Category: 3. Business

(Bloomberg) — Check a ranking of the best-performing equity indexes this year and the US doesn’t crack the Top 10. You won’t find it in the Top 25, either. Double that, and the S&P 500 is still absent.

The tally needs to unfurl all the way to 66 before the world’s most valuable equity index shows up — leaving it way behind Greece’s Athex and even Israel’s TA-35. It’s one of the worst relative performances since the global financial crisis for the US benchmark.

Most Read from Bloomberg

The underperformance is even more surprising given the S&P 500’s  rally to countless records in 2025. But it’s still trailing most developed market benchmarks like Germany’s DAX and Japan’s Nikkei 225, and lags behind gauges in South Korea, Spain and Ghana, when measured in dollars.

That last qualifier is critical, though not determinant. The US currency has fallen  this year, helping to boost returns on foreign bourses in dollar terms. That’s certainly the thrust behind gains of at least 39% in Colombia and Morocco.

But even in local-currency rankings, the S&P 500 comes in just 57th, hardly befitting of a measure home to the six most valuable companies in the world, along with the likes of Coca-Cola Co., McDonald’s Corp. and Walt Disney Co.

The underperformance, market participants say, owes just as much to a broader shift in the mindset among foreign investors, who have started targeting domestic champions as President Donald Trump wages a global trade war. Tensions ramped up on Friday after the president renewed threats of tariffs on China. Even in the US, they’re being more selective, with a focus on big tech rather than broad-based indexes.

Added to that is a growing sense of concern about political and fiscal stability in the world’s largest economy. Trump’s tax and spending bill is projected to blow out the deficit. The government has been shut down since the start of October, the president is increasingly threatening the central bank’s independence and public investment decisions have become less policy-based.

Together, the moves have shaken confidence in America, weakened the dollar and helped stoke a torrid rally in gold. While long-term Treasury yields haven’t exploded in any similar fashion, they’ve been elevated relative to recent years.

“The deteriorating US fiscal situation and increasing policy uncertainty are eroding investor confidence in the US market, weakening the dollar, and prompting investors to explore opportunities in non-US markets,” said Jasmine Duan, senior investment strategist at RBC Wealth Management Asia.

Of course, strategists have for years been predicting an imminent rotation away from US equities and those calls have fallen flat. The dollar’s slide has eased in recent weeks as political stresses mount around the world, from France to Japan to Argentina.

And while the S&P 500 is lagging well behind the top three — Ghana, Zambia and Greece with gains of at least 61% — its  rally this year has created about $6 trillion in market value, equivalent to more than a third of the entire capitalization of the Stoxx 600.

The US is also coming off of back-to-back years with gains north of 20%, easily outstripping the likes of the Euro Stoxx 50 and Nikkei 225. If you take stock of performances since the end of 2022 to 2024, the S&P 500 ranked 10th.

Lasting Outperformance

Still, there are evident reasons that global equity markets may continue to outperform. European interest rates are half the level in the US, giving corporates access to cheaper financing. Companies trade at valuations about 35% lower than in America.

And so in Germany, Rheinmetall AG has more than tripled to lead the DAX to a  gain as the government promises to step up defense spending. European banks, long laggards, have been revitalized. In Spain, Banco Santander SA has almost doubled in value.

South Korea’s Kospi index has risen  this year as investors speculate the new president’s push for shareholder-friendly policies will boost returns. The nation’s standing as a sophisticated chipmaker has given it domestic champions in artificial intelligence, with Samsung Electronics Co. and SK Hynix Inc. rising after deals to supply chips to OpenAI.

“Asia has been a great platform to bring diversification in our portfolio, and to express our preference for looking for alpha within asset classes,” said Sophie Huynh, portfolio manager and strategist at BNP Paribas Asset Management.

Similarly in Japan, expectations for a pro-stimulus lawmaker to become the next prime minister have pushed stocks to all-time highs. SoftBank Group Corp.’s  surge has powered the Nikkei 225. Defense equipment makers Mitsubishi Heavy Industries Ltd. and Japan Steel Works Ltd. also rallied this month on optimism around more government spending.

Global money managers are returning to China after years of aversion, drawn by advances in high-tech industries. Alibaba Group Holding Ltd.’s plans to ramp up AI spending, and Huawei Technologies Co.’s aim to challenge Nvidia Corp. helped Chinese stocks log their best run of monthly gains since 2018. The Hang Seng Tech Index’s year-to-date advance of  is more than double that of the Nasdaq 100.

Too Expensive

The S&P 500’s stellar run from its April low has stretched valuations to levels that have raised alarm and prompted investors to diversify exposure. The index trades at 22 times forward earnings, a premium of 46% to the rest of the world. It’s also famously top-heavy, with mega-cap tech and its smaller brethren accounting for more than one-third of the index by weighting. A 53% rally in the two years starting at the end of 2022 had left foreign investors over-exposed to American equities.

“Investors should be rebalancing, taking profits from their US allocation and increasing exposure to Europe, Asia and emerging markets,” said Kristina Hooper, chief market strategist at Man Group, the world’s largest publicly traded hedge fund. “The US will continue to lag other markets.”

For now, buying from foreign investors remains on pace for a record, as fears of a recession recede. Their purchases make sense given the US is home to the key players in the AI frenzy, led by Nvidia.

But many are moving money, according to a Bank of America Corp. survey of fund managers. Global investors were a net 14% underweight US stocks in September, while being 15% overweight euro-zone peers and 27% overweight emerging markets. There’s also evidence foreigners are being more selective, and why not? Just six stocks account for over 50% of the S&P 500’s gain this year. In fact, a gauge that strips out market-cap biases is up just  this year.

“The last two years have only been about the US and nothing else because tech earnings were surging while everything else was down to flat,” said Beata Manthey, head of European and global equity strategy at Citigroup Inc. “This year, the growth differential between the AI trade and the rest of the world has narrowed, and it’s going to narrow even more next year. So there are more themes to choose from.”

Most Read from Bloomberg Businessweek

©2025 Bloomberg L.P.

A new era of technology is taking off. AI is reshaping economies, industries, and governance. And just like the last time, we’re moving fast, breaking things, and building the plane while flying it (to use some common tech phrases). These mantras have driven innovation, but we’re now living with the unintended consequences.

For over a decade, I worked in the engine room of the social media revolution, starting in U.S. government, then at Twitter and Meta. I led teams engaging with governments worldwide as they grappled with platforms they didn’t understand. At first, it was intoxicating. Technology moved faster than institutions could keep up. Then came the problems: misinformation, algorithmic bias, polarisation, political manipulation. By the time we tried to regulate it, it was too late. These platforms were too big, too embedded, too essential.

The lesson? If you wait until a technology is ubiquitous to think about safety, governance, and trust then you’ve already lost control. And yet we are on the verge of repeating the same mistakes with AI.

The new infrastructure of intelligence

For years, AI was viewed as a tech issue. Not anymore. It’s becoming the substrate for everything from energy to defence. The underlying models are getting better, deployment costs are dropping, and the stakes are rising.

The same mantras are back: build fast, launch early, scale aggressively, win the race. Only now we’re not disrupting media instead we’re reinventing society’s core infrastructure. 

AI isn’t just a product. It’s a public utility. It shapes how resources are allocated, how decisions are made, and how institutions function. The consequences of getting it wrong are exponentially greater than with social media.

Some risks look eerily familiar. Models trained on opaque data with no external oversight. Algorithms optimised for performance over safety. Closed systems making decisions we don’t fully understand. Global governance void whilst capital flows faster than regulation.

And once again, the dominant narrative is: “We’ll figure it out as we go.”

We need a new playbook

The social media era playbook of move fast, ask forgiveness, resist oversight won’t work for AI. We’ve seen what happens when platforms scale faster than the institutions meant to govern them.

This time, the stakes are higher. AI systems aren’t just mediating communication. They’re starting to influence reality from how energy is transferred to how infrastructure is allocated during crises. 

Energy as a case study

Energy is the best example of an industry where infrastructure is destiny. It’s complex, regulated, mission-critical, and global. It’s the sector that will either enable or limit the next phase of AI.

AI racks in data centres consume 10-50 times more power than traditional systems. Training a large model requires the same energy as 120 homes use annually. AI workloads are expected to drive a 2-3x increase in global data centre electricity demand by 2030.

Already, AI is being embedded in systems optimising grids, forecasting outages, and integrating renewables. But without the correct oversights, we could face scenarios where AI systems prioritise industrial customers over residential areas during peak demand. Or crises where AI makes thousands of rapid decisions during emergencies that leave entire regions without power and no one can explain why or override the system. This is not about choosing sides. It is about designing systems that work together, safely and transparently.

