Healthcare (Commonwealth Union) – In recent years artificial intelligence has opened many new doors for a wide variety of fields, from commerce, technology to life science. The ability to analyse vast amounts of data which go back many decades within seconds have brought about many possibilities that were previously, severely limited.
This has resulted in greater treatment options, together with greater diagnostic capabilities and much better prognostic capabilities.
Researchers at McGill University have created a new artificial intelligence system capable of spotting disease indicators within individual cells that were previously undetectable.
Biomarkers and other indicators have existed and have been used well before artificial intelligence took off on a grand scale. However, the ability of artificial intelligence to analyse data with new tools as well as better technology has seen the combination of technology medicine rapidly improve.
In findings released in Nature Communications, the team showcased how the program, named DOLPHIN, could eventually allow doctors to identify illnesses at an earlier stage and make more precise treatment decisions.
“This technology could enable physicians to pair patients with the treatments most likely to benefit them, cutting down on guesswork in medical care,” explained senior author Jun Ding, assistant professor in McGill’s Department of Medicine and researcher at the McGill University Health Centre.
Researcher looking further into genetic structures had noted that disease signals often take the form of minute alterations in RNA expression. These small shifts can reveal whether a condition is present, how severe it may become, or how it could respond to particular therapies.
Traditional gene-level analysis methods compress these markers into a single measurement per gene, hiding important differences and revealing only a fraction of the picture, the researchers explained.
With new advances in artificial intelligence, it is now possible to explore the detailed complexity of single-cell data. DOLPHIN goes further than gene-level approaches by focusing on how genes are pieced together from smaller segments known as exons, offering a sharper view of cell states.
“Genes are not just one block, they’re like Lego sets made of many smaller pieces,” explained the first author Kailu Song, who is a PhD student at the McGill university, Quantitative Life Sciences program. “By looking at how those pieces are connected, our tool reveals important disease markers that have long been overlooked.”
In one trial, DOLPHIN examined single-cell data from people with pancreatic cancer and uncovered more than 800 disease indicators that traditional methods had overlooked. The system could separate patients with aggressive, high-risk cancers from those with milder forms—knowledge that could guide doctors toward more tailored treatment decisions.
In moving toward forming the ‘virtual cells ‘on a much larger scale, this advance moves science closer to the long-term vision of creating digital replicas of human cells. By producing far more detailed single-cell profiles than standard approaches, DOLPHIN makes it possible to run virtual experiments that predict how cells react to drugs before actual lab work or clinical testing—cutting both costs and development time.
The team’s next objective is to scale up the tool from a limited number of datasets to millions of cells, opening the door to even more precise and reliable virtual cell models in the future.
“DOLPHIN advances single-cell transcriptomics beyond gene level by leveraging exon and junction reads” was the name of the study by Kailu Song and Jun Ding et al., which appeared in the journal Nature Communications.
The study was backed by the Meakins-Christie Chair in Respiratory Research, the Canadian Institutes of Health Research, the Natural Sciences and Engineering Research Council of Canada as well as the Fonds de recherche du Québec.
