Ensuring AI Accuracy in Cancer Care and Its Impact on Oncology
Cancer Care Must Prioritize AI Accuracy
Cancer care must prioritize AI accuracy in oncology. Dr. Anil Parwani, MD, PhD, professor of pathology at The Ohio State University Wexner Medical Center, addresses potential biases or limitations in the artificial intelligence (AI) algorithm used at his institution, particularly concerning factors like patient demographics or socioeconomic status, and what their future goals entail.
Testing and Validating AI Algorithms
In our hospital, we know the type of population of patients we serve in the greater Columbus area. But what if the AI algorithm was designed for patients in different countries? Dr. Parwani emphasizes the necessity of extensive validation before implementing these algorithms. At Ohio State, the team conducts internal validations, using known patient cases to train and repeatedly test the algorithms.
Equality in Diagnosis: AI and Professionals
Recent publications demonstrate that the pathologist’s diagnosis and the AI diagnosis yield comparable results; neither is inferior to the other. This careful process is crucial because once the algorithm is secured, it becomes a standard routine for cancer care. In the coming months, the goal is to adapt commercially available algorithms, designed on international patient data, to the unique needs of Ohio patients, particularly for prostate cancer diagnosis.
Expanding AI Applications in Oncology
Ohio State is applying this framework not only to prostate cancer but also to breast cancer and gastric cancer. This innovative approach extends beyond cancers, detecting microorganisms such as H. Pylori in gastric biopsies. With advancements in technology, the future of patient care is bright.
I am excited and pleased to witness the tools available today for patient care in oncology. These developments represent a turning point in how we approach cancer treatment.
This article was prepared using information from open sources in accordance with the principles of Ethical Policy. The editorial team is not responsible for absolute accuracy, as it relies on data from the sources referenced.