UnpackAI student reflection: Deep learning, AI, and healthcare.
When Jeremy Howard mentioned a list of fields where deep learning produces the best results, my attention was immediately captured by the critical domain of healthcare. Having completed a Ph.D. at Taiyuan University of Technology, my research specialty was about deep learning applied to medical image analysis. I am now extending my deep learning knowledge with the help of UnpackAI.
To me, the promise of AI for health care, also the reason why I pursued this research topic, was rather obvious: the promise of specialized, advanced healthcare at scale. The country where I come is still on the path to economic development and but still has a lot to work on. In the medical sector, the number of generalist doctors per habitants is quite a few, especially in the rural areas, and that of specialized doctors is even less.
In short, while certain are worried about a risk of loss of labors due to automation and AI, for places such as my countries, AI is rather an opportunity to raise the quality of healthcare for the masses, because it holds the potential to multiply the capacity of rather limited medical workforces.
Better than humans?
In many domains, deep learning has demonstrated performance that is often equal if not superior to that of humans. The popular story of AI AlphaGod beating humans is interesting, but before anybody starting to picture the advent of real life Terminators, let's consider that AI can now do better diagnose work than human doctors sometimes. Truth is if something is done by humans, and there is available data that computers can learn from, chances are it can be done by machine, and soon done better. For example, Deep Learning tools are now better than human Dermatologists in Melanoma Detection, or they can do their jobs faster than humans.
There are countless examples like these and going through the scientific literature is just fascinating. I was fortunate enough to participate in research work that allowed me to witness firsthand how the best expertise of healthcare could be at the reach in remote locations with specific needs through machine learning in a world often worried about scarcity, and this is great.