Don’t repeat the past

We’re still early. We have time to shape the systems that will govern this technology. But that window is closing. So, we must act differently. 

We must understand that incentive structures shape outcomes in invisible ways. If models prioritise efficiency without safeguards, we risk building systems that reinforce bias or push reliability to the edge until something breaks.

We must govern from the beginning, not the end. Regulation shouldn’t be a retroactive fix but a design principle. 

We must treat infrastructure as infrastructure. Energy, compute, and data centres must be built with long-term governance in mind, not short-term optimisation. 

We cannot rush critical systems without robust testing, red teaming and auditing. Once embedded at scale, it’s nearly impossible to reverse harmful design choices.

We must align public, private, and global actors, which can be achieved through truly cross-sector events like ADIPEC, a global energy platform that brings together governments, energy companies and technology innovators to debate and discuss the future of energy and AI.  

No company or country can solve this alone. We need shared standards and interoperable systems that can evolve over time. The social media revolution showed what happens when innovation outpaces institutions. With AI, we get to choose a different path. Yes, we’ll move fast. But let’s not break the systems we depend on. Because this time, we’re not just building networks. We’re building the next foundation of the modern world.

The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.

Fortune Global Forum returns Oct. 26–27, 2025 in Riyadh. CEOs and global leaders will gather for a dynamic, invitation-only event shaping the future of business. Apply for an invitation.

Motion 035: Protection of Mesopelagic Ecosystem Integrity goes to vote at the World Conservation Congress this October. It urges nations not to authorize commercial fishing or other harmful activities in the deep ocean until we understand it better. If passed, this motion would be a crucial step toward ensuring that life in the deep layers of the ocean continues to thrive, and human activity in this space proceeds only when proven to pose no harm.

—

By Silvie Alexander, Kristin Kleisner, Dr. Lance Morgan, Chris Dorsett and Kristina Gjerde.

Beneath the waves, between 200 and 1,000 meters deep, lies one of Earth’s most mysterious and vital ecosystems: the mesopelagic zone, also known as the Ocean Twilight Zone. Though shrouded in darkness, this vast layer spans the globe and harbors an estimated 90% of all fish biomass, making it the largest unexploited ecosystem on the planet.

Until recently, this deep sea zone remained largely unknown. But as scientific discovery begins shedding light on the mesopelagic’s immense ecological and climate value, a new threat is rising: industrial exploitation. 

Fishing fleets are eyeing this zone for extraction as demand for fishmeal and fish oil increases. Other potential activities like deep-sea mining and marine carbon dioxide removal technologies also threaten the integrity of the mesopelagic ecosystem and its services. 

If we are serious about fighting climate change and preserving ocean health, the world must act now to protect this fragile, extraordinary ecosystem before it is too late. An upcoming vote at the International Union for the Conservation of Nature (IUCN) Conservation Congress is the first step. 

Lungs of the Ocean

Though invisible from the surface, the mesopelagic zone isn’t lifeless. It is teeming with bioluminescent fish, jellyfish, and otherworldly creatures, many of which remain unnamed, unstudied, or entirely undiscovered. But what makes this region truly exceptional is the essential role it plays in regulating our climate.

Each night, billions of mesopelagic organisms perform the largest animal migration on Earth: migrating to the surface to feed on carbon-rich plankton, then returning to the depths by day. This nightly movement pulls massive amounts of carbon from surface waters to the deep ocean, where it can be sequestered for hundreds to thousands of years.

It is estimated that mesopelagic species help transport 2-6 gigatons of carbon every year, more than double the annual emissions from all the world’s cars. And that is a conservative range based on our limited knowledge of the region’s biomass. Some scientists estimate migrating mesopelagic organisms facilitate the sequestration of over half the carbon stored by the ocean’s biologic carbon pump, with others positing they are responsible for up to 90% of deep ocean carbon storage. 

As science deepens its understanding of the mesopelagic zone and its species, it is increasingly evident that it is one of the planet’s most powerful climate stabilizers. Without it, Earth could be significantly hotter and climate impacts far more extreme. 

In addition to carbon transport, mesopelagic organisms form the foundation of oceanic food webs, serving as vital prey for economically and culturally significant species such as tuna, swordfish, sharks, sperm whales, and sea lions. In other words, what happens in the mesopelagic zone ripples out across entire ocean ecosystems and affects the communities and industries that depend on them.

Moreover, it is clear a vast number of species have evolved to the unique attributes of this dynamic deep sea environment, yet scientists have only catalogued a fraction. Considering the contributions biodiversity makes to science and life saving medicines every year, this is an invaluable benefit that we cannot afford to lose.

A New Gold Rush in the Deep Sea

Despite its societal, ecological, and climate value, the mesopelagic zone is now in the crosshairs of industrial-scale exploitation. Fishing companies are exploring it as a new source of fishmeal and fish oil (FMFO), used in aquaculture, livestock, and even pet food.

Currently, about 30% of global wild-caught fish are ground into FMFO. But as fisheries falter under climate stress and overfishing, attention is shifting to deeper, more abundant mesopelagic species. For industry, it is a business opportunity. For the planet, it is a dangerous gamble. 

We Have More to Learn – And We Must Learn Fast

It is worth emphasizing that we know shockingly little about the mesopelagic zone. We do not know how many species live there, how long they live, how they reproduce, or how resilient they are to disturbance. We do not know how fast these ecosystems recover from disruption, or if they can recover at all.

Unlike more familiar fisheries, there is no baseline data, no harvest limits, and no management frameworks. It is, quite literally, a scientific black box. While there may be some level of extraction that is sustainable, we do not know what these levels may be or how economic gains weigh against the damage exploitation could cause. There is research underway exploring this, but we need more before we alter this system. 

Ultimately, we cannot manage what we do not measure. Fishing before we have the necessary knowledge in hand is a reckless gamble we simply cannot afford. The stakes for ocean health, biodiversity, and the global climate are too high.

A Global Call to Action 

Recognizing this threat, the Marine Conservation Institute, the Environmental Defense Fund, and Ocean Conservancy are working to pass a motion at the International Union for Conservation of Nature (IUCN) that would place a precautionary pause on mesopelagic exploitation and spur the research needed to answer key questions.

Motion 035: Protection of Mesopelagic Ecosystem Integrity, goes to vote at the World Conservation Congress this October. It urges nations not to authorize commercial fishing or other harmful activities in the mesopelagic until we understand it better. If passed, this motion would be a crucial step toward ensuring the mesopelagic zone continues to thrive, and human activity in this space proceeds only when proven to pose no harm.

This is not about halting all human activity in the ocean – it is about acting responsibly, and understanding that the ocean, and particularly the mesopelagic zone, is more than a resource; it is a life-support system for us and our planet.

The mesopelagic zone is one of Earth’s last truly untouched frontiers. Once lost, we do not know what will happen and we have no guarantee we can restore it.

We have a narrow window of opportunity to make the right choice. The mesopelagic zone has served us and our planet silently for millennia. Now it is time we speak up for it.

What You Can Do

If you are an ocean advocate:

• Share the importance of Motion 035 and the mesopelagic zone.
• Amplify on social media and to your networks.
• Encourage IUCN members to vote “Yes.”

If you are an IUCN Member:

• Read, comment on, and vote in support of the motion.
• Urge others to protect this extraordinary and essential ecosystem.

—

About the authors: Silvie Alexander (Blue Carbon Intern at Environmental Defense Fund), Kristin Kleisner (Lead Senior Scientist and AVP, Ocean Science at Environmental Defense Fund), Dr. Lance Morgan (marine biologist and president of Marine Conservation Institute), Chris Dorsett (Vice President, Conservation, Ocean Conservancy) and Kristina Gjerde (Senior High Seas Advisor to IUCN’s Global Marine and Polar Programme).

Featured image: Paul Caiger, Nancy Copley, Larry Madin/Woods Hole Oceanographic Institution.

Cables have been stripped from an area near a major rail junction causing disruption for weekend passengers, a train operator said.

Greater Anglia said the theft between Shenfield and Brentwood, in Essex, resulted in a “loss of signalling” and the lines being blocked to London Liverpool Street on Saturday morning.

Network Rail and British Transport Police teams were sent to replace the cables in order to reopen the railway, with trains then resuming about lunch time.

A Greater Anglia spokesman said it was “sorry for the disruption” and affected passengers would be able to claim compensation for any delays.

The spokesperson added Saturday travel tickets could now be used on Sunday instead.

It was expected to take up to three hours before the train timetable was back to normal.

