AI 🤖 In the NICU : Ethics C🤍

Dear 99 NICU community I just completed Ethics of AI Course Diploma with the University of Michigan with Prof.H.V. Jagadish

This prompt to write a case study as applied to neonates. At the end ( sources ) you'll find the last article on the use of deep learning to predict outcomes

In the last decade, many different projects around the globe have attempted to use artificial intelligence to improve outcomess of premmies, estimate age, and predict the possible complications in the NICU stay.

Several ethical issues and how the neonatal community is trying to use it for the good. Each premature infant generates a huge amount of data.  This is where AI comes in bringing insight, but with an ethical implication beyond data privacy with many ramifications, extending data breaches to threaten health disparities due to population sample usesd to train AI models. Thus,

This Health data mining has a huge potential for harm.

In past, there have been many attempts by tech companies to monetize this data.

In the early 2000  with the advent of genomics legislators tried to protect the user by passing laws such as Gattaca, to prevent discrimination.  This was not enough as tech usually is ahead of legislation.

Starting with the bad. Here is an example that struck me the most as explained by the “ website.

A recent ethical misconduct case illustrated by David Marco is Google’s “project Nightingale”, where they were able to access personal health records to develop their AI. In doing so, violating several ethical principles starting with consent without asking for consent from the patient and without notifying medical professionals.Secondary violations in terms of data privacy and transparency about the objective of the project.  The company claimed they respect the HIPAA legislation business clauses. In my opinion, clearly not enough and not acceptable. In addition, breaking Dr. Jagadish's first rule - do not be surprised.

When I heard the episode  “Harnessing the Power of AI for Better Neonatal Outcomes (ft NeoMIND-AI founders Drs. Barry and McAdams)” I was intrigued but skeptical due to data privacy issues, discrimination and equity ramifications. After listening to the @nicupodcast episode It hits a lot of good points about how tech can bring better care to the NICU, in terms of insights and capabilities augmented for clinicians. Yet it still feels it needs care and consciousness. Here there are data privacy concerns, so we need to take care with parental consent.  In addition,  Algorithms need to be tweaked to prevent discrimination against minorities.  The doctor also explains that data sharing is tricky in terms of ethics.

https://podcasts.apple.com/es/podcast/the-incubator/id1566031191?i=1000655297640

To conclude that AI needs to be used with consciousness, transparency, solid data validation and engaging the stakeholders such as parents and clinicians.

Sources /
Data Ethics: Master Responsible Data Use: https://www.ewsolutions.com/understanding-data-ethics/

The ethics of data mining in healthcare: challenges, frameworks, and future directions - PMC

AI can predict preemies’ path, Stanford Medicine-led study shows