With the pandemic closing down the world and veiling the world with a fear of the virus, a good number of women were pushed to skip the routine check-ups. This will pose a major roadblock in the early detection and diagnosis of signs and symptoms which are necessary to provide the right treatment and more importantly, at the right time. And this where physicians decided to tap into an AI algorithm that could solve this problem.
The AI Approach
The challenge posed by the pandemic is what led Dr.Constance Lehman to modify the medical approach by fusing in the efficiency of artificial intelligence. Lehman has experience of working with MIT researchers to develop a way to apply AI to cancer screening. The AI algorithm used is capable of predicting who is at the most risk of developing the disease.
According to Lehman, about 20,000 women have missed their routine checkups due to the pandemic. And according to the estimates, it implies about 100 cancer cases that haven’t been diagnosed.
This is where the AI algorithm comes in. It has been of substantial help in facilitating the identification of women who have early signs of cancer. In addition, women who were flagged by the AI algorithm were three time more likely to develop the disease. This indicates the accuracy and precision with which the algorithm works in comparison to the statistical techniques used before.
How Does The Algorithm Work?
The algorithm works by analysing the previous mammograms and detecting warning signs that even the physicians missed in the initial scans. The algorithm’s eye for the subtleties could prove to be the difference between life and death. The potential of artificial intelligence to enhance medical imaging has long been discussed and some tools are already in the medical scene.
The AI is on point and accurate when it comes to predicting risk. The screening involves collecting details and information concerning the patient, in addition to mammogram examination. These are then fed into a statistical model which will determine the follow-up screening requirements.
The algorithm called Mirai was developed by Adam Yala, an MIT PhD student, even before covid with the objective of facilitating early detection.
In order to overcome the challenges that stalled the use of AI in radiology, Yala used an adversarial machine learning approach. The algorithm works by deceiving the other in order to determine the differences in scores among radiology machines. This indicates that patients, who are in the same risk levels can show different scores. The model is also more accurate, owing to its ability to incorporated data from several different years.
The standard four mammogram views are analysed by the algorithm, following which it infers patient information such as surgery history or hormone-related factors like menopause.
In comparison to the statistical models, Mirai proved to be more accurate. By comparing the historical data concerning patients, it was shown that about 42% who developed cancer in five years were put in the high risk category by the algorithm. The contrast in accuracy levels become even more evident with the fact that the best existing model only identified 23% in comparison.
It is widely accepted that this algorithm could be of immense help in cancer detection and diagnosis especially in a situation where the health care domain has undergone a lot of changes due to the pandemic. According to Charles Kahn( Professor of Radiology, University of Pennsylvania),with further improvements, it might be possible to develop a more personalized treatment and to provide the patients with a customized screening plan.
Dr. Lehman hopes that AI methods will prove to be beneficial to people from all walks of life, providing equal and quality healthcare at the right time.