Challenges Hinder Machine Learning's Potential in Oncology

James Zou, PhD, assistant professor of biomedical data science at Stanford University, elaborates on some of the challenges associated with machine learning techniques and its utilization in oncology.

While artificial intelligence has been an encouraging development and is already being used to aid in cancer diagnosis, treatment, and follow-up, it comes with some challenges. According to Zou, some of the main problems experts must address in order to move forward and be able to best utilize machine learning in oncology relate to data quality, access to results, interpretability, ethical considerations, and more.

Zou notes that more research is needed to better develop these models. Experts must putting in increased efforts to make sure the models are robust, and work to make sure they are overall more generalizable.

Transcription:

0:10 | I think there are several challenges. One is that often the models require a reasonably large amount of data to train and to validate. For example, let's say if we're trying to train and develop a machine learning model to analyze histopathology images to diagnose different cancers or diseases, then the model typically requires 1000s, sometimes 10s of 1000s of images that are annotated by clinicians in order to train.

0:40 | It's also possible that these kinds of machine models can pick up and learn about these artifacts or sort of spurious correlations in the images. Maybe it learns about a particular way that images are stained, and it thinks that those kinds of staining techniques are associated with patient outcomes, or it was cancer diagnosis, when those are just sort of technical artifacts. We develop these models, we spend a lot of effort trying to make sure that these models are robust, and they are trying to mitigate these correlations to make sure that they're actually generalizable across different hospitals [and] different conditions.