AI is widely used to cross computational hurdles and give the software the ability to learn. AI’s true limits is a question that still awaits an answer. Sure, it can drive a car, but can AI detect cancer before it even happens?
This is what a group of researchers in IBM have been trying to answer.
IBM AI Can Predict Brest Cancer
As per an IBM report, “A team of IBM researchers published research in Radiology around a new AI model that can predict the development of malignant breast cancer in patients within the year, at rates comparable to human radiologists.”
The model studied imaging data and a comprehensive patient’s health history to predict if the patient is likely to develop breast cancer soon. As per the results, the algorithm correctly predicted the development of cancer in 87% of patients. Further, it was also able to correctly interpret 77% of non-cancerous cases.
This can help the radiologists in confirming or denying the possibility of Brest Cancer. For patients, it can work as a second opinion.
The same AI model was put to test against 71 different cases that radiologists had originally determined as “non-malignant”. These cases were the ones who ultimately ended up being diagnosed with breast cancer within the year. Out of 71, the AI system was able to correctly identify breast cancer in 48 percent of individuals.
Google AI For Detection Of Lung Cancer Might Be Better Than Radiologists
Google used Deep Learning – a kind of AI, to detect malignant lung nodules on low-dose chest computed tomography (LDCT) scans. In doing so, it outperformed expert Radiologists.
Not only detection but the model was able to predict lung cancer for patients who had biopsy confirmed cancer within a year, with more accuracy than human experts. Also, it produced fewer false negatives and fewer false positives. This would lead to fewer unnecessary follow-up procedures and even less missed tumors that go latent in the medical tests.
Use of AI in Medical
From soon after its genesis, people have been trying to use AI to detect, and even to predict diseases. Hence, it doesn’t come as a surprise that there were already many prediction models that used AI to detect cancer. Although, they mostly rely on either mammogram images or on medical records of the patient.
IBM AI reached a success rate of 87% and proved to be more reliable than most others because it uses both. AI learns with anonymized mammography images which are linked to biomarkers and clinical data. This allows the creation of an algorithm with comparatively higher accuracy.
More than 42,000 chest screening images taken from nearly 15,000 patients trained Google’s model. 578 of these patients developed cancer in the year that followed. A Deep Learning Model (one which learns by example) then scanned the results which helped AI to detect cancer.
If we go by data, 1 in every 8 women in the US itself is likely to develop invasive Brest Cancer over the course of her lifetime – and in 85% of cases, it occurs in women who have no family history of breast cancer.
As per WHO data, Lung Cancer tops the table as it takes more number of lives annually than any other form of Cancer – Over 2 million lives per year. A lot of people suffer an enormous amount of pain due to Lung Cancer, and this could be mitigated if the cancer is detected at an early stage.
For a global population of over 7.5 billion, that’s a lot of people that can be helped by these initiatives.