The cases are from the journal’s weekly series that aims to baffle physicians by presenting difficult, complex situations in which the diagnosis isn’t immediately apparent. Using roughly 300 of these cases, Microsoft compared the performance of its MAI-DxO to 21 general-practice physicians. Microsoft’s team first developed a mechanism to simulate the iterative process that physicians usually follow when handling these cases: gathering data, evaluating it, ordering tests, and making decisions based on the findings.

“A collection of commercial AI models, including Claude, DeepSeek, Gemini, GPT, Grok, and Llama, were compared to the 21 physicians. The Microsoft team also created an Orchestrator, a virtual representation of the sounding board of colleagues and consultations that doctors frequently seek out in complex cases, to further mimic how human doctors handle such difficult cases,” the study stated.

Microsoft monitored the tests that the AI system and human doctors ordered to determine which approach could complete the work more affordably, as ordering medical tests in the real world is expensive. In addition to performing significantly better than physicians in determining the right diagnosis, MAI-DxO was able to do so at an average cost that was 20% lower.

“The four-fold increase in accuracy was more than previous studies have shown. Most of the time, there is a 10% absolute percentage difference, so this is a really big jump. But what really got his attention was the cost. Not only was the AI more accurate, but it was much less expensive,” Dr. Eric Topol, chair of translational medicine and director and founder of the Scripps Research Translational Institute, told Times Magazine.

Since MAI-DxO is still in development, it is not yet usable for purposes other than research. However, implementing such a model could potentially improve patient outcomes by reducing medical errors, which contribute significantly to healthcare costs, and by increasing the effectiveness of human physicians.

“This is a startling result. I think it gives us a clear line of sight to making the very best expert diagnostics available to everybody in the world at an unbelievably affordable price point. We are nearing AI models that are not just a little bit better, but dramatically better, than human performance: faster, cheaper and four times more accurate,” said Mustafa Suleyman, CEO of Microsoft AI, during an interview with the Financial Times, while describing the trial as a step toward “medical superintelligence.”

According to Suleyman, when AI algorithms were first used in medicine ten years ago, they were primarily used for binary tasks like tumour detection in image scanning. Microsoft’s research even suggests AI diagnostic tools could reduce unnecessary healthcare expenditures whilst improving accuracy – as the United States’ health spending is approaching 20% of GDP, with an estimated 25% providing minimal impact on patient outcomes.

“Important challenges remain before Gen AI can be safely and responsibly deployed across healthcare. We need evidence drawn from real clinical environments, alongside appropriate governance and regulatory frameworks to ensure reliability, safety and efficacy,” said Microsoft’s research team.

He claims that these models are currently having very high-quality, fluid conversations in which they ask the appropriate questions, probe in the right ways, and recommend the appropriate testing and interventions at the appropriate times. An AI system may also benefit from not having many of the biases that come with being human.

“Across Microsoft’s AI consumer products like Bing and Copilot, we see over 50 million health-related sessions every day. From a first-time knee-pain query to a late-night search for an urgent-care clinic, search engines and AI companions are quickly becoming the new front line in healthcare,” the company said in a blog post.

“As demand for healthcare continues to grow, costs are rising at an unsustainable pace, and billions of people face multiple barriers to better health. We want to do more to help and believe generative AI can be transformational. That’s why, at the 2024 end, we launched a dedicated consumer health effort at Microsoft AI, led by clinicians, designers, engineers, and AI scientists,” the tech giant stated further. According to Dominic King, vice president of Microsoft AI, “Confirmation bias affects everyone. Clinicians occasionally think, ‘I’m sure this is just like the patient I saw recently,’ after observing something. However, AI has a slightly different way of thinking.”

MAI-DxO is not just a simple spit-out system. It does its work in such a way that doctors may be able to study and examine its reasoning. However, some medical and AI experts point out that Microsoft’s method isn’t wholly original because its diagnoses relied on the combined output of several AI models.

