Subscribe for email alerts

Don’t miss a single Science & Enterprise post. Sign up for our daily email alerts.

Donate to Science & Enterprise

Please share Science & Enterprise

AI Analytics Boost Breast Cancer Assessment Diversity

Breast cancer cell

Scanning electron microscope image of a breast cancer cell. (Bruce Wetzel and Harry Schaefer, National Cancer Institute, NIH, Flickr https://flic.kr/p/K4F3ZT)

28 Oct. 2022. A genomic analytics company says an ethnicity algorithm added to a personal health questionnaire improves risk assessments for breast cancer across all populations. A team from MyOme Inc. in Menlo Park, California reported the findings yesterday at the annual meeting of American Society of Human Genetics in Los Angeles.

MyOme is a five year-old enterprise that applies artificial intelligence to provide a more detailed analysis of an individual’s entire genome. While genomic sequencing is becoming more common for disease diagnostics, most standard sequencing techniques test parts of the human genome, particularly those regions having genetic codes with instructions to cells for producing proteins. Advances in computing power and analytics make it possible to sequence and analyze a person’s whole genome more economically, with the technique adapted for disease detection and research.

MyOme offers whole genome sequencing and analysis for insights into the health of clients. Among its services are assessments of disease risk using machine learning and other artificial intelligence algorithms, including risks of hereditary diseases, responses to medications, and for family planning. The company says a subscriber to its service only needs a single whole-genome analysis that company uses for producing multiple assessments, as well as on-demand reports for clinical decision-making.

For the American Society of Human Genetics conference, a team led by MyOme biostatistician Placede Tshiaba reported on a statistical model for calculating breast cancer risk across a wide range of racial and ethnic populations. According to American Cancer Society, breast cancer is the second-leading cause of cancer among women, after skin cancers, responsible for thirty percent of new cancer cases among women each year. Some 288,000 new cases of breast cancer are expected in the U.S. this year, leading to more than 43,000 deaths.

Public data sets for training and validation

MyOme says the standard risk assessment calculator for breast cancer is the Tyrer-Cuzick test, which computes a woman’s risk of breast cancer for the next 10 years. The test gathers data on personal health measures and family history, including genetic susceptibility and ethnic heritage, e.g. Ashkenazi Jewish, associated with higher breast cancer occurrence. The company says most risk assessment models based on the Tyrer-Cuzick test are based on women of European ancestry, with a continuing need for a model that calculates breast cancer risk for a wider range of populations.

The MyOme team says they constructed a more inclusive breast cancer risk model using machine learning algorithms trained with data sets of more than 125,300 women representing ethnically and racially diverse populations. The researchers say they calculated polygenic or multi-ethnic breast cancer risk scores with the public 1000 Genomes Project database serving as a reference. The team combined their polygenic risk algorithms with the Tyrer-Cuzick model to create a cross-ancestry integrated risk score, or caIRS.

The researchers then validated the caIRS with nearly 25,000 cases in the Women’s Health Initiative and more than 119,000 women in the UK Biobank, open data sets for genomic researchers. Based on the validation data, the MyOme team says the caIRS risk assessments have stronger associations across all ethnic and racial groups than Tyrer-Cuzick calculations alone, covering Caucasian, Hispanic, African/African-American and Asian women. The researchers say the largest gain in performance is among Hispanic women with a 10 percent improvement over Tyrer-Cuzick alone.

“Commonly used breast cancer risk assessment tools based on clinical and family history,” says MyOme co-founder and chief medical officer Akash Kumar in a company statement, “are not well calibrated for women of certain ancestries. Developing inclusive tools that combine clinical and genetic factors is critical for better patient management.”

More from Science & Enterprise:

We designed Science & Enterprise for busy readers including investors, researchers, entrepreneurs, and students. Except for a narrow cookies and privacy strip for first-time visitors, we have no pop-ups blocking the entire page, nor distracting animated GIF graphics. If you want to subscribe for daily email alerts, you can do that here, or find the link in the upper left-hand corner of the desktop page. The site is free, with no paywall. But, of course, donations are gratefully accepted.

*     *     *

 

2 comments to AI Analytics Boost Breast Cancer Assessment Diversity