Lymphedema, a common complication of breast cancer treatment, can be accurately detected with a machine learning approach that uses real-time symptom reports, according to researchers.
Their study, “Machine learning for detection of lymphedema among breast cancer survivors,” appeared in the journal mHealth.
Lymphedema is a collection of fluid that causes swelling mostly in the arms or legs, and normally results from the removal of lymph nodes as a part of cancer treatment.
This common and chronic complication may happen right after surgery or months or even decades later. Research indicates that over 41 percent of breast cancer patients had lymphedema in the arms within 10 years of surgery.
Symptoms of lymphedema may include swelling in the arm, a feeling of heaviness or tightness, reduced range of motion, aching, recurrent infections, and skin fibrosis (or scarring). Although early intervention is able to ease symptoms, timely detection remains challenging.
“Clinicians often detect or diagnose lymphedema based on their observation of swelling. However, by the time swelling can be observed or measured, lymphedema has typically occurred for some time, which may lead to poor clinical outcomes,” Mei R Fu, PhD, the study’s lead author, said in a press release.
Computer science, such as machine learning, has emerged as a valuable tool to detect various medical conditions with greater accuracy over other approaches. Machine learning involves the creation of algorithms — a list of rules computers follow to solve problems — and, in the case of diseases, enables clinical decisions through the real-time reporting of symptoms.
The research team evaluated the accuracy, sensitivity, and specificity of machine learning using real-time symptom reports in the early detection of lymphedema in breast cancer survivors.
The scientists at New York University collected information from a total of 355 women in the U.S., covering demographic and clinical data, prior lymphedema diagnoses, as well as whether they were experiencing 26 different lymphedema symptoms. All participants had been treated for breast cancer, including surgery.
The study compared five different algorithms of machine learning. Results revealed that all five approaches outperformed a standard statistical method. The algorithm with the best performance to detect lymphedema, called artificial neural network, was 93.75 percent accurate in discriminating lymphedema from non-lymphedema cases based on reported symptoms.
“Such detection accuracy is significantly higher than that achievable by current and often used clinical methods,” Fu said.
This real-time lymphedema monitoring may encourage patients to track their lymphedema status without requiring a visit to a healthcare professional. Patients at risk could then be alerted to schedule an appointment for further evaluation.
“This has the potential to reduce healthcare costs and optimize the use of healthcare resources through early lymphedema detection and intervention, which could reduce the risk of lymphedema progressing to more severe stages,” Fu said.
“Use of a well-trained classification algorithm to detect lymphedema based on symptom features is a highly promising tool that may improve lymphedema outcomes,” the researchers wrote.