How A.I. will liberate doctors from keyboards and basements

Giving A.I. a role in health care can help both doctors and patients.

ERIC TOPOL: Technology can't enhance humanity, that it's depersonalizing, that it's going to detract. I actually think it's just the opposite in medicine, because if we can outsource to machines and technology, we can restore the human bond, which has been eroding for decades. So what I mean by deep medicine is really a three part story: The first is what we call deep phenotyping. And that is a very intensive, comprehensive understanding of each person at every level. So that's the idea of knowing all about their biology, not just their genome, their microbiome, and all the things the different layers of the person, but also their physiology through sensors, their anatomy through scans, their environment through sensors as well, and then traditional data.

So that's deep phenotyping. Now, no human being can process all that data, because it's dynamic, and it's actually quite large to deal with. That's why we have deep learning. That's a type of artificial intelligence which takes all of these inputs and it really crystallizes, distills it all. And that gets us to deep empathy. And the deep empathy is when we have this outsourcing to machines and algorithms. We have all of this data, and we now can get back to the human side of this connection. Well, where deep learning works the best today is with images. And so medical images are especially ideal because it turns out that radiologists miss things in more than 30 percent of scans that are done today. So in order to not miss these things, you can train machines to have vision better than humans. The difference here is that the radiologists can put more context in it, but the machines, they're very complementary.

They can pick up things that radiologists wouldn't see, like a nodule on a chest X-ray or an abnormality on an MRA that would be missed because radiologists read 50 to 100 scans per day. There's many times that we just don't see things. So when you bring the two together, you get the best economy. It doesn't mean we're going to eliminate the need for radiologists. It's going to make the accuracy and the speed much better. And what I project is that we're going to see a time when radiologists move out of the basement in the dark and actually connect with patients, because they want to see patients. They want to be able to share their expertise, and they don't have a vested interest about doing an operation or procedure. They just want to report what they find and communicate that. So I think we're going to see a reshaping of radiology because of this remarkable performance enhancement through AI.

There's a lot of use of AI in the hospital setting, because when patients come in, and trying to predict what's going to happen, we're not so good at that generally in medicine. So almost everything you can think of there have been algorithms tested. One example is sepsis. So what's going to happen? Does the person have sepsis, a serious infection? Are they going to decompensate and possibly die from sepsis? We're not so great at that, it turns out, by algorithm. But what we have learned is that we can use the same machine vision, whether it's nurses, doctors, people who are circulating in a room of a hospital, to see whether or not they're doing appropriate handwashing, and to set off a signal that, no, it wasn't done and needs to be done. So there's lots of things about patient safety with machine vision.

So for example, preventing falls, seeing that someone's walking is unsteady. Another great example is in the intensive care unit. Some people can pull out their breathing tube, and now we have machine vision that can monitor that so that we don't have a nurse that has to be in the room all of the time. Well, the biggest thing that we need down is the gift of time. Rather than to have this AI support? And it's at two levels. So if you can get rid of keyboards, or liberate from keyboards, reestablish face-to-face eye contact, that's a good start. It's going to happen. But also the patients now can have algorithms generating their own data, whether it's their heart rhythm, or their skin rash, or a possible urinary tract infection, they can get that diagnosed now by an algorithm. That frees up, again, doctors to take care of more serious matters, and that's what is so exciting if we grab this opportunity, which I don't know if we'll see it again for generations, if ever, because this technology offers that potential. But it won't happen by accident.

If we're not taking on this, really, activism to promote the gift of time and turning inward, as the medical community, if we don't do this, we're going to see even worse squeeze than we have now. This is an opportunity that we just can't miss.

  • Machines can help doctors by spotting abnormalities in X-rays or MRA scans that the physicians themselves may have missed.
  • A.I. can also help physicians by analyzing data and, through the use of algorithms, produce possible diagnoses.
  • The freed up time, as doctors make their rounds, can help physicians establish better connections with their patients, which in turn can lead to better treatment plans.

A map of America’s most famous – and infamous

The 'People Map of the United States' zooms in on America's obsession with celebrity

Image: The Pudding
Strange Maps
  • Replace city names with those of their most famous residents
  • And you get a peculiar map of America's obsession with celebrity
  • If you seek fame, become an actor, musician or athlete rather than a politician, entrepreneur or scientist

Chicagoland is Obamaland

Image: The Pudding

Chicagoland's celebrity constellation: dominated by Barack, but with plenty of room for the Belushis, Brandos and Capones of this world.

Seen from among the satellites, this map of the United States is populated by a remarkably diverse bunch of athletes, entertainers, entrepreneurs and other persons of repute (and disrepute).

The multitalented Dwayne Johnson, boxing legend Muhammad Ali and Apple co-founder Steve Jobs dominate the West Coast. Right down the middle, we find actors Chris Pratt and Jason Momoa, singer Elvis Presley and basketball player Shaquille O'Neal. The East Coast crew include wrestler John Cena, whistle-blower Edward Snowden, mass murderer Ted Bundy… and Dwayne Johnson, again.

