Google has published new research in Artificial Intelligence (AI), which can be used for tuberculosis (TB) screening. As per a blog post, the researchers at Google have developed AI research that builds on the existing medical imaging work and helps patients with Tuberculosis be identified for consecutive tests. The company confirms that this diagnostic procedure is a preliminary test and can save up to 80 percent of an expensive diagnostic test per TB patient.
The previous research which detected nodules or fractures in chest X-rays or collapsed lungs was used to build the latest AI tool. To ensure that the AI tool is effective on a wide range of races, Google trained the model using de-identified data from nine countries. It was then tested on cases from five countries.
In their study, the Google AI researchers have found that based on the chest X-ray, the deep learning system can identify patients who are likely to have active TB. It provides any number between 0 and 1, indicating the severity of tuberculosis in the patients.
Google’s AI tool was able to detect the cases of TB accurately. The blog also states that the detection rates of false-negative and false-positive cases were similar to 14 radiologists. One of the highlights of this latest AI tool is that it can maintain detection accuracy even while examining HIV patients. Generally, it is difficult to screen TB in HIV patients as their chest X-rays may differ from those of other TB patients.
For the implementation of this AI tool, the tech giant says that there needs to be agreement on the risk level, making a person eligible for additional testing.