The ongoing research and development in technologies such as AI and machine learning continue to prove their relevance in solving some critical issues in the world. For instance, introducing such technologies in wellness and the healthcare sector has led to revolutionary discoveries as well as making the delivery of medical treatment more seamless. In fact, Researchers based at the New York University recently reported that machine learning algorithms could currently identify lymphedema, which is a known side effect commonly associated with breast cancer treatment, with an impressive accuracy rate of 94%.
According to study author Mei R. FU, RN, PhD, and his colleagues, Lymphedema cannot be treated. The limited mobility, burning sensation, aches and swelling heaviness make the condition an unattractive candidate for treatment. As such, early diagnosis and intervention measures are the best and only ways that physicians can prevent the condition from deteriorating and minimize symptoms.
Fu, NYU Rory Meyers College of Nursing’s associate professor of nursing, revealed in the release that clinicians often diagnose or detect lymphedema based on their observation of swelling. Nonetheless, by the time the swelling can be measured or observed, lymphedema would have occurred for some time, which may translate to poor clinical results.
Fu and other co-authors also claimed in the release that lymphedema could take place after cancer surgery or even as late as two decades after the exercise. However, they added that only 41% of breast cancer patients experience the condition within ten years of treatment.
Fu and his team at the New York University focused their attention towards machine learning as the technology is popular for processing large volumes of data points that are autonomous from one another, as it is with lymphedema symptoms. The groundbreaking study involved 355 women who had previously gone through breast cancer treatment. The study involved the gathering of both clinical and demographic information prior to asking patients if they were going through any of the known 26 lymphedema symptoms.
After collecting the desired data from the patients, the researchers then embarked on an exercise to input the symptom information into five distinct machine learning algorithms. The algorithms comprised two Decision Tree models including a gradient improving model as well as a support vector machine and artificial neural network. In addition, the researchers said that all the five modalities detected lymphedema more accurately compared to the current standard statistical technique. However, they pointed out that the artificial neural network model proved to be the most successful one out of the two approaches, with a remarkable accuracy rate of 93.8%.
According to Fu, utilizing a well-trained classification algorithm in detecting lymphedema based on real-time symptom information is a promising tool with the potential to boost the outcomes of lymphedema. He also asserted that such detection accuracy is considerably higher than that attainable through the current and commonly-used clinical techniques. Additionally, Fu said that the approach also urges self- monitoring of symptoms, as it can send warnings to patients who are most likely to develop lymphedema.