The human body is a complex system, with an estimated 30 to 40 trillion cells and dozens of organs that all interact with each other in a network. Just like any complex system, mountains of data are generated when an individual’s body is tested, and a lot of important insights can be gleaned when the data of many people is studied.
Humans have done much of the legwork up to this point, but the democratization of artificial intelligence, particularly the subsets of deep and machine learning, has dramatically sped up our ability to discover new patterns and trends that cannot be identified by doctors and researchers alone. Deep learning, according to a paper published last year in The Lancet, was shown to perform equally to healthcare professionals when it came to diagnosing disease from patients’ medical imaging. More studies need to be done, but this finding already shows the huge potential of AI for patient care. AI will never replace doctors, but its use as a tool for medical practitioners will greatly benefit patients.
Answers In The Data
The success of AI thus far has implications for patient care in a variety of settings and has vastly sped up research and development efforts to find patterns, opening the door to faster diagnoses and/or potential treatments. There are three main areas in which AI will have the most immediate impact:
1. Curative Care: Technology has sped up the processes of most things (think instant online purchases), and with AI-powered tools, medical providers can diagnose patients quicker and at the point of care. Faster diagnoses allow health professionals to prescribe patients the best treatment plans on a quicker timeline.
2. Preventative Care: AI makes it easier for doctors to detect serious conditions or genetic disorders by revealing patterns that aren’t immediately apparent. But AI doesn’t just empower doctors; it empowers patients as well. Patients armed with wearables can get real-time feedback on the behaviors they need to maintain or change to stay healthy.
3. Outcome Management: Not only can AI-powered wearables and monitoring tools help prevent health conditions, but they can also be used to help patients better adhere to treatment plans. For example, AI could optimize the best times for medication reminders or send pushes to get patients to take a walk.
The Human Element
Of course, especially in medicine, AI should be used with caution. For example, algorithms have been shown to have bias in cases of criminal law, like when an algorithm unfairly sent cops into minority neighborhoods. Any bias in medical algorithms could be disastrous and harmful to patients, so code should be developed with scientific discipline and rigor.
It’s vital for systems to be trained with a large enough set of patient data that is representative across the gender and racial spectrum, with a meticulous control for bias. As The Economist puts it, “Finding digital truth is hard because the data come from many sources and in a staggering variety of formats — which makes them hard to integrate.”
Medical professionals and researchers also need to be cognizant of the constraints of algorithms, which are built upon specific data sets. This means they may only be appropriate for certain situations. Finally, they also need to keep in mind that just because software performed well in a computer lab setting doesn’t mean it will work in the real world.
In scientific studies, there’s the concept of double-blind, meaning both test subjects and researchers are ignorant of what they’re being tested for. For example, the health professional prescribing a potential treatment doesn’t know which groups are receiving placebos. Researchers need to be cautious to not trick themselves into seeing the results they want.
AI Is A Tool
All AI systems are fed by history, so we need to be careful not to give up medical decisions entirely to them, or we risk repeating the mistakes of the past. And since biology evolves and is extremely variable, algorithms can only learn from the data they were fed with from the past. Doctors and researchers should remain at the center of important decisions, using AI as a tool.
The tech mantra of moving fast and breaking things will not work in medicine, as the consequence of having an untested system can be harmful. There is little room for error, and quality control and discipline need to be top of mind in all AI tech applied to patient care. The role of regulation in AI-based systems and products will be increasingly important going forward, but it must undergo the required changes and become more sophisticated to adapt to this new system.
While the promise of AI in medicine is huge and will lead to many exciting breakthroughs, it’s important for the medical community and regulators to be wary of its use. There is no room to cut corners here.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
No comments:
Post a Comment