Greater Anglia said trains would be delayed, altered and cancelled in order to get crews and vehicles back into the correct places.

Signalling problems were first reported early on Saturday, before Greater Anglia later said the cable has been stolen.

Shenfield is a major junction for many services, including trains using the Great Eastern Main Line.

The blocked lines had prevented trains from running between Shenfield, Romford and London.

Passengers had also been unable to travel as normal on intercity trains between Norwich, Ipswich and London Liverpool Street.

Routes between Clacton-on-Sea, Colchester, Braintree Town and Southend Victoria to Liverpool Street were blocked too.

Passengers from Norwich were told to travel to London via Cambridge instead on GTR trains between Ely and London King’s Cross.

The incident also affected trains on the Elizabeth line between Stratford and Shenfield.

Greater Anglia, which runs trains across the East of England and into London, is to be brought into public ownership on Sunday.

“[The clutch] is like 1950s technology—it’s really boring,” Westerman said (“boring,” for grid operators, is the highest form of praise). ​“The marginal cost of putting this in is like nothing compared to the cost of the plant.”

A company called SSS has built these clutches for decades. One is nearly operational in the state of Queensland at the Townsville gas-fired plant, which Siemens Energy is converting into what it calls a ​“hybrid rotating grid stabilizer.” Siemens says this project is the world’s first such conversion of a gas turbine of this size.

That particular retrofit took about 18 months and involved some relocating of auxiliary components at Townsville to make room for the new clutch. So it’s not instantaneous, but far easier than building a new synchronous condenser from scratch, and about half the cost, per Siemens.

Some novel long-duration storage techniques also provide their own spinning mass. Canadian startup Hydrostor expects to break ground early next year on a fully permitted and contracted project in Broken Hill, a city deep in the Outback of New South Wales.

Broken Hill lent its name to BHP, which started there as a silver mine in 1885 and has grown to one of the largest global mining companies. More recently, the desert landscape played host to the postapocalyptic car chases of Mad Max 2. Now, roughly 18,000 people live there, at the end of one long line connecting to the broader grid.

Hydrostor will shore up local power by excavating an underground cavity and compressing air into it; releasing the compressed air turns a turbine to regenerate up to 200 megawatts for up to eight hours, serving the community if the grid connection goes down and otherwise shipping clean power to the broader grid.

But unlike batteries, Hydrostor’s technology uses old-school generators, and its compressors contribute additional spinning metal.

“We have a clutch spec’d in for New South Wales, because they need the inertia,” Hydrostor CEO Jon Norman said. ​“It’s so simple; it’s like the same clutches on your standard car.”

Transmission grid operator Transgrid ran a competitive process to determine the best way to provide system security to Broken Hill in the event it had to operate apart from the grid, Norman said. That analysis chose Hydrostor’s bid to simply insert a clutch when it installs its machinery.

The project still needs to get built, but if up-and-coming clean storage technologies could step in to provide that grid security, it wouldn’t all have to come from ghostly gas plants lingering on the system.

“It’s a different feeling [in Australia]—there’s a can do, go get ​’em, ​‘put me in coach’ attitude,” said Audrey Zibelman, the American grid expert who ran AEMO before Westerman. ​“When you’re determined to say how best to go about this, as opposed to why it’s hard or why it doesn’t work, the solutions appear.”

By Gordon Gottsegen

If legacy financial institutions don’t embrace new technology, their competitors may leave them in the dust

New York Stock Exchange parent company Intercontinental Exchange announced a $2 billion investment into prediction-markets platform Polymarket.

This past week, Intercontinental Exchange Inc. (ICE), the parent company of the New York Stock Exchange, announced a $2 billion investment into prediction-markets platform Polymarket. On the same day, S&P Dow Jones Indices (SPGI), the company behind stock-market indexes like the S&P 500 SPX, announced a partnership with fintech company Dinari to create a crypto-focused index.

Although these are two different legacy financial institutions partnering with two different fintech companies, both announcements referenced one thing: tokenization.

Tokenization refers to the creation of a digital identifier for a real-world asset, which allows that asset to trade on a blockchain, like bitcoin (BTCUSD) does. For now, tokenization of the entire stock market is a far-off dream. But that doesn’t mean financial institutions aren’t thinking about it: Companies like Robinhood Markets Inc. (HOOD) and Coinbase Global Inc. (COIN) have both experimented with tokenizing stocks.

But when you think about the New York Stock Exchange, which was started in 1792, and Standard & Poor’s, which traces its history back to 1860, you may not think of two companies that like to move fast and break things.

So why are these two financial-industry behemoths moving in the same direction all of a sudden? The answer is two-fold. Firstly, they see it as an opportunity to bring in new sources of revenue. And secondly, if they don’t innovate, their competitors may leave them in the dust.

Racing to where the puck is going

In its partnership announcement, S&P Dow Jones Indices said that it plans to create the S&P Digital Markets 50. Similar to the S&P 500, the S&P Digital Markets 50 will be a market-tracking index that follows 15 of the largest cryptocurrencies and 35 U.S.-listed companies in the crypto space.

S&P said this new index reflects a growing demand for crypto.

“Cryptocurrencies and the broader digital-asset industry have moved from the margins into a more established role in global markets. Our expanded index suite offers market participants consistent, rules-based tools to evaluate and gain exposure to this segment,” Cameron Drinkwater, chief product and operations officer at S&P Dow Jones Indices, said in the announcement.

That demand is also there for prediction markets. While the crypto industry may have had a head start on prediction markets, companies like Polymarket and Kalshi are growing rapidly – and now processing hundreds of millions, or even billions, of dollars in volume each month.

Also read: Here’s why Wall Street is betting against DraftKings and FanDuel – and going all in on Polymarket and Kalshi

“Polymarket knows where the puck is going,” Joe Saluzzi, head of equity trading at Themis Trading, told MarketWatch. “And where I think they’re going is something called tokenization.”

Saluzzi said that tokenization of all sorts of assets, including stocks, is something that many people in the financial system are talking about. While many are excited about it, he said, there’s also an undercurrent of competition.

In the movie “The Big Short,” there’s a scene where some of the main characters go to the S&P ratings agency and ask about how bonds get their AAA rating. When the woman at the agency reveals that all mortgage-backed bonds get AAA ratings, she blurts out that if she doesn’t give the banks the ratings they want, they’ll take their business elsewhere.

Although the movie is a Hollywood dramatization of what actually happened during the 2008 financial crisis, Saluzzi made an analogy to this scene and said it may be how financial institutions are now thinking. If they don’t move quickly to embrace crypto, tokenization and prediction markets, then their competitors will.

Financial incentives

Financial institutions don’t adopt new technology merely for the sake of innovation. They do it when there’s a business incentive.

For example, retail brokerage Robinhood makes money when its customers place trades. That tends to happen more when markets are up, noted Paul Rowady, director of research at Alphacution Research Conservatory, who tracks investor and market flows.

“When the market goes down, like it did in 2022, the client equity of Robinhood goes down in correlation with that. And I think that that’s true of all these guys,” Rowady told MarketWatch.

So what does Robinhood do to keep its customers active when market conditions aren’t favorable? They expand into new product verticals, like prediction markets.

“If the market goes down, the exchanges and Robinhood want their user base to be able to gamble on sports,” Rowady said. Getting into prediction markets is a way for these financial companies to hedge by diversifying their businesses.

Robinhood makes money on transaction-based revenue. For stock exchanges like Intercontinental Exchange, they make money providing financial data to institutional clients; that includes selling things like prop data, colocation fees and access to high-speed data ports.

As markets grow and see more volume, they also create more data. Prediction markets are still dwarfed by the stock market, but they’re growing rapidly. Thomas Peterffy, the founder and chair of Interactive Brokers Group Inc. (IBKR), said that he believes prediction markets will be larger than the stock market within the next 15 years. If that happens, stock exchanges may want to hedge their bets by looking into new markets.

Read: After a big 2024 election, why prediction markets could soon eclipse the stock market

But when S&P creates an index, the business incentive may be a little more simple.

“Create an index. Get paid,” Saluzzi said. “Somebody licenses it out, you can create an ETF based on your index, and the index companies just collect the toll.”

Saluzzi noted that it always comes back to financial institutions making sure they get paid. These are businesses, and they have to make money in order to continue operating. That doesn’t mean there’s some sort of malicious intent; crypto and prediction markets have grown organically based on retail-investor demand.

“In the end, it’s giving the people what they want. The retail [investors are] demanding this,” Saluzzi said.

When centuries-old financial institutions start moving quickly, it’s a sign that the demand is there.

