Clinical Data Standards
The data matters.
Why Does the Data Matter
Dear reader, you may have heard me rant in the past about the importance of awareness of algorithmic bias in clinical research and healthcare. Innovation in the area of precision medicine and personalized treatment plans rely heavily on data accuracy, accessibility, and integrity. To prevent algorithmic racism and sexism it is critical to 1) acknowledge these things exist in research and patient care, and is based on a deeply rooted history of racism and sexism in society, and 2) understand these biases can be trained into learning algorithms if we are not careful.
It makes sense that bias can be used to train artificial intelligence when we remember that humans are doing the training. All humans have bias. I have some, you have some, we all have some…. Becoming aware of our bias and being careful to avoid programming these biases into decisions about health care, research, and a number of other personal details requires the removal of ego and desire to promote equity in research and patient care for all.
Inserting oneself into groups that focus on ethical AI, diversity and inclusion in healthcare and research, and being dedicated to data fairness are ways to reduce the risk of coded bias.
Why I’m Excited About CDISC
The Clinical Data Interchange Standards Consortium (CDISC) was created to establish and promote the use of data standards for clinical data collection, analysis, and transmission. Ultimately the mission of CDISC is to make clinical data easy to understand and interpret. I’m excited about this because if an organization like CDISC can establish data standards based on data fairness, with a continual system of checks and balances to remain aware of any bias that may result from these standards, then over time there is a possibility of removing the racism and sexism that exists in our algorithms today.
On a practical level, if you are reading this and happen to be a minority or a female, you more than likely have experienced less-than-objective doctors and clinicians in your lifetime. I’ve mentioned before that I’ve been told I must have a higher pain tolerance because I’m a minority. Race correction exists in diagnostic algorithms, which is mind boggling because it has long been known in scientific communities that race is a social construct; it is not biological or genetic. There is no “Black gene” or “Hispanic gene”. There are social determinants of health, and inequalities of care, however, this is not an indicator of things like pain tolerance. I cannot express this point enough: the concept of race is an invented social category that has nothing to do with our DNA. In other words – a person’s skin color does not reflect their ancestry – so let’s stop coding that into our programs.
It’s 2021 and this still must be said. The frustration is maddening.
The Good News
I am optimistic and hopeful due to the increased awareness of this problem. There was a time when no one was really talking about this, and if it was brought up the reverse outrage was silencing.
Today is a new day and there is a real push to do better, be better, and promote equitable healthcare for all. There is still much to do, and I encourage all researchers, public health professionals, clinicians, data scientists, patient advocates, and any interested party to step in and get involved.