Balance of Hype and Caution with AI & ML - SMARTMD Balance of Hype and Caution with AI & ML - SMARTMD

Machine Learning and AI have been the center of much hype over the past few years. This hype is not entirely unwarranted. But, we have also seen how these new technologies can be misused or used blindly, having a negative impact on the very people they are intended to help. This week, we see some good and some not-so-good news about ML & AI.

Governance plays a key role in the equitable implementation of new technologies

At HIMSS21, Kevin Ross spoke about the need for a governance council that is made up of diversified fields. By creating a group with a wide scope of backgrounds and expertise, questions are more likely to arise that can help mitigate inequitable applications of ML & AI. Though well-intentioned, new technologies are developed by individuals with their own perspectives and bias. Having a diverse group overseeing the application of the new tools can help mitigate or avoid these biases.

Read more at MobihealthNews

Too many patients. Not enough time. Over-reliance on flawed AI tools

The US has been dealing with an opioid crisis as government agencies and healthcare providers have struggled to deal with the addiction epidemic.  As part of the effort, AIs have been developed to identify patterns of behavior that may indicate a patient is ‘shopping’ doctors to garner prescriptions to satiate their addiction. We are seeing patients who are flagged by the new AI driven-tools, getting kicked out of hospitals and dropped by their doctors. Unfortunately, two things may result: 1) Patients with addictions are being summarily dismissed with no assistance, and 2) the AIs are flagging patients who are not shopping doctors and not addicted but are truly in need of pain management. 

Read more on WIRED

Brain’s cellular changes are recognized as precursors to dementia

Pattern recognition has been one of the leading applications for AI & ML. Researchers at the University of Cambridge have taken this approach to detect Alzheimer’s before symptoms become apparent. By combining readings from brain scans with cognitive tests, the tools can predict with about 80% accuracy the likelihood of Alzheimer’s. The system has shown promise in a research setting and is heading to real-world testing.

Read more at MedicalXpress

As we gain experience with artificial intelligence and machine learning we will find ourselves in uncharted waters. We’ll learn about potential problems well before they manifest themselves in our daily lives, we’ll need to develop new ways to manage information and tools, and we have the real threat of seeing our own unknown biases integrated into far-reaching medical technology. With well-managed AI and ML, we will see a great deal of good coming to healthcare.