CheXzero’s algorithm has studied thousands of chest x-rays and their accompanying clinical reports to make doctor-level diagnoses of diseases from the images. About it declared experts from Harvard Medical School.
The researchers trained the model on a publicly available set of over 377,000 chest radiographs and 227,000 related clinical reports. The algorithm was able to independently associate certain types of images with their description. This avoided the need to structure and manually annotate the data.
The scientists tested the performance of CheXzero on separate datasets from two different institutions, one of which is located in another country. The researchers wanted to make sure the algorithm was able to match images with their descriptions, even if the reports contain different terminology.
According to experts, the AI model turned out to be more effective than analogs in diagnosing pneumonia, as well as damage and collapse of the lungs. The accuracy of CheXzero is almost comparable to the work of radiologists, they added.
“For the first time, the algorithm learned from unstructured text and was able to match physicians in performance, and also demonstrated the ability to predict multiple diseases from patient radiographs with a high degree of accuracy,” said Ekin Tiu, co-author of the report.
Scientists have opened the source code of the project to other researchers.
“We hope that the algorithm can be applied to CT scans, MRIs and echocardiograms to teach it to detect a wider range of diseases in other parts of the body,” said project leader Pranav Rajpurkar.
In his opinion, diagnostic AI models that require minimal supervision during operation will help people get medical care in countries with a shortage of specialists.
Recall that in September, American scientists announced that they would develop an artificial intelligence algorithm to detect diseases by voice.
In March 2021, researchers used AI to decipher radiographs of 300-year-old letters.
Subscribe to Cryplogger news in Telegram: Cryplogger AI – all the news from the world of AI!
Found a mistake in the text? Select it and press CTRL+ENTER
CheXzero’s algorithm has studied thousands of chest x-rays and their accompanying clinical reports to make doctor-level diagnoses of diseases from the images. About it declared experts from Harvard Medical School.
The researchers trained the model on a publicly available set of over 377,000 chest radiographs and 227,000 related clinical reports. The algorithm was able to independently associate certain types of images with their description. This avoided the need to structure and manually annotate the data.
The scientists tested the performance of CheXzero on separate datasets from two different institutions, one of which is located in another country. The researchers wanted to make sure the algorithm was able to match images with their descriptions, even if the reports contain different terminology.
According to experts, the AI model turned out to be more effective than analogs in diagnosing pneumonia, as well as damage and collapse of the lungs. The accuracy of CheXzero is almost comparable to the work of radiologists, they added.
“For the first time, the algorithm learned from unstructured text and was able to match physicians in performance, and also demonstrated the ability to predict multiple diseases from patient radiographs with a high degree of accuracy,” said Ekin Tiu, co-author of the report.
Scientists have opened the source code of the project to other researchers.
“We hope that the algorithm can be applied to CT scans, MRIs and echocardiograms to teach it to detect a wider range of diseases in other parts of the body,” said project leader Pranav Rajpurkar.
In his opinion, diagnostic AI models that require minimal supervision during operation will help people get medical care in countries with a shortage of specialists.
Recall that in September, American scientists announced that they would develop an artificial intelligence algorithm to detect diseases by voice.
In March 2021, researchers used AI to decipher radiographs of 300-year-old letters.
Subscribe to Cryplogger news in Telegram: Cryplogger AI – all the news from the world of AI!
Found a mistake in the text? Select it and press CTRL+ENTER