The medical application of AI is opening up new horizons. The diagnosis of lung cancer, the deadliest form of the disease in the world, can be greatly enhanced thanks to an algorithm developed by Google, with machine learning proving more effective at detecting malignant tissue than the naked eye, thus reducing the risk of false positives.
According to Accenture, the US healthcare system could save up to $150bn a year through to 2026 thanks to the use of artificial intelligence (AI). The AI healthcare market is set for a boom in the next five years, with investment potentially reaching $6.6bn in 2021, as reported by the website Frenchweb.
Such an expansion will not come at the expense of the role people play in the health sector, however. The contribution AI has to make is as an additional technology that can be applied solely to reduce costs, improve the quality of healthcare, and make it more accessible.
Improved detection of malignant tissue
AI offers many potential healthcare benefits, including the detection of lung cancer. According to a study published recently in the journal Nature Medicine, it could usher in remarkable advances in the treatment of the disease. A research project headed up by Google has demonstrated that some algorithms are better at diagnosing lung cancer than radiologists, a development that will give genuine hope to sufferers of the disease, which is the deadliest form of cancer in the world and causes 1.7 million deaths every year.
Making use of machine learning, Alphabet’s researchers and engineers trained their AI on 45,856 patient CT scans, including lung cancers at varying stages of progression, provided by the National Institutes of Health and Northwestern University. Capable of generating a 3D model of the scan, the algorithm then analyses it to detect malignant tissue that radiologists working with 2D images would find difficult to spot. The algorithm is even capable of estimating the degree of malignancy of the cancer.
A reduction in false positives of over 11%
During the course of the study, the Google team compared the accuracy of its tool with that of six qualified radiologists. Not only was the algorithm capable of detecting 5% more cancer cases, it also cut false positives by 11%. The programme achieved a 94.4% success rate.
“These very promising results represent a revolution,” said Dr Alain Livartowski, data director at the Curie Institute’s hospital complex in Paris. Google has confirmed that these initially encouraging results will be followed by additional studies that will seek to assess their impact and usefulness in clinical practice. The company also said that it intends to serve the algorithm model through its Cloud Healthcare API.
Frenchweb reports that Google began exploring AI’s healthcare capabilities in 2017, mainly with a view to improving the diagnosis of eye diseases.
Cover photo credits: British Medical Journal
Contact Allianz Partners
Feb 11, 2019
American researchers have developed an algorithm that can accurately detect precancerous cells in the cervix. In the long term, healthcare staff might be able to screen for this type of [...]
May 24, 2018
According to the results of research carried out by scientists at the University of California at San Francisco (UCSF) and the Mayo Clinic, software aimed at evaluating breast density c [...]