Biomedical ethics has been largely reactive.Dr Matthew Strother
If you’ve never even thought about health data and ethics, you’d be one of the vast majority of people out there. But not for health data scientists and clinicians.
The field of data science and healthcare is growing at a rapid speed and ethics is a hot topic.
I was recently invited to be a writer for HACK Aotearoa 2020, a conference held at the University of Auckland around healthcare, technology, and artificial intelligence (AI).
Here’s a summary of a keynote speech delivered by Dr Matthew Strother, an oncologist, clinical pharmacologist, clinical senior lecturer at University of Otago in Christchurch, and PhD candidate in evaluating machine learning governance in healthcare.
The interface of data and ethics
Dr Matthew Strother led us through how medical ethics has evolved over the years and how that affects data ethics in healthcare.
He painted the picture of how the very different worlds of biomedical ethics and data science ethics will inevitably converge, bringing with it unique conversations and challenges that must be overcome.
While medicine and ethics have been entwined for decades, “biomedical ethics has been largely reactive” says Dr Strother, referencing the Nuremberg code formed after the reveal of inhumane Nazi experimentation, and mandates around human radiation experimentation following non-consensual experiments evaluating the effect of radiation on people.
“The legal and political responses have formed the core biomedical ethic principles” of autonomy, beneficence, non-maleficence, and justice that all clinicians are intimately familiar with.
Autonomy is expounded as informed consent. Beneficence and non-maleficence as weighing up risks and benefits for the individual, and justice as balancing risks and benefits for the society.
Dr Strother states that biomedical ethics is culturally embedded in medicine with normalised standards with a strong emphasis on individual researcher responsibility, patient consent and autonomy. Data science ethics, on the other hand, is none of these.
Data science ethics
“This is where I might get controversial. I would make the statement that data science ethics is underdeveloped. Data science has largely evolved in the business arena and arguably business ethics have been the dominant paradigm”Dr Matthew Strother
But it hasn’t been without concern.
He mentions the self-regulation of data ethics by the industry but notes that “conversations have been dominated by technical discussions and the issue of fairness and bias” says Dr Strother, “while these discussions are helpful, they are unlikely to be stand-alone” as a solution to the problems of data ethics that we are beginning to foresee.
Interestingly, Dr Strother alludes to the ethical challenges that were encountered around genetics.
“There is already a roadmap for bringing ethical norms into new areas of biomedicine from our experience with genetics” and the integration of genetics in modern medicine.
Yet despite significant work and discussion with genetic ethics, many of the issues are unresolved and overlap with issues from data science. Issues such as:
- Consent – if you consent to future yet-unknown research, is this really “informed” consent?
- Data sharing – sharing is required for publishing peer-reviewed work and incredibly important for the development of data science in healthcare
- As well as issues of privacy, indigenous rights, ownership of commercial data and withdrawal of consent.
“There’s no quick solution,” says Dr Strother, “there’s not going to be a right or wrong with most of these, it’s about having the discussions… and acknowledging the concerns.”
What do you think? Let us know in the comments below.