“In my mind, they are not testing any individual model that is optimised for healthcare. They are testing the concept of testing all of the models out there today and combining their decision-making. That part, to me, is not surprising,” Keith Dreyer, chief data science officer at Massachusetts General Hospital and Brigham and Women’s Hospital Centre for Clinical Data Science, said.

Additionally, Dreyer notes that the findings do not necessarily mean that regulatory bodies like the US Food and Drug Administration, which has yet to make a determination on whether or not such systems qualify as medical devices, will approve them.

Microsoft is not alone in its pursuit of an AI-powered medical diagnostic programme. Google is creating a dialogue-based system to simulate the back-and-forth between a doctor and patient, replicating how real doctors gather data from patients and analyse those symptoms to arrive at a diagnosis. In preliminary testing, the system performed better than physicians in correctly diagnosing case studies of simulated patients. The previous iteration of Google’s system correctly identified 59% of cases in a 2024 test that was comparable to the one Microsoft conducted using case studies, while human doctors only did so in 33% of cases.

The true test will be how well these AI systems function in real-world healthcare settings. Understanding how AI could enhance or support a physician’s role in disease diagnosis is the next step. Topol remarks, “What they accomplished is impressive.” However, until they are implemented in actual medical environments, it won’t alter medical practice. Topol hopes the AI systems will be tested in various health systems so that physicians and the AI platform can be compared on a variety of more common and diverse cases.

A comprehensive clinical trial and regulatory agency approval are necessary to ensure that patients won’t suffer any harm from a greater reliance on AI-based decision-making in the provision of care. Dominic King states, “We are very much on that journey to create the evidence base required to support doctors and patients to make a difference in their health.”

Microsoft’s MAI-DxO marks a significant milestone in the integration of AI into clinical diagnostics, showcasing higher accuracy and lower costs than human doctors in complex cases. While still in development, the system’s potential to enhance medical decision-making, reduce diagnostic errors, and ease healthcare costs is compelling. However, the road to real-world adoption remains cautious. Experts stress the need for extensive clinical trials, transparency, and regulatory approval to ensure safety and reliability. Though not a replacement for physicians, AI systems like MAI-DxO could become powerful collaborators in modern medicine, offering second opinions and data-driven insights.

As tech giants like Microsoft and Google race to refine these tools, the healthcare industry stands on the brink of transformation. This shift must be approached with scientific rigour, ethical foresight, and a strong commitment to patient safety. Microsoft has already built other healthcare AI tools like RAD-DINO (for radiology workflows) and Dragon Copilot (a voice assistant for clinicians). It will now be partnering with hospitals, clinicians, and health organisations to further validate the technology in real-world settings. That’s where the real test will start for the tech giant.

The post Is Microsoft’s AI better than your doctor? appeared first on International Finance.

Cases included false billing of the federal medicare insurance programme for elderly and disabled Americans and paying illegal kickbacks, illicit diversion of expensive prescription medications and the improper dispensing of highly addictive opioid painkillers.

“Among those facing charges include 24 doctors, nurses and other licensed medical professionals, as well as healthcare executives including the current and former CEOs of a durable medical equipment online platform accused of falsely billing $1.9 billion in fraudulent claims,” Reuters stated.

The rotten state of affairs

As per the Justice Department, out of the $2.5 billion in fraudulent medicare claims, state Medicaid programmes serving the poor and private medicare insurance programmes, about $1.1 billion was actually paid out to the fraudsters.

Some cases were related to expensive HIV medications, which can fetch medicare reimbursement rates as high as $10,000 for a month’s supply.

In one case, the owner of a pharmaceutical wholesale distribution company was charged in New Jersey with illegally purchasing diverted HIV drugs and then reselling the medication by falsely claiming it was acquired through legitimate channels. In another incident, federal officials announced an indictment against a Wisconsin-based business owner accused of preying on low-income pregnant women by enticing them to sign up for prenatal care services, and submitting phoney claims for services never rendered.

