Reading time: 2 minutes
#Healthcare
In an inefficient sector like #medicine, where allocation of resources and workforce is far from optimal,
AI could be a disruptive innovation making us able to overcome all the challenges we will face in the
future: aging population, scarcity of resources, rising costs for new technologies.
One of the most promising AI is called #Watson, a supercomputer capable of answering questions posed
in natural language and developed by #IBM that became famous some years ago beating two champions
during the American TV quiz show Jeopardy!
In healthcare, IBM is investigating to see how Watson may contribute to clinical decisions: a physician
can pose a question to the machine describing symptoms and other related factors; the computer identifies
the most important ones and then checks patient data to find useful reports about medical and hereditary
history; finally, it examines available data sources and provides a list of individualized recommendations.
Recently Watson had some problems that slowed down its path to a real implementation in hospital
routine: it turned out that advices given by the supercomputer were not adequate probably because most
of the data fed to it were from hypothetical and not real cases.
Although Watson is probably the best known, it’s not the only one. Google and Amazon are both heavily
involved in AI, healthcare and analytics.
A study published in the “Annals of Oncology” describe a Google deep learning convolutional
#neuralnetwork (CNN) that was trained to identify skin cancer.
The researchers showed it more than 100,000 images of malignant skin cancers and the computer had to
indicate the diagnosis for each. The results displayed that expert dermatologists were significantly
outperformed by the program.
The war among big tech companies is open and the first to come out with a reliable AI system will have
the opportunity to expand in this new unexploited field.
How long do you think it will take and how will these discoveries impact on our #health?
#Shareyouropinion in the comments below
Credits @ Alessandro Carli