can be easily updated if the standards for diagnosing kidney disease are changed in the future and are freely available to other institutions.
A disadvantage of the algorithm is that it depends on the availability of relevant blood and urine tests in the medical record. The blood test is pretty routine, but the urine test is underutilized in clinical practice, says Kiryluk.
Despite these limitations, algorithmic diagnosis could raise awareness of kidney disease, Kiryluk says, and with earlier treatment, it could potentially reduce the number of people who lose kidney function.
Powerful tool for research
The algorithm has other important advantages for researchers. Because the algorithm can be applied to EHR records with millions of patients and identifies all patients with chronic kidney disease, not just those diagnosed with the disease, it improves the performance of many research studies.
The researchers have already applied the algorithm to a database of millions of Columbia patients to find previously unrecognized links between chronic kidney disease and other conditions. For example, depression, alcohol abuse, and other psychiatric illnesses were significantly more common in patients with mild kidney disease than in patients with normal kidney function, even after considering age and gender differences.
"Our analysis also confirmed that blood relatives of patients with kidney disease often have mild degrees of renal impairment," says Dr. Ning Shang, Associate Research Scientist at the Kiryluk Laboratory and lead author of the paper. "These results support a strong genetic determination of kidney disease, even in its mildest form."
In the future, Kiryluk said, the algorithm could be used to better understand the congenital risk of chronic kidney disease, as the algorithm enables genetic analysis of millions of people to discover new kidney genes.
The risk of chronic kidney disease is even more critical due to the COVID-19 pandemic
Ning Shang et al., Chronic Record-Based Chronic Kidney Disease Phenotype for Clinical Care and Observational and Genetic Studies Using Big Data, npj Digital Medicine (2021). DOI: 10.1038 / s41746-021-00428-1
Columbia University Irving Medical Center
The algorithm searches electronic health records to uncover hidden kidney diseases (2021, April 28).
accessed on April 28, 2021
This document is subject to copyright. Apart from fair treatment for the purpose of private study or research, no
Part may be reproduced without written permission. The content is provided for informational purposes only.