Then there were fraud cases targeting elderly/disabled patients. These victims were tricked into providing their insurance information to telemarketers in lieu of receiving testing, medical equipment or other “free service” paid for by medicare. Doctors having no relationship with the patients were then rubber-stamping the orders by falsely certifying that they were medically necessary. These claims were getting submitted to federal/state insurance programmes for reimbursement.

Persons contributing to the scheme were receiving illegal kickbacks. The Justice Department also cited examples of services like durable medical equipment, genetic testing and lab diagnostic services, which were targeted by the fraudsters.

In January 2024, reports emerged about YouTube deleting a thousand videos circulating on its platform that purportedly showed singer Taylor Swift, television personality Steve Harvey and podcaster Joe Rogan pitching a ‘Medicare Scam’.

These videos were artificial intelligence-generated deepfakes. YouTube told Newsweek about the video streaming platform being “aware” that it was containing such fake advertisements and deleted over 1,000 videos, 90 YouTube channels and multiple advertiser accounts linked to the scams.

In 2023, The United States National Health Care Anti-Fraud Association estimated that tens of billions of dollars per year were lost to healthcare fraud. Many of these losses were attributed to the fact that medicare is required to pay medical claims quickly. As a result, claims were getting paid long before they could be flagged for potential fraud, which meant that these monetary losses were evading the government’s radar.

Some of these scams also preyed upon the disabled and elderly, making them unwitting victims of large criminal schemes. One such was the Durable Medical Equipment (DME) companies paying illegal kickbacks and bribes in exchange for referrals of medicare beneficiaries. Authorities described this scam, which involved telemedicine, as one of the largest healthcare frauds they had ever investigated. The scam targeted hundreds of thousands of elderly/disabled patients and involved medical professionals and call centres in other countries in pulling off the heinous act.

The scammers didn’t spare the COVID outbreak either. Medicare advocates noticed an eleventh-hour rise in complaints from beneficiaries who received tests, sometimes by the dozen that they never requested. For these experts, the trend was a signal that someone might have been using, and could continue to use, seniors’ medicare information to improperly bill the Joe Biden government.

In 2023, the United States Department of Health and Human Services’ Office of Inspector General received complaints about unsolicited tests being billed to medicare, as per the reports.

As per María Alvarez, who oversees New York State’s Senior Medicare Patrol, an organisation which helps identify and educate beneficiaries about medicare fraud, a stolen medicare number can be used repeatedly to get payment for all kinds of things.

According to Nancy Moore, the Senior Medicare Patrol programme director for Indiana, one beneficiary suspected something was amiss after receiving 32 unrequested tests over a 10-day period. In fact, complainants didn’t give out their medicare numbers to random people. Lisa Dalga, project manager for Ohio’s Senior Medicare Patrol, stated in 2023 about medicare being paid for tests for some beneficiaries who never received them.

Scams in Catheter Bills as well

Recently, the New York Times reported about a person named Linda Hennis, who, while checking her medicare statement in January 2024, saw a company, which she had never heard of, had been paid about $12,000 for sending her 2,000 urinary catheters. The fact is that Hennis never needed/received any catheters.

Chicago-based Hennis was among over 450,000 medicare beneficiaries whose accounts were billed for urinary catheters in 2023, according to the “National Association of Accountable Care Organisations,” an advocacy group that represents hundreds of healthcare systems across the country.

“The massive uptick in billing for catheters included $2 billion charged by seven high-volume suppliers, potentially accounting for nearly one-fifth of all medicare spending on medical supplies in 2023. Doctors, state insurance departments and health care groups around the country said the spike in claims for catheters that were never delivered suggested a far-reaching medicare scam,” The New York Times noted.

“Pretty in Pink Boutique”, the organisation under the lens here, billed medicare at least $267 million for catheters between October 2022 and December 2023.

Talking about medicare billing scams, if patients do not pay the bills themselves, more spending by the government can increase the premiums paid by enrollees in the future.

