Fluid forms that strike a delicate balance between openness and intimacy: the Kering Pavilion draws inspiration from the natural environment to reflect the Group’s living, evolving force of creativity. Gently curved enclosures define individual spaces where each House reveals its distinctive universe and savoir-faire. Meanwhile, open display tables, accentuated by curvilinear panels, form a landscape that invites exploration. Light filters through the panels, casting shadows and animating the displays with movement and depth.
The construction itself reflects Kering sustainability commitments: modular prefabricated elements reuse 50% of materials from the previous year while, for part of the furniture, traditional leather is replaced with full-plant-life-cycle leather provided by PEELSPHERE, the first-place winner of the 2nd Kering Generation Award China.
A collection of Polaroids from Hidden’s September event. Illustration: Guardian Design/Photos courtesy of Stella Barey
Stella Barey has an hour for lunch. At 1.30pm, she loads her banged-up Tacoma with her three Belgian malinois and drives to a secret Los Angeles hiking trail. There, she gulps down a tapioca pudding and laces up her sneakers. After checking over her shoulder for foot traffic, she pulls down her brown sweatpants and jiggles her bare ass for the camera. Then come the undies. Her coiffed landing strip hovers above the rocks as a rush of urine floods the trail. Every mile she walks, she films another video: a flash, a moon, a finger up the ass.
When Barey decided in 2020 to pursue porn full-time, she did not imagine that at 28 she would spend more time hunched over a desk – not in the fun way – making flow charts, scheduling Zoom calls, and sending pitch decks. “I’m at my happiest when I’m making a video like putting a strawberry in my butt and pushing it out,” she says. “Now I’m on calls all day and I have tech neck.” Known online as the “Anal Princess”, with large, blinking Shelley Duvall eyes and an American Girl doll pout, she will try anything once – even the title “tech founder”.
Barey is looking to disrupt the porn landscape with Hidden, a site designed to alleviate creator burnout and restore the fun in making and consuming adult content – a place where, unlike OnlyFans, she can post public exposure and piss clips from her daily lunch-break hikes.
Hidden arrives during an uncertain time for porn, especially for gen Z. They are skeptical of it, raised on it and increasingly behind the camera themselves. The platform reflects the DIY sensibility that Barey’s generation grew up with, and its mission speaks to their conflicted relationship with big tech, sex work and making a living online. “At the end of the day, Hidden is more than porn,” Barey says. “It’s a political statement” – and one of the only sex worker-founded sex-tech companies hosting adult content today.
Born in San Juan Capistrano, California and raised by a gynaecologist mother, Barey cannot remember a time when she was not insatiably curious about her body. At NYU, she invented her own major – ethical healthcare systems and policy – then finished med school prerequisites for gynaecology at University of California, Los Angeles, where she gravitated to the city’s X-rated underbelly: sex parties, industry mixers, late-night friendships with porn girls. Trapped in her apartment during the first wave of Covid, grinding through schoolwork, she started experimenting with anal sex with her then boyfriend – and sharing her sexploits via dirty story times on TikTok. “This is when TikTok still felt like this hidden little piece of the internet,” she recalls.
Her third video went viral. Men in the comments wanted her OnlyFans. Women wanted more story times. By 2021, Barey had given them both. OnlyFans bikini shots and nude selfies pulled in more than $40,000 a month – enough to make medical school look optional. She dropped out, figuring the white coat would wait for her. On TikTok, she became the horny professor to a cult following of gen Z women, quoting Michel Foucault, Sigmund Freud and Marquis de Sade as she spoke about the color of her vagina, STD checks, treating bacterial vaginosis (BV) and inserting a menstrual sponge – and the delights of anal sex.
“I was making the sacrifice of potentially ruining my reputation to speak about sexual health and sexual topics, things 99% of the girls my age are also thinking about, in a non-stigmatized way,” Barey says. (Having a gynaecologist mom on speed dial helped.) By 2022, she had hit 750,000 TikTok followers and a $285,000 month on OnlyFans.
It didn’t last. Four years prior, in 2018, federal law made websites legally liable for hosting material linked to sex trafficking. Spooked by the law as well as mounting pressure from banks, advertisers and religious lobbying groups, Instagram and TikTok began issuing “community violations” for even the hint of sexual content. The word obscenity, long the legal standard for what counts as “too sexual”, remained deliberately undefined, giving them carte blanche to remove anything that made them nervous. Adult creators were shadowbanned, deleted and demonetized without warning. But they could not afford to leave the mainstream platforms altogether – that’s where they advertised their content and built fanbases.