-Gordon Gottsegen

This content was created by MarketWatch, which is operated by Dow Jones & Co. MarketWatch is published independently from Dow Jones Newswires and The Wall Street Journal.

  (END) Dow Jones Newswires

  10-11-25 0700ET
Copyright (c) 2025 Dow Jones & Company, Inc.

Research published this week indicates that North Korean scammers are trying to trick US companies into hiring them for architectural design work, using fake profiles, résumés, and Social Security numbers to pose as legitimate workers. The hustle fits into longstanding campaigns by the hermit kingdom to steal billions of dollars from organizations around the world using careful planning and coordination to pose as professionals in all different fields.

Under pressure from the Department of Justice, Apple removed a series of apps from its iOS App Store this month related to monitoring US Immigration and Customs Enforcement activity and archiving content related to ICE’s actions. As more apps are removed, multiple developers told WIRED this week that they aren’t giving up on fighting Apple over the decisions—and many are still distributing their apps on other platforms in the meantime.

WIRED examined increasing warnings from software supply chain security researchers that the proliferation of AI-generated software in codebases will create an even more extreme version of the code transparency and accountability issues that have come up with widespread integration of open source software components. And Apple announced expansions of its bug bounty program this week, including a maximum $2 million payout for certain exploit chains that could be abused to distribute spyware, and additional bonuses for exploits found in Apple’s Lockdown Mode or in beta versions of new software.

But wait, there’s more! Each week, we round up the security and privacy news we didn’t report in depth ourselves. Click the headlines to read the full stories. And stay safe out there.

The notorious spyware vendor NSO Group, known for developing the Pegasus malware, has faced financial issues since losing a long legal battle against the secure messaging platform WhatsApp as well as a lawsuit filed by Apple. Now, the company, which has long had Israeli ownership, has been purchased by a group of US-based investors led by movie producer Robert Simonds, who helped finance Happy Gilmore, Billy Madison, The Pink Panther, Hustlers, and Ferrari, among many other films. The deal is reportedly worth “several tens of millions of dollars” and is close to completion. Israel’s Defense Export Control Agency (DECA) within the Ministry of Defense will need to approve the sale. Use of mercenary spyware has increased within some US federal government agencies since the beginning of the Trump administration.

Hundreds of national security and cybersecurity specialists who work in the US Department of Homeland Security have faced mandatory reassignment in recent weeks to roles related to President Donald Trump’s mass deportation agenda. Bloomberg reports that affected workers are largely senior staffers who are not union eligible. Workers who refuse to move roles will reportedly be dismissed. Members of DHS’s Cybersecurity and Infrastructure Security Agency (CISA) who have faced reassignment reportedly worked on “issuing alerts about threats against US agencies and critical infrastructure.” For example, CISA’s Capacity Building team has faced a number of reassignments, which could hinder access to emergency recommendations and directives for high-value federal government assets. Workers have been moved to agencies including Immigration and Customs Enforcement, Customs and Border Protection, and the Federal Protective Service.

A recent breach of a third-party customer service provider used by the communication platform Discord included a trove of data from more than 70,000 Discord users that contained identification documents as well as selfies, email addresses, phone numbers, some home location information, and more. The data was collected as part of age verification checks, a mechanism that has long been criticized for centralizing users’ sensitive information. 404 Media reports that the breach was perpetrated by attackers who are attempting to extort Discord. “This is about to get really ugly,” the hackers wrote in a Telegram channel on Wednesday while posting the stolen data.

US Immigration and Customs Enforcement inked a $825,000 contract in May with TechOps Specialty Vehicles (TOSV), a Maryland-based company that manufactures equipment and vehicles for law enforcement. The company provides products including rogue cellphone towers that are used for phone surveillance and sometimes called “stingrays” or “cell-site simulators.” Public records reviewed by TechCrunch show that the agreement describes how the company “provides Cell Site Simulator (CSS) Vehicles to support the Homeland Security Technical Operations program” and is a modification for “additional CSS Vehicles.” TOSV also began a similar $818,000 contract with ICE in September 2024, prior to the start of the Trump administration. In an email to TechCrunch, TOSV president Jon Brianas declined to share details about the contracts but confirmed that the company does provide cell-site simulators. The company does not manufacture them itself, he said.

Introduction

Breast cancer (BC) has become the most commonly diagnosed cancer and remains a leading cause of cancer-related mortality among women worldwide.¹ Ki-67, a nuclear protein associated with cellular proliferation, serves as an essential marker for evaluating tumor growth dynamics.² Elevated Ki-67 expression is associated with an increased risk of tumor invasion and recurrence.³ Moreover, it correlates with the pathological complete response rate to neoadjuvant therapy (NAT) in BC patients.⁴ A study by Chen et al found that higher pre-treatment Ki-67 levels were associated with a better clinical response to neoadjuvant chemotherapy in luminal BC subtypes. Specifically, a Ki-67 cutoff value of 25.5% was identified as a predictor of treatment response, indicating that Ki-67 could serve as a valuable biomarker for guiding individualized treatment strategies.⁵ Furthermore, the POETIC trial demonstrated that changes in Ki-67 levels during preoperative endocrine therapy were predictive of long-term outcomes, underscoring the critical role of Ki-67 in informing treatment strategies and optimizing patient management.⁶ Currently, Ki-67 expression is assessed via immunohistochemical (IHC) analysis, which typically requires an invasive core needle biopsy (CNB) prior to surgery. However, due to intratumoral heterogeneity and the limited sampling of CNB, discrepancies between CNB and postoperative specimens are frequently observed, with reported inconsistency rates ranging from 10% to 40%.⁷ Therefore, there is a pressing need for an accurate, comprehensive, and non-invasive method to predict preoperative Ki-67 expression, which is essential for clinical decision-making.

Ultrasound (US) is widely used for the diagnosis of BC due to its simplicity, low cost, and non-invasive nature.⁸ Compared to conventional US, ABVS enhances reproducibility through automated scanning,⁹ representing a significant technological advancement in US imaging.¹⁰ Its standardized coronal imaging offers detailed lesion information, making it particularly suitable for radiomics analysis.¹¹ Radiomics enables the extraction of high-throughput quantitative features from medical images, allowing for a comprehensive characterization of tumor phenotype.¹² It has been increasingly applied in tumor diagnosis, treatment planning, and prognosis prediction.¹³ The significant potential of radiomics has been fully demonstrated across numerous oncology applications, particularly in improving the predictive performance of ABVS in BC management.^11,14–17 This progress marks a major step toward non-invasive tumor biological profiling and further integrates medical imaging with personalized medicine.¹⁸ However, tumor heterogeneity, arising from variations in cell composition and spatial distribution, poses a challenge to accurate characterization.¹⁹ To better visualize and quantify this heterogeneity, voxel clusters with similar tumor biological characteristics can be grouped into sub-regions.²⁰ Habitat radiomics, an emerging approach, segments tumors into biologically similar sub-regions and extracts features from these areas to enhance the assessment of tumor heterogeneity. A recent study,²¹ involving multi-modal logistic regression models based on magnetic resonance imaging (MRI), US, and mammography revealed that incorporating peripheral tumor features (within 5 mm) yielded the best performance in distinguishing benign from malignant breast nodules, with an AUC of 0.905 (95% CI: 0.805–1). This highlights the increasing value of multimodal radiomics approaches. Previous US-based radiomics studies have shown promise in predicting Ki-67 expression in BC.^16,17 However, these studies were limited by small sample sizes and single-modality imaging, which may not fully capture the complex characteristics of tumors. Additionally, habitat radiomics has not been widely explored in this context. As tumor heterogeneity plays a crucial role in understanding tumor biology, habitat radiomics may enhance predictive accuracy. To address these limitations, our study integrates multimodal radiomics derived from ABVS and 2D US, along with habitat radiomics, to provide a more comprehensive and accurate prediction of Ki-67 expression in BC.

Motivated by these insights, we aim to evaluate the predictive value of radiomics and habitat radiomics features, captured from entire tumors and their sub-regions, using a machine learning (ML) model. Our goal is to establish a robust, non-invasive model for predicting Ki-67 expression in BC, thereby supporting personalized treatment strategies and improving prognostic assessment.

