Discover more from Stephen's People
Who needs the human touch?
(No. 26) I do. I miss it; AI instead of humans; Frida Kahlo, live. By Stephen P. Williams
(please touch the little heart, above)
But first, this:
I miss people. I’m sheltering largely alone (with my 19 year old daughter and her girlfriend, who I rarely see). Even though I’m 62, I miss hugs and handshakes and kisses and all the rest. Here’s a good article about the all the rest.
Let the machines handle it
AI and Covid
I can’t help but admire the intelligence of the coronavirus. The efficient system that distributes it around the world. The myriad and horrifying ways it manifests in our bodies, primarily as means to distribute itself across the landscape. Its unemotional, implacable advance through populations with no regard for our greatness (hey, the Royal Tom Hanks got it) or vulnerability (Corona, the Queens, NY neighborhood hit hardest by the virus is filled with economically vulnerable people), would be admirable if the virus weren’t so destructive. It plays tricks on our bodies, with inexplicable symptoms that unleash the worst ageist impulses in society. At first we thought the old were vulnerable -- and many of us said, “Well, let them die.” Now it’s shown that the virus ravages younger people as well, unleashing strokes and lung damage and brain mishaps in people we’d never expect to suffer. The virus, as a system, possesses a natural intelligence that we seem unable to challenge.
So why not turn to artificial intelligence to deal with this scourge? The supposed rationality of AI might actually work in my old body’s favor.
If you, like me, are over 62, the knowledge that some doctors in the US and elsewhere have chosen not to use ventilators for people 62 and above when the devices are in short supply might not seem readily logical. How about letting AI do the job? Hospitals across the country face supply shortages and complex moral decisions about a disease that seems to strike young adults, middle agers, and the elderly in different ways, often with ferocity. They are turning to artificial intelligence for help in deciding how sick these people are, and what treatments they deserve.
X-ray by Qure.ai's qXR system showing lung problems in a patient.
In the UK, the Royal Bolton Hospital, which is part of the National Health Service, has designed an AI system that can read lung x-rays to assess whether or not a patient has the distinct lung markings that appear in some COVID-19 patients, thereby considerably speeding up the diagnosis. At first the tests were just used to second guess the doctor’s decisions. Now the AI systems replace the doctor in many cases.
A large number of AI companies around the world are developing systems for diagnosing COVID-19, deciding treatments, and tracking potential carriers. One system was developed by qure.ai, beneficiary of one of the tech world’s most marvelous intentional misspellings. In France at this moment, it can take several days to have a lung scan read, because the radiologists are overloaded with patients. A company called Vizyon has developed an AI diagnosis system that’s accurate and fast. The various AI innovators aren’t driven by altruism. They see COVID-19 diagnosis as a way to get hospitals comfortable with AI in a way that will keep them coming back even after the pandemic fades. It’s called market penetration.
Qure.ai points out abnormalities in a chest x-ray, and determines the likelihood that it is COVID-19.
In Israel, Maccabi Healthcare Services uses AI to figure out which of its 2.4 million subscribers stand the biggest chance of getting severe COVID-19 symptoms. The company that designed the system had already trained it to decide who was most at risk from flu. They adapted that data to COVID-19 symptoms and decided that 40,000 of their members were at heightened risk of suffering severe cases. Those people are the first to receive tests for the virus. The AI doesn’t interact with the people to make its decisions. Rather, it combs through their medical records.
So far, AI has proved generally good at predicting potential cases of COVID-19, though it isn’t scaling rapidly. Qure.ai worked with experts to list the typical signs of COVID-induced pneumonia in the lungs: opaque spots, ground glass appearance and other signs. Then it trained the AI to recognize these symptoms, using x-ray images as teaching tools. An initial study of 11,000 people to validate the process found the tool worked with 95 percent accuracy.
A lot of the AI tools are hastily being put into action, because the pandemic is so threatening. Hopefully, no patients will be harmed. Some people predict that many doctors will flee the profession after witnessing the horrors of this pandemic. If so, it might be time for more machines to join the profession. Let’s just hope the hospitals vet the AI tools well, because some doctors say there’s a lot of AI junk out there. No surprise: there’s plenty of AI junk everywhere.
The state of AI in medicine. (Long and complex, but interesting):
Actual footage of Frida Kahlo, who in some ways transcended pain and aging, or so I like to think
Music: Esa Noche by Cafe Tacuba, one of my favorite bands
The other day one of my kids felt hopeless about the future. No wonder, given the times we’re in. My middle aged and older friends have shared similar feelings. I’m lucky to rarely feel hopeless, but I’m sympathetic to those who do. Maybe this will help.
“This is the time to be slow,
Lie low to the wall
Until the bitter weather passes.
Try, as best you can, not to let
The wire brush of doubt
Scrape from your heart
All sense of yourself
And your hesitant light.
If you remain generous,
Time will come good;
And you will find your feet
Again on fresh pastures of promise,
Where the air will be kind
And blushed with beginning.”
―John O'Donohue, To Bless the Space Between Us: A Book of Blessings.
Fascinating news for the hopeless, I hope
What is panic?
Are you anxious about your anxiety?
Beautiful insight into a rare form of dementia:
I found this video in Aeon magazine, a consistently remarkable publication.