In 2019, an international fraud ring involving over $1 billion in phoney billing for back and knee braces got busted. The trend is clear here-medical supplies have become the sources of scams, thanks to relatively low-quality bars that result in the mushrooming of medical supply companies.

“The companies don’t need much to show why grandma needs a urinary catheter,” said Eva Gunasekera, who previously led healthcare fraud investigations at the Department of Justice.

A scary future indeed

Patients and doctors who have been reporting mysterious catheter claims to medicare for months expressed their frustration, as there was a lack of communication from the Biden government about whether billions of dollars have been lost to an ongoing billing scam.

Dr. Bob Rauner told The New York Times that his patients got collectively billed nearly $2 million in 2023 for “Phantom Catheters”.

He had to file a complaint with the federal health department’s Office of Inspector General in December 2023.

“Pretty in Pink Boutique is registered with medicare to a street address of a house in El Paso. Its phone number goes to an auto body shop called West Texas Body and Paint, where an employee who answered a call from a reporter said the shop receives ‘calls all day, every day’ from medicare enrollees concerned about fraudulent bills,” NYT stated further.

Pamela Ludwig, who runs an unrelated business in Nashville that is also called “Pretty in Pink Boutique”, got so many catheter complaints (as the medical supply company was sharing the same brand name) that she had to give an online clarification that her business was not part of any scam.

She complained to medicare in September 2023. After that, her husband heard from a New York City banker about several men coming to his office and asking to set up an account for “Pretty in Pink Boutique”.

The issue landed on the radar of the Oklahoma Insurance Department in July 2023, when it was investigating fraudulent medicare claims for COVID-19 kits. The officials then noticed a surprisingly high number of claims for catheters as well. The massive spike in billing for at-home catheters cost medicare as much as $2 billion.

The healthcare structure in the United States is in dire straits, with the scamsters having field days at the cost of the Americans’ well-being. Will the government ramp up its regulatory watch? Because being extra vigilant will help things more, rather than launching the regulation investigation after the occurrence of the crimes.

The post US healthcare under siege? appeared first on International Finance.

Steven Riznyk

March 20, 2017: The H1b visa has made it to the news through the recent halt to Premium Processing. The impact has created an opportunity for analysis that will hopefully create some positive changes for the H1b visa (usually referred to as ‘H1”, but not to be confused with the H1A).

Although it is easy to make broad brush strokes that change the system, the time has come for a multi-dimensional analysis of the H1 visa so that the appropriate changes can be made. The first thing I would do is carve the H1 visa up into different groups, each with its own set of narrowly defined regulations, or alternatively, split it up into a number of visas.

Hypothetically, let’s say either Microsoft or Apple (hereinafter ‘company’) are working on a high-level project that competes with the other operating system and require someone with a certain rare skillset. If that person is not available in the United States, they will have to open the demographic search to the entire planet to find the right person; that will require an H1 visa application. As projects at this level require collaboration amongst a group of people, it cannot be outsourced, there are no choices.

Challenges of this nature are painful for a business. First, if they apply for the H1 visa, they have to wonder whether the person will be one of three who wins the H1 lottery. Second, even if that applicant prevails, the company will have to wait until October 1st for that skillset. Third, even if the person prevails on the H1 visa, and starts on October 1st, that person requires training and has to be brought up to speed. Fourth but not last, after all that, the person may not be the right choice for a number of reasons.

In the interim, while waiting until October 1st, what is the company to do with the project? Does our tech sector wait for rivals in other countries to beat them to the punch? Does one company prevail over the other because they hired the last person in the US who can do what it is they require?

If one company loses over the other due to the unavailability of key people, their stock will surely suffer. The people who have money in their retirement accounts will be affected. If it is severe enough, people may be laid off. Houses go up for sale, and then the supply & demand curve kicks in; the house prices in that city will spiral downward, and so on. These are just a few aspects of the domino effect that occurs.