Materials and Methods

Study Population

This retrospective study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of the First Affiliated Hospital of Anhui Medical University (Approval No. PJ2023-07-11). Given the retrospective nature of the study, which involved the use of previously collected cases and medical records without any new clinical interventions, the requirement for informed consent was waived in accordance with relevant ethical guidelines. To ensure data reliability, strict inclusion and exclusion criteria were applied. The inclusion criteria were as follows: (1) Patients pathologically diagnosed with BC who underwent both ABVS and conventional US examinations within two weeks prior to surgery; (2) No NAT administered before surgery; (3) Availability of complete clinical and pathological information. The exclusion criteria were: (1) Incomplete clinical or pathological data; (2) History of other malignant tumors; (3) Receipt of preoperative NAT; (4) Maximum tumor diameter > 50 mm as measured by US; (5) Presence of bilateral BC. A total of 426 patients with BC met the eligibility criteria. The cases were randomly divided into a training set (n = 297) and a validation set (n = 127), following a 7:3 ratio (Figure 1).

Figure 1 Flowchart of the entire research. Abbreviations: ROI, Region of Interest; ABVS, automated breast volume scanner; RadABVS+2D, Radiomics ABVS and 2D model; HadABVS+2D, Habitat radiomics ABVS and 2D model; Rad-HabABVS+2D, Radiomics and Habitat radiomics ABVS and 2D model; CMClinical + Rad-Hab, Clinical–Radiomics–Habitat ABVS+2D Combined Model.

Clinical and Histopathological Data Acquisition and Assessment

Details regarding image acquisition, evaluation procedures, and US equipment specifications are provided in Supplementary Material S1. Breast lesion classification was performed according to the fifth edition of the American College of Radiology (ACR) Breast Imaging Reporting and Data System (BI-RADS).²² The Ki-67 proliferation index was calculated based on the percentage of malignant cells showing positive nuclear staining for Ki-67. A Ki-67 score ≥ 20% was defined as high expression, while a score < 20% was considered low expression.^23,24

Segmentation of Regions of Interest (ROIs) and Generation of Habitat Sub-Regions

To segment the lesion regions on ABVS coronal images and 2D US images, the ITK-SNAP software (version 3.8, website: www.itk-snap.org) was used for tumor segmentation. Details of the tumor ROI delineation process are provided in Supplementary Material S2. Before generating habitat sub-regions, the tumors were first localized on both ABVS and 2D images. Each image was paired with a tumor mask of identical dimensions. The bounding box of the tumor region was extracted from the corresponding mask file, and this region was used to define the area for further analysis. Radiomic features were then extracted from these sub-regions. For each pixel within the tumor, a 5 × 5×5 sliding window was applied, expanding outward by two pixels in each direction. Radiomic features within this window were extracted using the pyradiomics library. To enable subsequent clustering analysis, all features were normalized to a range of 0 to 1. This standardization mitigates the influence of features with larger numerical scales and enhances the accuracy and robustness of clustering.²⁵

Although larger windows and a greater number of features may improve robustness to noise, they also significantly increase computational complexity, especially when extracting features at the pixel level. Therefore, in this study, the number of radiomic features was limited to five, specifically those derived from the Gray-Level Co-occurrence Matrix (GLCM). The GLCM effectively captures subtle textural variations in the image, reflecting microscopic irregularity and complexity. It has been widely used to reveal the potential relationship between tumor tissue structure and biological behavior, making it an important tool in the study of tumor heterogeneity^25,26 Subsequently, to further quantify image data, each pixel’s local histological features were transformed into a five-dimensional feature vector. This vector integrates multiple aspects of local feature information, such as texture, contrast, and uniformity, facilitating subsequent quantitative analysis and model construction.

A Gaussian mixture model (GMM) clustering algorithm was employed to identify tumor sub-regions composed of biologically similar pixels. Clustering was performed at the cohort level, rather than the individual patient level, to ensure consistent cluster assignment across patients and to allow the propagation of cluster centers from the training set to the test set, ensuring consistent clustering during model application. To determine the optimal number of clusters (ie, habitats), the Silhouette coefficient was used to evaluate clustering performance across a range of k values from 2 to 10. Following clustering, each cluster was assigned a unique color label to generate a cluster label map, which reflected the global distribution of internal regions within the tumor.

Radiomics Feature Extraction

Radiomics features were extracted from both tumor regions and sub-regions within the ABVS and 2D US images. Prior to feature extraction, image preprocessing was performed to ensure consistency across datasets. First, the signal intensity of the original images was normalized and standardized to a range of 0–100 to minimize intensity variations between images. Next, spatial resampling was conducted to achieve a uniform in-plane pixel resolution of 2 mm in the XY direction. All feature extraction was confined to the two-dimensional plane. Image gray levels were discretized based on a predefined bin width (eg, grouping every 5 intensity values), enabling standardized texture quantification. All processing was applied exclusively to regions defined by the specified labels in the tumor mask. A wavelet filter was used for feature enhancement prior to extraction. Radiomics features were extracted using the open-source Python package PyRadiomics (https://pyradiomics.readthedocs.io/en/latest/index.html, version 3.0.1), developed by the Computational Imaging Bioinformatics Laboratory at Harvard Medical School. Both unfiltered (from original images) and filtered features were included in the analysis. A total of 464 features were extracted and categorized into the following classes: 90 first-order features, 9 shape features, 110 Gray-Level Co-occurrence Matrix (GLCM) features, 80 gray-level size zone matrix (GLSZM) features, 80 gray-level run length matrix (GLRLM) features, 25 neighboring gray-tone difference matrix (NGTDM) features, 70 gray-level dependence matrix (GLDM) features. PyRadiomics adheres to the standards of the Imaging Biomarker Standardization Initiative (IBSI), ensuring consistent definitions and methodologies for radiomic feature extraction.

Radiomics Feature Selection

The selection of predictive features associated with Ki-67 expression was performed through a multi-step process. This included screening radiomics features, sub-regional radiomics features, and clinical US features. The detailed feature selection workflow is presented in Supplementary Material S3.

Construction and Validation of ML Models

After feature selection and fusion, five predictive models were developed, namely the clinical model, the radiomics model (Rad _{ABVS + 2D}), the habitat radiomics model (Hab _{ABVS + 2D}), the combined radiomics model (Rad-Hab _{ABVS + 2D}), and the Clinical–Radiomics–Habitat _ABVS+2D Combined Model(CM _{Clinical + Rad-Hab}). The entire model construction workflow is shown in Figure 2. All models were established using radiomics features derived from ABVS and 2D images. Four ML classifiers were employed: logistic regression (LR), ExtraTree (ET), EXtreme Gradient Boosting (XGBoost), and Light Gradient Boosting Machine (LightGBM). To reduce overfitting, 5-fold cross-validation was applied within the training cohort to optimize hyperparameters for each classifier. Model performance was evaluated by plotting receiver operating characteristic (ROC) curves and calculating the area under the curve (AUC). The DeLong test was applied to statistically compare ROC performance across different models, while the Hosmer-Lemeshow test assessed the models’ goodness-of-fit. Clinical utility was further evaluated using the DCA to estimate the net benefit of each model in guiding clinical decision-making.

Figure 2 Workflow of radiomics analysis. This figure illustrates the segmentation, feature extraction, and feature selection process for ABVS and 2D images in breast cancer. Abbreviations: ABVS, automated breast volume scanner; ICC, intraclass correlation coefficient; LASSO, least absolute shrinkage and selection operator; PCC, Pearson correlation coefficient; RadABVS+2D, Radiomics ABVS and 2D model; HadABVS+2D, Habitat radiomics ABVS and 2D model; Rad-HabABVS+2D, Radiomics and Habitat radiomics ABVS and 2D model; CMClinical + Rad-Hab, Clinical–Radiomics–Habitat ABVS+2D Combined Model.

Statistical Analysis

All statistical analyses and data visualizations were performed using R software (version 4.4.2) and JD_DCPM (V6.03, Jingding Medical Technology Co., Ltd.) and Python (version 3.8; https://www.python.org). Continuous variables were presented as mean ± standard deviation, while categorical variables were expressed as counts (n) and percentages (%). For quantitative data following a normal distribution, Student’s t-test was used. Levene’s test was employed to assess the homogeneity of variance. The Kruskal–Wallis test was used for non-normally distributed data. The Chi-square test was applied to compare categorical data. The DeLong test was used to compare the ROC performance among different models. All statistical tests were two-sided, and statistical significance was set at P < 0.05.

Results

Comparison of Baseline Data

A total of 426 eligible BC patients were included in this study, with 297 patients assigned to the training set and 129 to the validation set. A summary of baseline clinical and US characteristics is presented in Table 1. Among these variables, multivariate logistic regression analysis identified T-stage and US-ALNs as independent predictors. These two factors were therefore incorporated into the clinical model (Table 2). Using LR, the clinical model achieved an AUC of 0.720 (95% CI: 0.662–0.775) in the training set and 0.648 (95% CI: 0.557–0.734) in the validation set.

