What if there was a way that illnesses, including cancer, could be detected in an individual simply by analyzing their breath? While this may seem like the premise of a Science Fiction movie, a recent announcement made by Nvidia says that artificial intelligence can smell illnesses within the breath of a human.
New Technology Can Predict Illnesses Better Than Humans
The News Center on Nvidia’s website stated on Wednesday the development of an AI program that can predict illnesses with better than-human average performance of an individual by analyzing their breath. According to the website, researchers from the Edinburgh Cancer Center in the United Kingdom, Loughborough University, the University of Edinburgh and Western General Hospital have developed recently a deep learning-based method capable of analyzing compounds within the breath of a human and detect illnesses that includes cancer.
Researcher Andrea Soltoggio posted on Smithsonian.com that “the sense of smell is used by animals and even plants to identify hundreds of different substances that float in the air. But compared to that of other animals, the human sense of smell is far less developed and certainly not used to carry out daily activities. For this reason, humans aren’t particularly aware of the richness of information that can be transmitted through the air, and can be perceived by a highly sensitive olfactory system.”
Developing the Efficiency of Their Neural Network
The team was able to build their neural network by using TensorFlow deep learning frameworks, Nvidia Tesla GPUs and the cuDNN-accelerated Keras. The information that was utilized for the growing of the neural network came from volunteers that had various forms of cancer that were receiving radiotherapy; this is according to researcher Angelika Skarysz who is a PhD research student from Loughborough University.
The team said that to increase the efficiency of the neural network, they increased the first training information by utilizing data augmentation; the convolutional neural network received one-hundred times of augmentation. According to the team, this is the first successful machine learning attempt at learning ion patterns and detecting compounds from raw GC-MS data. The researchers explained that the convolutional neural network achieved the best performance when implemented with two particular features: one-dimensional filters to adapt to the particular structure of GC-MS data, and a three-channel input to read high, medium, and low-intensity signals from the highly variable GC-MS spectrum. The novel approach was shown to discover labeling errors from human experts, suggesting better-than-human average performance.
Also, the researchers utilized the GPUs for inference; with this case, it called for scanning of breath samples. Soltoggio said that computers equipped with this technology only take minutes to autonomously analyze a breath sample that previously took hours by a human expert. While artificial intelligence is making the complete process less expensive, more importantly is the results are more reliable.
Next month, the finding will be presented in Rio Janeiro, Brazil at the International Joint Conference on Neural Networks (IJCNN). Recently, the research paper was posted on Research Gate.