Researchers are developing a deep learning network capable of identifying disease biomarkers with much greater accuracy.
Experts from the Cheriton School of Computer Science at the University of Waterloo have created a deep neural network that achieves 98 percent recognition of peptide features in a data set. Scientists and medical professionals thus have a greater chance of discovering possible diseases by analyzing tissue samples.
There are several existing techniques for detecting disease by analyzing the protein structure of biosamples. Computer programs are playing an increasingly important role in this, by examining the large amounts of data that arise from such tests in order to localize specific disease markers.
“But existing programs are often inaccurate or can be limited in their underlying functions by human error,” said Fatema Tuz Zohora, a PhD student at the Cheriton School of Computer Science.
“In our research, we have created a deep neural network that achieves 98 percent recognition of peptide features in a data set. We are working to make disease detection more accurate to provide doctors with the best tools.”
Peptides are the chains of amino acids that make up proteins in human tissue. It is these small chains that often show the specific disease markers. With better tests, diseases can be detected earlier and with greater accuracy.
The Zohora team calls their new deep learning network PointIso. It is a form of machine learning or artificial intelligence that has been trained on a huge database of existing sequences of biosamples.
“Other methods of detecting disease biomarkers usually have many parameters that have to be set manually by field experts,” said Zohora. “But our deep neural network learns the parameters itself, which is more accurate, and automates the approach to disease biomarker discovery.”
The new program is also unique in that it is not trained to look for just one type of disease, but rather to identify the biomarkers associated with a range of diseases, including heart disease, cancer and even COVID-19.
“It is applicable to any type of disease biomarker discovery,” said Zohora. “And because it’s essentially a pattern recognition model, it can be used to recognize any small object in large amounts of data. There are so many uses for medicine and science; it is exciting to see what opportunities this research opens up and how it can help people. “
Source of the story: