This idea that artificial intelligence can predict and diagnose for conditions and signs of imminent mortality has evolved a lot in the last couple of years.
Even dumb AI is doing things now in terms of survival rate probability and early medical diagnosis that is more accurate than the existing tools we have today, but researchers don't understand fully how the AI arrives at such accurate predictions.
Even dumb AI is doing things now in terms of survival rate probability and early medical diagnosis that is more accurate than the existing tools we have today, but researchers don't understand fully how the AI arrives at such accurate predictions.
Brandon Fornwalt at healthcare provider Geisinger in Pennsylvania, US and colleagues tasked an AI with examining 1.77 million electrocardiograms (ECG) results from nearly 400,000 people to predict who was at a higher risk of dying within the next year. I first read this in the New Scientist.
As you probably know, an ECG records the electrical activity of the heart. Its pattern changes in cardiac conditions including heart attacks and atrial fibrillation. So the team They measured the AI’s performance using a metric known as AUC, which measures how well a model distinguishes between two groups of people – in this case, patients who died within a year and those who survived.
The Probability of Everything Scenario is Coming
The AI consistently scored above 0.85, where a perfect score is 1 and a score of 0.5 indicates no distinction between the two groups. That's really high.
However, the AI also accurately predicted death in people that doctors believed were healthy based on "normal" ECG results. The AI seemed somehow to be able to see risk factors that traditional methods miss. That is, the model even detected heart problems in patients who were previously cleared by cardiologists. How is that even possible?
In the 2020s we are entering an age where AI will increasingly be implicated in early diagnosis and is being trained in top universities and by Google to do this better and better. AI models have also been used to diagnose heart disease and lung cancer — in some instances more accurately than human doctors.
“That finding suggests that the model is seeing things that humans probably can’t see, or at least that we just ignore and think are normal,” said Fornwalt.
The AI is able to crunch the numbers in a way that is not explainable in easy terms. The Geisinger’s AI and many other models as well have experts struggling to explain how the AI works (MIT). Since no one really knows how the most advanced algorithms do what they do, that is making many feel a bit uneasy.
We Cannot Explain How AI is Superior to our Detection Methods
Our AI systems are regularly becoming so complicated that even the engineers who designed it may struggle to isolate the reason for any single action. The AI is doing "spooky action at a distance" in ways that are already unexplainable to human intelligence, while simultaneously outperforming human models currently in use.
Last year researchers at Google in Mountain View, California created a predictive model using electronic health records to predict a patient’s length of stay and time of discharge, as well as the time of death.
Hospital officials say that strict controls will limit Google’s access to Mayo’s data. However, with the Project Nightingale controversy, it appears Google might be probed by federal regulators about a potential HIPAA violation.
HIPAA is the federal Health Insurance Portability and Accountability Act of 1996, which established national standards to protect individuals' personal health information. It's not clear to what lengths Google and others will go to super-charge their AI in healthcare, but it is a $3.5 Trillion industry so their incentive to cross some boundaries is very high. Few companies have the talent of a DeepMind in their back pocket as Google does.
AI now no longer just controls the spread of information, data harvesting, and ad-targeting, but will soon impact our health and finances. AI implementations still feel like the wild wild west of profitability models where AI could significantly reduce healthcare costs by assisting physicians and optimizing patient journeys.
Does Google's Project Nightingale Infringe upon HIPAA?
With Project Nightingale, it appears (according to the WSJ) that neither patients nor doctors were notified, and at least 150 Google employees have access to the data, which includes lab results, diagnoses, and hospital records and provides detailed information on people's health histories.
As many Millennials will likely have reduced life expectancies as compared to more wealthy Boomers and GenXers, this ability to know who is at risk to die will become far more lucrative. If loneliness and poverty can kill, AI will be able to reduce to predictive analytics EMR records reduced to coefficients and probabilities. AI playing God in our lives will go from science fiction to reality within our lifetimes.
How can you go against companies that are now inventing the future of healthcare and finance, and later, education? Google is using the patient data to tune artificial-intelligence software that may help improve patient care to reduce the financial burden of skyrocketing costs of healthcare that could cripple a generation. It's important to work but the incentives are so high, the ethics aren't even framed in an appropriate way for the kind of society we are building.
What's in an ECG or an AppleWatch, really? The algorithm performed better than the traditional measures used by cardiologists. Geisinger points to a continual movement of AI being used to diagnose earlier. This has implications for the future treatment of heart disease across the board.
Why AI Hype is Dangerous to the Future We are Building
This very idea that AI has "limitless potential" to benefit humanity is also dangerous. The highest official regulator in all of Europe even seems to be an AI enthusiast. What does that say about humanity's opt-in to AI consuming some of the most important life and death questions of our existence? Margrethe Vestager, the European Commissioner for Competition, offered a highly optimistic assessment of AI's potential impact on society, saying she sees "no limit to how AI can support what we do as humans".
Do they not understand the kind of society AI will enforce? When we have to rely on Europe to regulate BigTech we clearly have a problem, as China will begin to out-innovate the United States in the decades ahead, launching a world state where AI surveillance capitalism is the dominant paradigm. Congress can't afford to regulate Google's profiteering here because it's a question of national security in their strategic rivalry with China.
AI won't just know when you will die but will know you better than you know yourself. Google and Facebook already see patterns in your life that your best friend or Mother would miss. AI is giving white-collar professionals new tools to do their jobs better, as our cities become embedded with machine intelligence more and more each decade.
Google's DeepMind has built a team to tackle some of healthcare’s most complex problems, developing AI research and tools that could disrupt the future of healthcare as we know it. On the research side, we’ve seen major advances with Moorfields Eye Hospital NHS Foundation Trust in detecting eye disease from scans as accurately as experts, predicting patient deterioration as much as 48 hours in advance, with so many other breakthroughs in the pipeline.
Death, Taxes and AI Surveillance
Death, taxes and AI. Welcome to the future of humanity.
An artificial intelligence system developed at the John Radcliffe Hospital diagnoses heart scans much more accurately than doctors can. We know that those suffering from chronic loneliness are at greater risk for early mortality. How soon before they can model my mental health and know how many years I have left? For aging seniors, what interventions might we make to help those who are at risk for deaths of despair?
I know I will feel a bit more lonely when AI is given such a central place in our processes, interactions, judgments, and technological reductionism, just as the entry of Algos dehumanized so many of our institutions.
I worry that companies like Google are ushering in a dark age of technology where human life is trivialized in a "cost-efficient capitalism" that squeezes human value for ROI in ways we cannot even yet imagine. What does materialism + AI eventually lead to? A utopia where we live longer or a dystopia with more levels of control and surveillance? It's no longer hard to imagine the final answer to this.
0 टिप्पणियाँ