Recently, researchers from Case Western Reserve University reveled an underlying cellular difference between Black and White cancer patients with the help of artificial intelligence (AI). The research team studied close to 400 male patients with prostate cancer in six locations over a period of three years.
AI was used to analyze digitized images of cancer issues extracted from these patients. The finding believed would build on existing population specific information, especially the biological disparities at a cellular level that found to exist between races and improve overall care for Black patients with prostate cancer as more targeted medical interventions can be generated for individuals within specific population.
Incorporate an element of race into AI
According to Anant Madabhushi, F. Alex Nason Professor II of Biomedical Engineering and Director of the Center for Computational Imaging and Personalized Diagnostics (CCIPD) at Case Western Reserve University who headed the research team, their primary aim was to understand and find out if there’s any biological difference in prostate cancer. If there is, can it be a function of one’s race and ethnicity?
They examined disease management; as in which patient following a prostatectomy will have a higher risk of disease recurrence and who may benefit from adjuvant therapy (i.e., chemo- or radiotherapy after a surgery to decrease the risk of cancer recurring). At the same time, the research team also used AI to seek out patterns from both tumor and stroma (i.e., tissues outside the tumor) images, for a better understanding whether patients had responded well to a particular treatment, their chances of survival and rate of relapse.
However, as researchers look at the patient pool, there was an overwhelming percentage of White patients (i.e., 80%). They immediately knew they could be doing a disservice to the non-White patients, so there’s a need to incorporate the race element in their AI model. Once they have got that fixed, the research team realized the accuracy in determining whether Black prostate cancer patients will suffer a relapse increased by six-fold.
Implications for future studies
The research team noted while there are clear scientific evidences that racial differences are present in all cancers but it’s unclear whether they can be explained by access to care, socioeconomic status, or mere biological variances.
AI had made it easier for researchers to highlight subtle distinctions that are present in patients from a diverse background because it “look at, and actually measure, hundreds of thousands, even millions, of cancer cells to see features that a human could never see – including structural characteristics,” said Hersh Bhargava, a PhD student at the University of California-San Francisco who was involved in the study.
At a higher level, the research team will also like to know if such discovery will disclose a possible research bias at a human level and emphasis on the need to build AI models that are “meant for a particular population.
“We know now that the risk is that if you just build a model for all patients, you will actually perform worse for patients in the minority and that’s something we cannot accept, even if it’s not something we did intentionally. So, if you want to work on patients from all populations, you have to deliberately include a population-specific aspect,” added Patrick Leo, a graduate Biomedical Engineering student at Case Western Reserve University.