The Rock pops up in both Hayward, CA and Southwest Ranches, FL, but he's not the only one to appear twice on the map. Wild West legend Wyatt Earp makes an appearance in both Deadwood, SD and Dodge City, KS.

How is that? This 'People's Map of the United States' replaces the names of cities with those of "their most Wikipedia'ed resident: people born in, lived in, or connected to a place."

‘Cincinnati, Birthplace of Charles Manson'

Image: The Pudding

Keys to the city, or lock 'em up and throw away the key? A city's most famous sons and daughters of a city aren't always the most favoured ones.

That definition allows people to appear in more than one locality. Dwayne Johnson was born in Hayward, has one of his houses in Southwest Ranches, and is famous enough to be the 'most Wikipedia'ed resident' for both localities.

Wyatt Earp was born in Monmouth, IL, but his reputation is closely associated with both Deadwood and Dodge City – although he's most famous for the Gunfight at the O.K. Corral, which took place in Tombstone, AZ. And yes, if you zoom in on that town in southern Arizona, there's Mr Earp again.

The data for this map was collected via the Wikipedia API (application programming interface) from the English-language Wikipedia for the period from July 2015 to May 2019.

The thousands of 'Notable People' sections in Wikipedia entries for cities and other places in the U.S. were scrubbed for the person with the most pageviews. No distinction was made between places of birth, residence or death. As the developers note, "people can 'be from' multiple places".

Pageviews are an impartial indicator of interest – it doesn't matter whether your claim to fame is horrific or honorific. As a result, this map provides a non-judgmental overview of America's obsession with celebrity.

Royals and (other) mortals

Image: The Pudding

There's also a UK version of the People Map – filled with last names like Neeson, Sheeran, Darwin and Churchill – and a few first names of monarchs.

Celebrity, it is often argued, is our age's version of the Greek pantheon, populated by dozens of major gods and thousands of minor ones, each an example of behaviours to emulate or avoid. This constellation of stars, famous and infamous, is more than a map of names. It's a window into America's soul.

But don't let that put you off. Zooming in on the map is entertaining enough: celebrities floating around in the ether are suddenly tied down to a pedestrian level, and to real geography. And it's fun to see the famous and the infamous rub shoulders, as it were.

Barack Obama owns Chicago, but the suburbs to the west of the city are dotted with a panoply of personalities, ranging from the criminal (Al Capone, Cicero) and the musical (John Prine, Maywood) to figures literary (Jonathan Franzen, Western Springs) and painterly (Ivan Albright, Warrenville), actorial (Harrison Ford, Park Ridge) and political (Eugene V. Debs, Elmhurst).

Freaks and angels

Image: Dorothy

The People Map of the U.S. was inspired by the U.S.A. Song Map, substituting song titles for place names.

It would be interesting to compare 'the most Wikipedia'ed' sons and daughters of America's cities with the ones advertised at the city limits. When you're entering Aberdeen, WA, a sign invites you to 'come as you are', in homage to its most famous son, Kurt Cobain. It's a safe bet that Indian Hill, OH will make sure you know Neil Armstrong, first man on the moon, was one of theirs. But it's highly unlikely that Cincinnati, a bit further south, will make any noise about Charles Manson, local boy done bad.

Inevitably, the map also reveals some bitterly ironic neighbours, such as Ishi, the last of the Yahi tribe, captured near Oroville, CA. He died in 1916 as "the last wild Indian in North America". The most 'pageviewed' resident of nearby Colusa, CA is Byron de la Beckwith, Jr., the white supremacist convicted for the murder of Civil Rights activist Medgar Evers.

As a sampling of America's interests, this map teaches that those aiming for fame would do better to become actors, musicians or athletes rather than politicians, entrepreneurs or scientists. But also that celebrity is not limited to the big city lights of LA or New York. Even in deepest Dakota or flattest Kansas, the footlights of fame will find you. Whether that's good or bad? The pageviews don't judge...

Keep reading Show less

Thumbs up? Map shows Europe’s hitchhiking landscape

Average waiting time for hitchhikers in Ireland: Less than 30 minutes. In southern Spain: More than 90 minutes.

Image: Abel Suyok
Strange Maps
  • A popular means of transportation from the 1920s to the 1980s, hitchhiking has since fallen in disrepute.
  • However, as this map shows, thumbing a ride still occupies a thriving niche – if at great geographic variance.
  • In some countries and areas, you'll be off the street in no time. In other places, it's much harder to thumb your way from A to B.
Keep reading Show less

Michio Kaku: Genetic and digital immortality are within reach

Technology may soon grant us immortality, in a sense. Here's how.

Videos
  • Through the Connectome Project we may soon be able to map the pathways of the entire human brain, including memories, and create computer programs that evoke the person the digitization is stemmed from.
  • We age because errors build up in our cells — mitochondria to be exact.
  • With CRISPR technology we may soon be able to edit out errors that build up as we age, and extend the human lifespan.
Keep reading Show less