Skip to main content

This week’s AI Talk…

Precision medicine

I saw an article in Newsweek this week talking about precision medicine. It was written by Brandon Suh, CEO of Lunit. The article itself largely underscores what one would expect is happening in this area of health care. Precision medicine, that is care targeted specifically to the genotype and phenotype characteristics of individual patients, is showing great promise. The article provides an interesting statistic: For 10 percent of patients suffering from cancer the survival rate has dramatically increased. Current approaches are more genome-centric. The other 90 percent are going to need targeted AI, looking at all of the data available about a patient, Suh suggests. I looked up what his company does. Lunit focuses on artificial intelligence and deep learning technology as applied to cancer care. They have a number of solutions in the radiology domain predicting cancerous nodules in chest X-rays and detecting suspicious lesions in mammograms. Personalized precision medicine is going to need lots of data and lots of AI---in this, Suh is indeed right!

Musk’s new gambit

Well, it turns out it is a two-year-old gambit, one that adds to his illustrious series of ventures from Tesla to SpaceX. Two years ago, Musk co-founded Neuralink, a company focused on brain-computer interfacing. This week, the company announced that it had made significant progress---and every media outlet, including late night comedy shows, carried this news!  They have managed to implant electrodes (thousands of them) in monkeys and rats, using a sewing machine-like robot. They are applying for FDA approval to carry out a small field study on human patients, perhaps a year from now. The goal is to monitor neurological activity to gain a better understanding of the brain. If successful, this technology can be used to treat a range of disabilities. Musk never ceases to amaze me!

Almost 21

I am not referring to the drinking age, but to a tennis match! Roger Federer had two match points against Novak Djokovic last Sunday at the Wimbledon final. If he had captured either one, he would have had a record of 21 grand slam victories! But he didn’t. As a huge Federer fan, I was depressed all of Sunday over his loss. The really sad part was Roger outplayed Novak for almost the entire 5 hours of this epic 5-set match, but Novak managed to play extremely well when it counted. I relieved part of my misery by looking at the highlights posted on the Wimbledon site. Turns out these highlights are automatically curated by IBM Watson AI technology. They have been tweaking this AI engine for the past few years and this year they have improved it even further and claim that they have removed bias in their algorithms. The technology monitors, among other factors, crowd noise and player’s reactions to determine when a good point has been played. Bias creeps in when crowd noise favors one player over another or when some players are over demonstrative or have an emotional response to a point played. I have a suggestion for IBM for next year: Perhaps monitor the royal box for emotions! (Check out this article about Kate Middleton reactions to the match).

I am always looking for feedback and if you would like me to cover a story, please let me know. “See something, say something”!  Leave me a comment below or ask a question on my blogger profile page.

V. “Juggy” Jagannathan, PhD, is an AI Evangelist with four decades of experience in AI and Computer Science research.