Artificial Intelligence (A.I.) has been around for a long time. Neural networks, decision support systems and bayesian inference were all around when I was an A.I. researcher in the late eighties and early nineties. But the sudden jump to prominence is a result of one major breakthrough – deep learning neural networks and computing power.
It is now possible to recognize deeper patterns as the networks are continuously trained using vast amounts of data. Any signal (sound, image, language, stock ticker, genetic codes, EKG, etc.) can be put through deep learning systems to be trained for recognition in real-time or offline. Patterns in these signals that are not evident to humans can be discovered by computing power. You can recognize this in our everyday life through smartphones, smart gadgets and Teslas.
However, A.I. still lags in reasoning. Reasoning is critical to calling any machine “intelligent.” This is precisely the reason why A.I. will never catch up with humans. Apart from emotions, reasoning and abstract thinking are uniquely human characteristics that are hard to replicate.
What is the implication of this trend in healthcare for the future? How does this affect the healthcare workforce? Will this drastically change healthcare outcomes for the masses? Answers to these questions are unraveling continuously as breakthroughs occur periodically.
In diagnostic radiology and other image recognition fields, A.I. is already here. Several reports have shown that computers can recognize most imaging better than professionals. A.I. will become part of standard imaging equipment and the workforce requirements will adapt to this reality. The role of professionals will be supervisory with routine imaging but more hands-on with studies that require complex, context-sensitive interpretation.
Significant breakthroughs are bound to happen in clinical genomics and individualized therapies as more data is accumulated. Targeted preventive health will come more prevalent with increasing accuracy of genomic prediction. Individualized therapies will become far more effective as we accumulate more data and evidence. A.I. is already being applied to drug discovery and vaccine formulation techniques.
In general, A.I. will aid in discovering clinical evidence from a vast sea of data and convert this into decision support protocols that can supplement clinical practice. Our increased demand on the physician workforce due to demographic shifts will be shaped with these tools as well as increased participation of adjunct providers. Another side effect could be a better lifestyle option for physicians, as A.I. and technology could lessen the burden on their time with efficiencies afforded by automation.