Table 1 Baseline Clinical Ultrasound Characteristics in the Training and Validation Sets

Table 2 Univariate and Multivariate Logistic Regression Analyses of Clinical Ultrasound Characteristics in the Training Set

Screening of Radiomics Features

Rad _{ABVS + 2D} Feature Selection

In the training set, the tumor ROI were delineated on both ABVS and 2D US images, and a total of 464 radiomics features were extracted from each image. After standardization, features with an intraclass correlation coefficient (ICC) > 0.75 were retained for subsequent analysis, resulting in 898 features. Subsequently, univariate t-tests and Pearson correlation coefficient (PCC) analyses were conducted to evaluate the relationships among these features. Finally, the least absolute shrinkage and selection operator (LASSO) algorithm was used to select the most predictive features based on the optimal λ value. The Rad _{ABVS + 2D} model identified 15 radiomics features, 9 from ABVS images and 6 from 2D US images, that were significantly correlated with Ki-67 expression (λ=0.037, Figure S1).

Hab _{ABVS + 2D} Feature Selection

In this study, the Silhouette Coefficient was used to evaluate clustering performance and determine the optimal number of clusters (Figure 3A). The analysis revealed that when the number of clusters was set to 3, the silhouette coefficient reached its highest value, indicating the best clustering performance. Accordingly, the tumor ROI was divided into three habitat sub-regions for subsequent feature extraction and model construction. For each sub-region, 464 radiomics features were extracted, resulting in a total of 2694 features (898 features × 3 sub-regions), following intraclass correlation coefficient (ICC) filtering. Feature selection was then conducted, and the final Hab _{ABVS + 2D} model identified 13 ABVS and 12 2D radiomics features that were significantly correlated with Ki-67 expression (λ=0.018, Figure S2). The habitat feature maps and corresponding sub-region segmentations are shown in Figure 3B and C.

Figure 3 Determination of optimal clusters and visualization of habitat clusters with corresponding feature maps. (A) Silhouette coefficient plot was used to determine the optimal number of clusters (k), indicating that k=3 is optimal. (B) Distribution of habitats under different clustering numbers, with different colors representing different clusters. (C) Feature maps for the following parameters: original_glcm_DifferenceVariance, original_glcm_Idm, original_glcm_InverseVariance, original_glcm_JointAverage, original_glcm_JointEnergy, original_glcm_MaximumProbability.

Rad-Hab _{ABVS + 2D} Feature Selection

According to the methodology described above, the Rad-Hab _{ABVS + 2D} model initially included a total of 3592 features (898 + 898×3). After feature selection and dimensionality reduction, the final model identified 16 ABVS features and 8 2D radiomics features that were significantly correlated with Ki-67 expression (λ=0.012, Figure S3).

Construction and Validation of Each Models

The optimal model for the traditional Rad _{ABVS + 2D} configuration was LR, as detailed in Table S1 and Figure S4A and B. In the training set, the model achieved an AUC of 0.755 (95% CI: 0.688–0.813), with accuracy, sensitivity, specificity, and F1-score of 0.670, 0.824, 0.602, and 0.605, respectively. In the internal validation set, the AUC was 0.603 (95% CI: 0.515–0.690), with corresponding values of 0.682 for accuracy, 0.442 for sensitivity, 0.802 for specificity, and 0.481 for F1-score.

Figure 4 Comparison of the performance of different radiomics models in the training and validation sets. This figure shows the distribution of radiomics model outputs between the high- and low-expression Ki-67 groups. (A and B) RadABVS+2D; (C and D) HabABVS+2D; (E and F) Rad-HabABVS+2D; (G and H) CMClinical + Rad-Hab. Abbrteviations: RadABVS+2D, Radiomics ABVS and 2D model; HadABVS+2D, Habitat radiomics ABVS and 2D model; Rad-HabABVS+2D, Radiomics and Habitat radiomics ABVS and 2D model; CMClinical + Rad-Hab, Clinical–Radiomics–Habitat ABVS+2D Combined Model; ABVS, automated breast volume scanner.

For the Hab _{ABVS + 2D} model, the best-performing algorithm was ExtraTree, as detailed in Table S2 and Figure S4C and D). In the training set, the model achieved an AUC of 0.779 (95% CI: 0.712 −0.825), with accuracy, sensitivity, specificity, and F1-score of 0.781, 0.505, 0.903, and 0.586. In the validation set, the AUC was 0.664 (95% CI: 0.579–0.755), and accuracy, sensitivity, specificity, and F1-score were 0.605, 0.721, 0.547, and 0.549, respectively.

The Rad-Hab _{ABVS + 2D} model performed best when using the XGBoost, as detailed in Table S3 and Figure S4E and F. In the training set, the model achieved an AUC of 0.935 (95% CI: 0.910–0.962), with accuracy, sensitivity, specificity, and F1-score of 0.869, 0.868, 0.869, and 0.802, respectively. In the validation set, the AUC was 0.850 (95% CI: 0.789–0.918), with values of 0.806 for accuracy, 0.744 for sensitivity, 0.837 for specificity, and 0.719 for F1-score.

After combining Rad-Hab_{ABVS + 2D} features with clinical features, the ML model achieving the best performance was based on LightGBM, as detailed in Table S4 and Figure S4G and H. This resulting model was designated CM _{Clinical + Rad-Hab}. In the training set, the model achieved an AUC of 0.951 (95% CI: 0.928–0.973), with accuracy, sensitivity, specificity, and F1-score of 0.886, 0.890, 0.883, and 0.827, respectively. In the validation set, the AUC was 0.884 (95% CI: 0.831–0.949), with corresponding values of 0.783 for accuracy, 0.860 for sensitivity, 0.744 for specificity, and 0.725 for F1-score. Across the radiomics-based models, statistically significant differences in Ki-67 expression were observed between the high- and low-expression groups, with the exception of the Rad _{ABVS + 2D} in the validation set (P <0.01, Figure 4). The differences in radiomics features from specific habitat sub-regions included in the CM _{Clinical + Rad-Hab} model are shown in Figure 5.

Figure 5 Box plot showing differences in ABVS and 2D image habitat radiomics features between the low- and high-expression groups in the CMClinical+Rad_Hab model. The P value indicates statistical significance and is displayed in each feature chart. Black dots represent individual data points; the horizontal line within each box indicates the median (50th percentile), and the upper and lower edges represent the 25th and 75th percentiles, respectively. Abbreviation: ABVS, automated breast volume scanner.

Comparison of Model Performance

In the training set, the DeLong test results indicated that the AUC differences between the CM _{Clinical + Rad-Hab} model and the Clinical, Rad _{ABVS + 2D}, Hab _{ABVS + 2D}, and Rad-Hab _{ABVS + 2D} models were all statistically significant (Z = 7.979, P < 0.001; Z = 7.162, P < 0.001; Z = 6.017, P < 0.001; Z = 2.669, P = 0.007). Similar results were observed in the validation set (Z = 4.829, P < 0.001; Z = 5.665, P < 0.001; Z = 4.885, P < 0.001; Z = 2.662, P = 0.009). These findings indicate that the CM _{Clinical + Rad-Hab} model significantly outperformed the other models in predicting Ki-67 expression, as presented in Figure 6A and B, Table 3. The calibration curves of the CM _{Clinical + Rad-Hab} model exhibited strong agreement between expected and observed outcomes in both the high and low Ki-67 expression groups, surpassing the calibration performance of the other models, as illustrated in Figure 6C and D. The Hosmer-Lemeshow test further confirmed good model calibration for the CM _{Clinical + Rad-Hab} model, with P = 0.645 and 0.587 for the two sets.

Table 3 Comparison of Radiomics and Sub-Region Features Across Different Models Using ABVS and 2D Imaging

Figure 6 ROC curves (A and B), calibration curves (C and D), and DCA curves (E and F) for different models in the training and validation sets. Abbreviations: DCA, Decision Curve Analysis; ROC, Receiver Operating Characteristic; ABVS, automated breast volume scanner; RadABVS+2D, Radiomics ABVS and 2D model; HadABVS+2D, Habitat radiomics ABVS and 2D model; Rad-HabABVS+2D, Radiomics and Habitat radiomics ABVS and 2D model; CM Clinical + Rad-Hab, Clinical–Radiomics–Habitat ABVS+2D Combined Model.

The DCA demonstrated that CM _{Clinical + Rad-Hab} achieved superior net clinical benefit compared to all- or no-treatment strategies, as shown in Figure 6E and F. Furthermore, incorporating the Rad-Hab _{ABVS + 2D} model with clinical risk factors (T-stage, US-ALNs) significantly improved the predictive performance of the CM _{Clinical + Rad-Hab}. This improvement was confirmed by significant increases in both the net reclassification improvement (NRI) and integrated discrimination improvement (IDI) indicators in the training and validation sets (Table 4). NRI and IDI values for the other models can be found in Table S5.

