AI trained to “listen” for disease – will this create a new surveillance system to penalize people if the system “finds” something in their coughs, sniffles, sneezes, or breath?

Google and others are already screening people in India

Google is training AI to use “sound signals to predict early signs of disease, opening up a world of possibilities,” Bloomberg reported on August 29, 2024.

The new aim is to train AI so it can hear when someone is sick. Ostensibly, AI can learn to identify how body sounds are linked to diseases, and this will facilitate the early detection of diseases - even those without visible symptoms.

Source of the video: Google Research

Bloomber further promises that

“Body sounds are filled with information about our well being, containing near-imperceptible clues that can help screen, diagnose and manage health conditions.”

Ok, the premise is that AI will be able to make sense of those near-imperceptible clues. The idea had been around for some time but it was simply too costly and difficult to process the huge amount of information. Google and others claim they now have made major improvements to cut through all the data via “self-supervised” machine learning models. Thus, the new promise: what previously was invisible can now be heard -  via the AI analysis.

Yet, how on earth can “body sounds” help “manage” a disease? At best, one may envision that changes in sound patterns could indicate if a certain intervention works or not.

However, it’s not the body sounds themselves that can do this. But they may be captured for something else, even beyond the promotion of certain drugs and therapies. In the context of global censorship and Big-Industries trying to fight “misinformation” like hell, it is not difficult to envision how this technology could be gravely misused.

Is the plan to “screen” and “diagnose” so that “high-risk” individuals (e.g. those who oppose an approved narrative and refuse certain medical “interventions”) can be “managed?” I have no proof of this. But neither is there any proof that various infectious diseases, and others, can accurately be diagnosed by AI, even from laboratory tests and clinical samples. Rather, you end up with a computer output that could be rather arbitrary. Extending the “diagnosis” to the sound of the human voice, their cough, breath, or related, is rooted in even more convoluted computer models and AI interpretation. AI can tell you whichever it wants!

“The cough sound is the equivalent of giving a blood sample, only this particular sonic sample is processed on the cloud rather than in a laboratory.”

This raises a huge issue. It’s processed in the cloud. The decision, aka, “diagnosis,” is done in the absence of physical samples, based on digital data, and theoretical modeling only. This type of abstraction is in itself worrisome because there is no strong link to a physical or biological reality. In an ideal world, we would have tons of independent analyzes of this.

Since AI is involved, thanks to its inherent black-box features, it is even more opaque, After all, AI is employed because of the “near-imperceptible clues” that can no longer be verified by humans. Good luck with unbiased testing, analysis, and oversight! For example, Hyderabad-based Salcit Technologies, a collaborator of some Indian tech giants, boasted that their AI model that analyzes cough samples tests diseases with 94% accuracy.

How has this been validated? For which diseases? Is it all open to public scrutiny or the company’s own estimates? As we know from the COVID-19 pandemic era, drug efficacy data from BigPharma’s own trials get too readily broadcasted by MSN, parroted by high-ranked public health officials as facts, and taken by policymakers as gospel truth. 

Now, granted, the Boomerang article does admit some “challenges.”

“But there are challenges. While the new tech is exciting doctors in the field by opening up a new frontier, it’s not easy to change routine clinical practices. The screening tool will need to find acceptance.”

So, the challenge is that independent doctors may not like it when AI comes up with a diagnosis that they do not support. The “challenge” is to get them to do what AI tells them to do!!!!!

Never mind that medical doctors have gone through extensive training, see the effects on real human beings, and have accumulated tons of insight knowledge that distinguishes them from any algorithm or machine! And never mind the value of possibly decades of experience that some doctors may bring to the table, or that of pure and strong instinct the youngest or poorest patient may have! Healthcare professionals better “accept” that some abstract data evaluated in the cloud somewhere, without any tangible explanation, and likely even in the absence of symptoms, ought to overrule everything they know!

As for the challenge of “background noise.” I find it shocking that the real problem is not mentioned. AI systems are biased and known to “hallucinate.” They frequently come up with unjustifiably flawed statements, and it is incredibly difficult to re-train them. Some of the machine learning systems even support their claims with convoluted and lengthy “arguments” when further queried. It takes someone with real knowledge to cut through the crap and recognize the flawed conclusions.

According to Bloomberg, another “challenge” is “the problem of ensuring audio samples don’t come with an abundance of background noise” and that “[r]ural users, unfamiliar with technology, may be unable to record coughs on the app.” But there seems to be a “solution,” as specified in the very next sentence: “Yet, the tech is finding supporters, including those like the StopTB Partnership, a UN-backed organization, which aims to end TB by 2030.”

Some may wonder what the roll-out of these technologies will cost, compared to how much it would be to provide funding for clean water and healthy food to those in those high-risk geographies that the technology purports to save.

Amidst the global push to vaccinate and medicate every living thing, the following paragraph makes my heart sink.

“Montreal-based Ubenwa has built a foundation model for infant cries, and interprets infant’s needs and health by analyzing the biomarkers in their cry sounds.”

I can see the enormous new business opportunity! If those poor babies make any strange sound, then there will probably be an expensive drug to solve the non-verifiable (and possibly non-existing) problem!

“And others are working on AI tools that can detect autism based on oohs, aahs and gurgling sounds.”

And what will be the result of such “early detection?” Classifying children as autistic, even if they are not?! Or diagnosing adults with various mental problems? The more I think about this, the more it is clear there are enormous ways in which all this could be misused.

The technology could screen our voices, coughs, or even breathing for “harmful patterns” and, in turn, cause the population deemed at high “risk” to be quarantined, detained, and cut off from society. 

The technology can “ride in a smartphone” and be deployed in “tricky geographics.” Importantly, body sounds can also be analyzed remotely and at scale!

What today is in the name of “early disease detection” could tomorrow be used to identify individuals who for their political, public-health-related, or other views, be deemed dangerous or unacceptable. The limits seem endless. As soon as someone claims that AI “bioacoustics” can use our voice and other body signals to find who is infected with a dangerous pathogen, who is not adequately “immunized,” who is “lying,” “spreading hate speech,” or disseminating other forms of mis-, dis-, or malinformation, such a “diagnosis” could be cheaply made, processed on the cloud, and be irrevocable by real-world evidence. 

If it was really about individual and public health, there would be much better ways to come up with a tangible solution: clean up the governments and agencies captured by the industries they're supposed to regulate.