The use of artificial intelligence (AI) in health can save the US health care system $ 150 billion a year between 2017 and 2026, according to Accenture. Investments in the AI-based health care market are expected to reach $ 6.6 billion by 2021. In total, the market will have grown to more than ten times its original size in five years. But far from replacing humans in the health sector, AI and its transformative power should complement it. Technology will reduce costs while improving the quality of care and access.
Detection of malignant tissue difficult to see with the naked eye
In a study published Monday in the journal Nature Medicine, Google has reinforced this promise. A research project at Alphabet’s subsidiary investigated the use of AI in lung cancer screening. He concluded that some algorithms may be more effective than radiologists in detecting this most deadly cancer in the world with 1.7 million deaths a year.
As part of the initiative, Google researchers and engineers have developed a machine-learning algorithm based on 45,856 chest scanners. These came from the National Health Institute and Northwestern University. They included lung cancers at different stages. The algorithm generates a 3D model of the scanner and then searches for malignant tissue in lung nodules. These can be difficult to see with the naked eye, especially since radiologists usually look at 2D images. The algorithm finally generates an estimate of the malignancy of lung cancer. It can also rely on data from previous scanners.
Increase in detected cases and reduction of false positives
Google researchers and engineers then applied this algorithm to a single scanner to compare its accuracy with those of six certified radiologists. As a result, the algorithm detected 5% more cases of lung cancer and reduced false positives by more than 11%.
” These initial results are encouraging, but additional studies will evaluate their impact and utility in clinical practice, ” said Google. The algorithm template will be available through Google’s Cloud Healthcare API.
The Mountain View firm had begun late 2017 to apply artificial intelligence (AI) to various health problems. Other initiatives include projects to improve the diagnosis of eye diseases or the prediction of patient outcomes.