Sept. 1, 2022 Its hard determining what the street ahead can look like for a cancer patient. Plenty of evidence is known as, just like the patients health insurance and genealogy, grade and stage of the tumor, and traits of the cancer cells. But ultimately, the outlook boils down to medical researchers who analyze the reality.
That may result in large-scale variability, says Faisal Mahmood, PhD, an assistant professor in the Division of Computational Pathology at Brigham and Womens Hospital. Patients with similar cancers can end up getting completely different prognoses, with some being more (or less) accurate than others, he says.
Thats why he and his team developed an artificial intelligence (AI) program that may form a far more objective and potentially more accurate assessment. The purpose of the study was to inform if the AI was a workable idea, and the teams results have already been published in Cancer Cell.
And because prognosis is type in deciding treatments, more accuracy could mean more treatment success, Mahmood says.
Building the AI
The researchers developed the AI using data from The Cancer Genome Atlas, a public catalog of profiles of different cancers.
Their algorithm predicts cancer outcomes predicated on histology (a description of the tumor and how quickly the cancer cells will probably grow) and genomics (using DNA sequencing to judge a tumor at the molecular level). Histology has been the diagnostic standard for a lot more than 100 years, while genomics can be used a growing number of, Mahmood notes.
“Both are actually popular for diagnosis at major cancer centers, he says.
To check the algorithm, the researchers find the 14 cancer types with data available. When histology and genomics were combined, the algorithm gave more accurate predictions than it did with either information source alone.
Not just that, however the AI used other markers just like the patients immune reaction to treatment without having to be instructed to do so, the researchers found. This may mean the AI can discover new markers that people dont even understand about yet, Mahmood says.
While more research is necessary including large-scale testing and clinical trials Mahmood is confident this technology will undoubtedly be useful for real-life patients someday, likely within the next 10 years.
“In the years ahead, we will have large-scale AI models with the capacity of ingesting data from multiple modalities, he says, such as for example radiology, pathology, genomics, medical records, and genealogy.
The more info the AI can element in, the more accurate its assessment will undoubtedly be, Mahmood says.
“Then we are able to continuously assess patient risk in a computational, objective manner.”