Table 4 Evaluation of CM Clinical + Rad-Hab and Clinical Models Using NRI and IDI

Discussion

In this study, we proposed a new approach that integrates ABVS- and 2D-based lesion imaging with radiomics and habitat radiomics to predict the expression of Ki-67 in BC. Our results demonstrated that the Rad-Hab _ABVS+2D model could accurately predict Ki-67 expression, and that incorporating clinical factors into this model (CM _{Clinical + Rad-Hab}) further enhanced predictive performance. This approach provides a new strategy for constructing Ki-67 prediction models in BC. More importantly, by leveraging ML to integrate radiomics features, our approach addresses tumor heterogeneity and enables complex nonlinear feature interpretation, thereby improving predictive accuracy.

A previous study utilized six ML algorithms to integrate US radiomics with postoperative pathological features to predict Ki-67 expression in breast malignancies, reporting that the LR achieved the best average predictive performance (training AUC: 0.793, validation AUC: 0.798).¹⁷ Another similar study also attempted to predict Ki-67 expression, but the predictive performance was only moderate.²⁷ In our study, the Rad-Hab_{ABVS + 2D} model achieved AUC values of 0.935 in the training cohort and 0.850 in the validation cohort, demonstrating strong predictive capability. Furthermore, by integrating US features (US-ALNs and T-stage), the LightGBM-based CM_{Clinical + Rad-Hab} model achieved even higher performance (training AUC: 0.951, validation AUC: 0.884). The calibration curve demonstrated excellent predictive accuracy, while DCA showed good clinical benefits. These findings highlight the potential of ML-based multimodal radiomics as a non-invasive tool for personalized clinical diagnosis and treatment planning.

Radiomics has emerged as a specialized field within medical imaging, enabling the extraction of high-throughput quantitative features from medical images. This approach captures the intrinsic characteristics of breast tumors, and to some extent, provides additional insights into Ki-67 expression in BC.²⁸ By applying various ML or deep learning algorithms to different imaging modalities,^2,11,27 key information about Ki-67 expression in BC can be obtained, supplementing conventional pathological assessments. In our study, dimensionality reduction of the Rad-Hab_{ABVS + 2D} model yielded 24 key radiomics features related to the Ki-67 status. These features, combined with clinical variables, were evaluated using four ML algorithms (LR, ET, XGBoost, and LightGBM). The results showed that LightGBM demonstrated the best predictive performance. As a gradient boosting decision tree algorithm, LightGBM enhances performance by increasing the number of boosting trees and offers efficiency and flexibility in modeling nonlinear relationships. Its ability to handle large datasets with high-dimensional features makes it particularly effective for predictive modeling.²⁹ Numerous studies have confirmed the robustness and reliability of LightGBM in classification and regression tasks,^29–31 further proving its robustness and reliability as a powerful classification tool.

The results of this study indicated that, within our developed CM _{Clinical + Rad-Hab} model, the Rad-Hab _{ABVS + 2D} component was the most influential predictive factor. Tumors represent complex ecosystems, and intratumoral heterogeneity is often distributed unevenly throughout the lesion.³² However, regional heterogeneity within the tumor is frequently overlooked. Recent studies have shown that voxels at different spatial locations within an image may share similar imaging features, and these sub-regions may exhibit comparable biological characteristics.³³ Therefore, to better study and quantify such regional heterogeneity, habitat radiomics divides tumors into sub-regions consisting of voxel clusters with similar characteristics for unsupervised analysis.³⁴ Recently, this approach has achieved promising results in the assessment of parotid gland tumors, liver cancer, and BC, among others.^3,32,35 In our CM_{Clinical+Rad-Hab} model, it is noteworthy that most of the 24 selected radiomics features were derived from wavelet features (16 features). These wavelet features effectively capture heterogeneity at multiple spatial scales (Figure S3),³⁶ patterns that are typically imperceptible through visual inspection but can be extracted using radiomics and mathematically correlated with the Ki-67 status. Although the Rad_Hab_ABVS+2D model alone demonstrated strong predictive performance, incorporating clinical parameters (T-stage, OR = 3.078; US-ALNs, OR = 4.759) significantly improved the overall performance of the CM_{Clinical+ABVS+2D} model (training set: Z = 2.669, P = 0.007; validation set: Z = 2.662, P = 0.009). This observation result is consistent with the studies,^9,11,37 emphasizing that only by fully leveraging multi-dimensional data can the radiomics features reach their full potential in predicting Ki-67 expression (Table 3 and Figure 6).

Although previous studies have used either ABVS or 2D single-modality images to predict Ki-67 expression, to our knowledge, this study is the first to integrate radiomics features from both ABVS and 2D US images, including sub-regional (habitat) features, for Ki-67 prediction. However, several limitations should be noted. First, as a single-center retrospective study, potential selection bias may limit the generalizability of the findings; future multicenter, prospective studies are needed to address this issue. Second, while this study is the first to employ multimodal sub-region radiomics analysis for Ki-67 prediction, model interpretability remains challenging due to the complexity of habitat radiomics, and incorporating explainable AI techniques could improve clinical interpretability. Third, our study was restricted to ABVS and 2D US images, excluding other modalities such as mammography and MRI, which could provide complementary diagnostic information; integrating these modalities in future studies may further enhance model performance. Lastly, radiomics features were extracted only from ABVS coronal images and the maximum 2D tumor section; extending analyses to 3D imaging or multiple planes may provide a more comprehensive characterization of tumor heterogeneity. In the future, we plan to explore correlations between radiomics or deep learning features and biomarkers such as Ki-67 and HER-2 expression, as well as treatment responses to NAT, using multimodal imaging, to provide deeper insights for the precise treatment of BC.

Conclusion

In conclusion, this study demonstrates that integrating ML with ABVS and 2D US tumor- and sub-regional-based radiomics features can effectively predict Ki-67 expression in BC. The developed CM_{Clinical + Rad-Hab} model, which combines US indicators with radiomics features, achieves excellent classification performance and shows substantial clinical value. This approach holds significant potential for improving preoperative diagnostic accuracy and facilitating therapeutic efficacy assessment of BC biomarkers.

Ethics Statement

This retrospective study was approved by the Ethics Committee of the First Affiliated Hospital of Anhui Medical University, with a waiver of informed consent. All research data were de-identified and processed in strict accordance with relevant privacy protection regulations to ensure the confidentiality of participant information.

Acknowledgments

We sincerely acknowledge all the staff involved in implementing the intervention and assessing the research components. We also acknowledge the PixelMed AI platform and its developers for their valuable assistance with the code used in this revised manuscript.

Funding

This study was supported by Anhui Provincial Natural Science Foundation (Grant number: 2308085MH278), Health Research Program of Anhui (Grant number: AHWJ2023A10017), Anhui Provincial Health Commission Scientific Research Project (Grant number: AHWJ2024Aa30096), and Scientific Research Foundation for High-level Talents of First Affiliated Hospital of Wannan Medical College (Grant number: YR202436).

Disclosure

The author declares no potential conflicts of interest with respect to the research, authorship, and publication of this article.

