New Machine Learning Algorithm for Breast Cancer Detection
Artificial intelligence (AI) is very quickly expanding in the healthcare industry. AI’s primary role has been to automate processes to make it easier for employees to work on more complex people-related concerns. Yet as the technology develops, Healthcare Central reports that AI is also being used as a diagnostic tool. With a faster, more accurate way of detecting illnesses, and administering treatments, the mortality rate could be significantly reduced. This includes the treatment of conditions like cancer.
A Dartmouth research team is harnessing machine learning to predict the chances of breast cancer. The assistant professor of biomedical data science and epidemiology in Geisel School of Medicine, Saeed Hassanpour, along with his team, are currently developing the technology. The AI looks at the possibility that a breast lesion found during a medical examination is or can be cancerous. This comes after breast cancer screenings inducing false negatives have put women in danger of overdiagnosis and overtreatment.
According to Hassanpour, doctors currently perform a core needle biopsy whenever a marker for high-risk breast cancer is found—and it doesn’t matter whether the lesion is malignant or benign. “Seventy to 80 percent of women didn’t need this surgery. Only 20 to 30 percent of patients are found to have cancerous lesions,” he explains. This led to a search for a more personalized and less invasive alternative using machine learning.
Geisel School of Medicine’s associate professor of radiology and gynecology and obstetrics Roberta DiFlorio Alexander, used AI in radiology research. She found that the use of digital imagining allows researchers to mine vast amounts of data. After the data is combined with molecular and proteomic data it can help “create personalized medicine and a better understanding of a patient’s disease”. She believes that this will eventually allow for a “personalized treatment for each patient’s unique cancer.”
With the breakthroughs in algorithm techniques that yield a surplus of data, medical professionals can diagnose illnesses much faster. A number of computer-assisted diagnoses through deep learning algorithms are already in place, with some being applied to the three common diagnoses in dermatopathology, according to an article in the Journal of Pathology Informatics. The article found that there was a high degree of accuracy “with 113 out of 114 cases being correctly diagnosed by the algorithm with sensitivity of 98.8% and specificity of 100%.”
With machine learning, patients can also be diagnosed not only faster, but earlier, too. This will have a huge impact on healthcare, especially since the number of nurses and doctors are consistently decreasing and time makes a huge difference in a patient’s chances of recovery. A post by Maryville University details how there is expected to be a shortage of 100,000 primary care physicians by 2025. This means that more people with conditions like breast cancer may not be diagnosed early or correctly. Machine learning in healthcare will help make up for this huge shortfall.
All this is possible through machine learning algorithms that we’ve discussed on Stratio. AI is being used to solve real world problems, and in the healthcare industry this has the potential to make up for the lack of doctors and nurses, and most importantly save lives.
Flora Clover spends most of her time writing tech articles. She is particularly interested in the increasing application of AI and machine learning across industries. She enjoys solving crossword puzzles in between coding and checking the latest trends online.