On a smaller level, but a more direct and severe one, smaller and mid-size firms suffer the impact faster.  An example is someone who called his office. The person is Italian and a brilliant architect who outfits high-end homes with marble floors, tables and walls that are hand-designed with inlays that are works of art. He found a position with a company that was launching in two months’ time. The owner could not wait until October 1st and took a pass, however very regretfully. The owner is now relegated to hiring someone for her Italian-styled interior design firm who is not from Italy, or who is, but lacks the architectural and artistic skills of the applicant. She will settle for second best because that is the only choice she has: the left-overs.

The point is that the H1 visa, for business purposes, needs to be effective in getting someone to the United States rapidly. Businesses cannot wait 6-17 months for key employees; they need them yesterday.

From this point of view, we could have a division of the H1 visa called H1RN (for clarity’s sake only) for Right Now. These people could apply with Premium Processing and get in immediately (ie in under 2 months’ time). This would have to be a skills-based visa, wherein special circumstances could be pleaded as one of the base requirements.

This could work for Microsoft or Apple, as well as for the interior designer. The CIS could ask a premium for the filing fees for this special consideration, and it would compensate for the extra 15 minutes it would take to determine the viability of this type of application. They could also charge another filing fee that would transfer it to another, less urgent category should it not be approved so that no time is wasted in reapplying.

Another category is for doctors who have a 212(e) 2-year foreign residency requirement as a result of the J1 visa. One way they can avoid the foreign residency requirement is by working in underserved areas. This would be great for America as many parts of the country desperately require high-end doctors who frankly would find those areas too remote or even un-exciting to live in. Whoever thought of this was brilliant.

The problem again, is that these doctors are competing with techs that are being outsourced for the very same slots. I would call this the H1MD for Medical Doctor. This category in itself would have to be spilt into two sub-categories to cover the J1 persons and all other doctors and dentists. In the interests of being thorough, this category should perhaps have a third category covering personnel in the medical field who are not doctors or dentists. In this manner, the sub-category would be comprehensive when covering medical personnel of all types.

A third category would be the H1TB for Technology-Based and a fourth would be the H1OP for Outsourced Personnel. The technology-based visa would be the visa that tech companies would use when they require people with tech skills and could either not qualify is H1RNs (Right Nows), or they had lesser skills. The H1OP would be for all of the consulting firms who are bringing people into the country and then outsourcing them.

This is an area of much controversy. The H1b grants a visa to these consultants who often are programmers with no skills that are absent in this country and the competition they bring to the application process keeps high-level doctors and scientists out.

With those major categories taken care of, the H1 visa is still responsible for many more potential jobs. The public, for the most part, thinks of it as a visa for tech-related positions, but the positions it is also responsible for includes, but certainly is not limited to: aeronautical engineers, as well as many other fields of engineering, mathematics, astronomy, physics, pharmacy, university and college education, writing and editing, accounting, management, economics, museum curation, music, and yes, law.

Our country can prosper by getting the best and brightest people here, but not as the system stands; it is too slow and too disruptive on their (potential workers’) lives. These other categories need to be grouped on their own, but definitely separated from the ones numerated above; even then, it is still too broad a category, but much better than what we have now.

The great advantage that we will have if we break down the H1b into separate categories is that we will be able to promulgate regulations for each of the categories with their own mandates, priorities, and requirements. In many ways, we can call these growing pains, which are a good thing. They mean America has grown, and therefore outgrown its older broader categories. We are standing on a more highly specialised work platform that has to cater to the needs of today’s business world, not the world in 1990 when the category was created. We can no longer group all of the job positions into one visa, nor can share the same regulations across the board.

Once this is resolved, our country will be in a great position to continue as a world leader, with the best and the brightest at the helm.

Steven Riznyk is CEO of San Diego Biz Law, a business & immigration law firm

RELATED ARTICLES

A tech visa is way overdue

Married to US citizens but overstaying visa is not good

Immigration problems cannot be repaired by one entity

The post Time for a multi-dimensional analysis of the H1 visa appeared first on International Finance.