These Algorithms Could Bring an End to the World's Deadliest Killer:
In rural India and other places where tuberculosis is rampant, A.I. that scans lung X-rays might eliminate the scourge.
The Chinchpada Mission Hospital in the rural Indian state of Maharashtra.
In some of the most remote and impoverished corners of the world, where respiratory illnesses abound and trained medical professionals fear to tread, diagnosis is increasingly powered by artificial intelligence and the internet.
In less than a minute, a new app on a phone or a computer can scan an X-ray for signs of tuberculosis, Covid-19 and 27 other conditions.
TB, the most deadly infectious disease in the world, claimed nearly 1.4 million lives last year. The app, called qXR, is one of many A.I.-based tools that have emerged over the past few years for screening and diagnosing TB.
The tools offer hope of flagging the disease early and cutting the cost of unnecessary lab tests. Used at large scale, they may also spot emerging clusters of disease.
"Among all of the applications of A.I., I think digitally interpreting an image using an algorithm instead of a human radiologist is probably furthest along," said Madhukar Pai, the director of the McGill International TB Center in Montreal.
Artificial intelligence cannot replace clinicians, Dr. Pai and other experts cautioned. But the combination of A.I. and clinical expertise is proving to be powerful.
"The machine plus clinician is better than the clinician, and it's also better than machine alone," said Dr. Eric Topol, the director of the Scripps Research Translational Institute in San Diego and the author of a book on the use of A.I. in medicine.
In India, where roughly one-quarter of the world's TB cases occur, an app that can flag the disease in remote locations is urgently
(Score: -1, Redundant) by Anonymous Coward on Monday November 23 2020, @08:33PM (4 children)
Cool story, brah...
Let's add to that Deification of The Algorithm, all Hail The Algorithm, do not question The Algorithm, especially not when The Algorithm tells us that YOU must die.
Listen, I'm all in favor of helping folks, but let's not over-attribute qualities to some pattern matching toolkit!
(Score: 0) by Anonymous Coward on Monday November 23 2020, @09:13PM (1 child)
Save your rant for a stupider story. This is the kind of application is the best use case!
(Score: 0) by Anonymous Coward on Monday November 23 2020, @09:25PM
RANT IS RANT! https://imgflip.com/i/4nkxxe [imgflip.com]
(Score: 0) by Anonymous Coward on Monday November 23 2020, @09:33PM
I for one welcome our new algorithmic A.I. overlords, even if that means my life being classified as not mattering by the algorithm.
(Score: 0) by Anonymous Coward on Monday November 23 2020, @09:50PM
Languishig in the queue is this story, https://soylentnews.org/submit.pl?op=viewsub&subid=45705 [soylentnews.org]
[pasted in the story below, without the original formatting]
Technology Review describes results from 40 researchers across seven different teams at Google: https://www.technologyreview.com/2020/11/18/1012234/training-machine-learning-broken-real-world-heath-nlp-computer-vision/ [technologyreview.com] [technologyreview.com] or https://archive.is/xhEGp [archive.is] [archive.is]
It’s no secret that machine-learning models tuned and tweaked to near-perfect performance in the lab often fail in real settings. This is typically put down to a mismatch between the data the AI was trained and tested on and the data it encounters in the world, a problem known as data shift.
... identified another major cause for the common failure of machine-learning models. Called “underspecification,” it could be an even bigger problem than data shift. “We are asking more of machine-learning models than we are able to guarantee with our current approach,” says Alex D’Amour, who led the study.
Underspecification is a known issue in statistics, where observed effects can have many possible causes. D’Amour, who has a background in causal reasoning, wanted to know why his own machine-learning models often failed in practice. He wondered if underspecification might be the problem here too. D’Amour soon realized that many of his colleagues were noticing the same problem in their own models. “It’s actually a phenomenon that happens all over the place,” he says.
D’Amour’s initial investigation snowballed and dozens of Google researchers ended up looking at a range of different AI applications, from image recognition to natural language processing (NLP) to disease prediction. They found that underspecification was to blame for poor performance in all of them. The problem lies in the way that machine-learning models are trained and tested, and there’s no easy fix.
The paper is a “wrecking ball,” says Brandon Rohrer, a machine-learning engineer at iRobot, who previously worked at Facebook and Microsoft and was not involved in the work.
The article has more detail and examples. Full paper (PDF warning) is available here, https://arxiv.org/pdf/2011.03395.pdf [arxiv.org] [arxiv.org]
While the article focuses on image recognition (primarily medical) and speech, your AC submitter suspects that this is going to push off the timing for self driving cars by a significant amount of time. The article suggests that 50 times the training might be required compared to what is being used now.
(Score: 2) by krishnoid on Monday November 23 2020, @08:59PM
I'll post this again [mit.edu]; it's worth a shot to see if it can be trained to differentially select [reddit.com] audio signatures in respiratory ailments. After all, it's just looking at audio waveforms along with other entered data; who knows what it could correlate between slight changes to the auditory response in the lungs and upper respiratory tract, perhaps based on an audio profile of some coughs, deep breaths, and maybe a few sung notes.
Submit a sample yourself [mit.edu] and tell your friends to tell everyone else to try it too.
(Score: 0) by Anonymous Coward on Monday November 23 2020, @10:27PM (3 children)
If US elections couldn't bring an end to the man with a quarter million deaths on his watch, I doubt those algorithms will stop him.
(Score: 3, Insightful) by barbara hudson on Tuesday November 24 2020, @01:53AM (1 child)
1. If it weren't for covid, Trump would have way a second term. Now that the transition is officially beginning, there's at least one silver lining.
2. Related to my first point, the headline is wrong. The #1 killer is stupidity. Call me back when there's an app to detect that and we'll talk.
SoylentNews is social media. Says so right in the slogan. Soylentnews is people, not tech.
(Score: 2) by Muad'Dave on Tuesday November 24 2020, @02:19PM
There are several, actually. Two of the most popular stupid-detecting apps are called 'Facebook' and 'TikTok'. The algorithm is simple - if you use the app, you're likely stupid.
(Score: 0) by Anonymous Coward on Tuesday November 24 2020, @01:32PM
A quarter-million is nothing compared to recent ghouls. W was responsible for hundreds of thousands and probably a million deaths as a result of his crusade and how many people do you think Obama killed with his refusal to get even a public option into Obamacare? Trump has been terrible, but let's be honest about the fact that it's been ages since we've had a President that was at all acceptable.
(Score: 0) by Anonymous Coward on Monday November 23 2020, @10:56PM
"Except COVID..."
"What?"
"You mean the world's deadliest killer except for COVID"
"Naturally except for COVID."
Humperdinck: "Ah, my dulcet darling! Tonight, we marry. Tomorrow morning your men will escort us to Florin Channel, where every ship in my armada waits to accompany us on our honeymoon."
Buttercup: "Every ship but your four fastest, you mean. Every ship but the four you sent."
Humperdink: "Yes. Yes of course. Naturally not those four."
(Score: 2) by deimtee on Tuesday November 24 2020, @02:32AM
Is anyone aware of any studies on the effects on other diseases of the COVID lockdowns and precautions?
It seems like most other communicable diseases should be getting knocked on their arse too, but nobody talks about beneficial side-effects of the lockdowns.
One job constant is that good employers have low turnover, so opportunities to join good employers are relatively rare.