References

1. Sung H, Ferlay J, Siegel RL, et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2021;71(3):209–249. doi:10.3322/caac.21660

2. Li HE, Cheng C. Development and assessment of a predictive model for Ki-67 expression using ultrasound indicators and non-morphological magnetic resonance imaging parameters before breast cancer therapy. Ultrason Imaging. 2024;46(6):332–341. doi:10.1177/01617346241271107

3. Yerushalmi R, Woods R, Ravdin PM, Hayes MM, Gelmon KA. Ki67 in breast cancer: prognostic and predictive potential. Lancet Oncol. 2010;11(2):174–183. doi:10.1016/S1470-2045(09)70262-1

4. Mukai H, Yamaguchi T, Takahashi M, et al. Ki-67 response-guided preoperative chemotherapy for HER2-positive breast cancer: results of a randomised Phase 2 study. Br J Cancer. 2020;122(12):1747–1753. doi:10.1038/s41416-020-0815-9

5. Chen R, Ye Y, Yang C, et al. Assessment of the predictive role of pretreatment Ki-67 and Ki-67 changes in breast cancer patients receiving neoadjuvant chemotherapy according to the molecular classification: a retrospective study of 1010 patients. Breast Cancer Res Treat. 2018;170(1):35–43. doi:10.1007/s10549-018-4730-1

6. Smith I, Robertson J, Kilburn L, et al. Long-term outcome and prognostic value of Ki67 after perioperative endocrine therapy in postmenopausal women with hormone-sensitive early breast cancer (POETIC): an open-label, multicentre, parallel-group, randomised, Phase 3 trial. Lancet Oncol. 2020;21(11):1443–1454. doi:10.1016/S1470-2045(20)30458-7

7. Rossi C, Fraticelli S, Fanizza M, et al. Concordance of immunohistochemistry for predictive and prognostic factors in breast cancer between biopsy and surgical excision: a single-centre experience and review of the literature. Breast Cancer Res Treat. 2023;198(3):573–582. doi:10.1007/s10549-023-06872-9

8. Fang J, Zhao W, Li Q, Zhang B, Pu C, Wang H. Correlation analysis of conventional ultrasound characteristics and strain elastography with Ki-67 status in breast cancer. Ultrasound Med Biol. 2020;46(11):2972–2978. doi:10.1016/j.ultrasmedbio.2020.06.024

9. Li F, Zhu TW, Lin M, et al. Enhancing Ki-67 prediction in breast cancer: integrating intratumoral and peritumoral radiomics from automated breast ultrasound via machine learning. Acad Radiol. 2024;31(7):2663–2673. doi:10.1016/j.acra.2023.12.036

10. Vourtsis A. Three-dimensional automated breast ultrasound: technical aspects and first results. Diagn Interv Imaging. 2019;100(10):579–592. doi:10.1016/j.diii.2019.03.012

11. Wu Y, Ma Q, Fan L, Wu S, Wang J. An automated breast volume scanner-based intra- and peritumoral radiomics nomogram for the preoperative prediction of expression of Ki-67 in breast malignancy. Acad Radiol. 2024;31(1):93–103. doi:10.1016/j.acra.2023.07.004

12. Lambin P, Rios-Velazquez E, Leijenaar R, et al. Radiomics: extracting more information from medical images using advanced feature analysis. Eur J Cancer. 2012;48(4):441–446. doi:10.1016/j.ejca.2011.11.036

13. Mayerhoefer ME, Materka A, Langs G, et al. Introduction to radiomics. J Nucl Med. 2020;61(4):488–495. doi:10.2967/jnumed.118.222893

14. Jiang W, Deng X, Zhu T, Fang J, Li J. ABVS-based radiomics for early predicting the efficacy of neoadjuvant chemotherapy in patients with breast cancers. Breast Cancer. 2023;15:625–636. doi:10.2147/BCTT.S418376

15. Ma Q, Shen C, Gao Y, et al. Radiomics analysis of breast lesions in combination with coronal plane of ABVS and strain elastography. Breast Cancer. 2023;15:381–390. doi:10.2147/BCTT.S410356

16. Wang SJ, Liu HQ, Yang T, et al. Automated Breast Volume Scanner (ABVS)-based radiomic nomogram: a potential tool for reducing unnecessary biopsies of BI-RADS 4 lesions. Diagnostics. 2022;12(1). doi:10.3390/diagnostics12010172

17. Wu J, Fang Q, Yao J, et al. Integration of ultrasound radiomics features and clinical factors: a nomogram model for identifying the Ki-67 status in patients with breast carcinoma. Front Oncol. 2022;12:979358. doi:10.3389/fonc.2022.979358

18. Chen Y, Xie Y, Li B, et al. Automated Breast Ultrasound (ABUS)-based radiomics nomogram: an individualized tool for predicting axillary lymph node tumor burden in patients with early breast cancer. BMC Cancer. 2023;23(1):340. doi:10.1186/s12885-023-10743-3

19. Li J, Qiu Z, Zhang C, et al. ITHscore: comprehensive quantification of intra-tumor heterogeneity in NSCLC by multi-scale radiomic features. Eur Radiol. 2023;33(2):893–903. doi:10.1007/s00330-022-09055-0

20. Jardim-Perassi BV, Huang S, Dominguez-Viqueira W, et al. Multiparametric MRI and coregistered histology identify tumor habitats in breast cancer mouse models. Cancer Res. 2019;79(15):3952–3964. doi:10.1158/0008-5472.CAN-19-0213

21. Wu J, Li Y, Gong W, Li Q, Han X, Zhang T. Multi-modality radiomics diagnosis of breast cancer based on MRI, ultrasound and mammography. BMC Med Imaging. 2025;25(1):265. doi:10.1186/s12880-025-01767-1

22. D’Orsi C, Morris E, Mendelson E. ACR BI-RADS® Atlas, Breast Imaging Reporting and Data System. 2013:121–140.

23. Jiang M, Zhang D, Tang SC, et al. Deep learning with convolutional neural network in the assessment of breast cancer molecular subtypes based on US images: a multicenter retrospective study. Eur Radiol. 2021;31(6):3673–3682. doi:10.1007/s00330-020-07544-8

24. de Gregorio A, Friedl TWP, Hering E, et al. Ki67 as proliferative marker in patients with early breast cancer and its association with clinicopathological factors. Oncology. 2021;99(12):780–789. doi:10.1159/000517490

25. Zhang Y, Ma H, Lei P, Li Z, Yan Z, Wang X. Prediction of early postoperative recurrence of hepatocellular carcinoma by habitat analysis based on different sequence of contrast-enhanced CT. Front Oncol. 2024;14:1522501. doi:10.3389/fonc.2024.1522501

26. Zhang J, Wang X, Zhang L, et al. Radiomics predict postoperative survival of patients with primary liver cancer with different pathological types. Ann Transl Med. 2020;8(13):820. doi:10.21037/atm-19-4668

27. Zhu Y, Dou Y, Qin L, Wang H, Wen Z. Prediction of Ki-67 of invasive ductal breast cancer based on ultrasound radiomics nomogram. J Ultrasound Med. 2023;42(3):649–664. doi:10.1002/jum.16061

28. Fan M, Liu Z, Xu M, et al. Generative adversarial network-based super-resolution of diffusion-weighted imaging: application to tumour radiomics in breast cancer. NMR Biomed. 2020;33(8):e4345. doi:10.1002/nbm.4345

29. Basha SM, Rajput DS, Vandhan V. Impact of gradient ascent and boosting algorithm in classification. Int J Intell Eng Syst. 2018;11(1):41–49. doi:10.22266/ijies2018.0228.05

30. Zhang J, Mucs D, Norinder U, Svensson F. LightGBM: an effective and scalable algorithm for prediction of chemical toxicity-application to the Tox21 and mutagenicity data sets. J Chem Inf Model. 2019;59(10):4150–4158. doi:10.1021/acs.jcim.9b00633

31. Yanagawa R, Iwadoh K, Akabane M, et al. LightGBM outperforms other machine learning techniques in predicting graft failure after liver transplantation: creation of a predictive model through large-scale analysis. Clin Transplant. 2024;38(4):e15316. doi:10.1111/ctr.15316

32. Ma Q, Wang J, Tu Z, et al. Prediction model of axillary lymph node status using an automated breast volume ultrasound radiomics nomogram in early breast cancer with negative axillary ultrasound. Front Immunol. 2025;16:1460673. doi:10.3389/fimmu.2025.1460673

33. O’Connor JP, Rose CJ, Waterton JC, Carano RA, Parker GJ, Jackson A. Imaging intratumor heterogeneity: role in therapy response, resistance, and clinical outcome. Clin Cancer Res. 2015;21(2):249–257. doi:10.1158/1078-0432.CCR-14-0990

34. Kim J, Ryu SY, Lee SH, Lee HY, Park H. Clustering approach to identify intratumour heterogeneity combining FDG PET and diffusion-weighted MRI in lung adenocarcinoma. Eur Radiol. 2019;29(1):468–475. doi:10.1007/s00330-018-5590-0

35. Xiao Y, Huang P, Zhang Y, et al. Component prediction in combined hepatocellular carcinoma-cholangiocarcinoma: habitat imaging and its biologic underpinnings. Abdom Radiol. 2024;49(4):1063–1073. doi:10.1007/s00261-023-04174-8

36. Xie C, Yang P, Zhang X, et al. Sub-region based radiomics analysis for survival prediction in oesophageal tumours treated by definitive concurrent chemoradiotherapy. EBioMedicine. 2019;44:289–297. doi:10.1016/j.ebiom.2019.05.023

37. Liu J, Wang X, Hu M, et al. Development of an ultrasound-based radiomics nomogram to preoperatively predict Ki-67 expression level in patients with breast cancer. Front Oncol. 2022;12:963925. doi:10.3389/fonc.2022.963925