Stella Barey. Photograph: Courtesy of Stella Barey
To survive, they developed their own internet survival codes, resorting to “algospeak” to circumvent obscenity guidances (“corn” for porn, “seggs” for sex, “accountant” for sex worker), VPNs and burner accounts to evade detection, and private Discords to swap intel. “We became outlaws,” Barey says. “You have no clue what is allowed or not allowed until you get hit with a violation. It’s all word of mouth.”
By the end of 2022, Barey had gone through 22 TikTok accounts, many with more than 600,000 followers, buying burner phones to start from scratch each time she got booted from the platform.
Severely limited by mainstream sites, sex workers sought out new havens for posting adult content, the most popular of which is OnlyFans. Whereas Pornhub and its peers monetize traffic through ads and streaming, OnlyFans monetizes relationships, letting creators sell directly to fans. By 2024, more than 4.6 million creators were pulling in $7.2bn from subscribers. But OnlyFans comes with its own set of problems. The platform is notorious for lacking creator-friendly tools. Unlike Instagram or TikTok, there is no “explore” page or discovery feed; the burden of finding an audience falls entirely on the performer, which means relentless self-promotion on the very sites that are so hostile to them. “People don’t realize that most of these girls don’t want to be doing social media – they just want to make porn,” Barey says.
Once fans do make it to a creator’s page, monthly subscriptions account for only a fraction of potential earnings. The real money comes from time-consuming manual engagement, such as selling custom videos, sexts and one-to-one messaging. At one point, 70% of Barey’s income came from these direct messages. “You can’t be spending your entire day making content, promoting it on socials, and also be on your account selling to fans 24/7,” she says. “It’s unsustainable.”
The idea for Hidden began in 2023, when a high school friend and Wharton business school grad approached Barey about co-founding a porn site. Barey set the terms: a platform designed by and for sex workers, built to promote passive income in an industry where constant performance is often the price of survival. The platform launched on 12 April with a sleek black interface and artful branding. Focused on amateur content, it recalls the Tumblr-era cam girl aesthetic – a period many creators look back on with fondness. (“Hidden” is a reference to the phone folder in which normies keep their nudes.)
After clicking “18+”, users land on a TikTok-style “For You” page that serves clips tailored to their taste. Scrolling through videos and photos, one can find anything from a girl-next-door-type smiling in her pyjamas to a performer gyrating on a lubed-up dildo. See someone you like? A quick swipe to that creator’s profile is usually where the first paywall appears: a small fee to unlock their feed, view explicit posts or send a message.
Barey is most excited about the ways Hidden, unlike OnlyFans, helps creators keep earning without constantly filming or messaging. The site’s algorithm promotes old videos as much as new ones, and each profile includes a built-in store where fans can buy clips and pay-per-view posts – content that creators have already made, now working for them in the background.
None of Hidden’s features are brand new, Barey admits. The scrolling feed is lifted from RedGifs; the store from ManyVids; and the chargeback protections, popular among creators for making it harder for customers to dispute charges and get refunds, from SextPanther. But they are consolidated on Hidden, which also takes the smallest cut in the industry (18% of creators’ earnings, compared with OnlyFans’ 20%).
For Leila Lewis, 28, a Philadelphia-based creator making over $30,000 a month on OnlyFans, the appeal was immediate. “Everyone is getting sick of the OnlyFans model. We’re exhausted and burnt out,” she says. After a consultation with Barey, she said Hidden felt like a return to the golden days – something that finally made her excited about the work again. “You can’t do fisting or pee content on OnlyFans,” which is the content her fans like best but is prohibited on the more skittish platform, Lewis explains. “That’s why I love Hidden, because they just let you do pretty much anything.”
Barey speaks at a Hidden event for creators in September. Photograph: Stella Barey
Barey oversees a 40-person software team, a product designer and six content moderators, and more features are on the way. Barey and her team are building a takedown bot to scan the internet for leaks and stolen content with a single click. She is experimenting with AI tools that would let fans request personalized clips generated from a performer’s likeness (for instance, “me in a red dress on a plane”), while safeguarding ownership of their likeness from sites that are already selling nonconsensual AI versions of them. Barey even wants Hidden to handle its own payments instead of relying on third-party payment processors – an unheard-of move in the adult industry that would cut out the middlemen that drive up fees. (If she could buy a bank outright, she says, creators might one day keep nearly all of what they earn.)
Ultimately, she insists it’s sex workers who will decide what comes next for Hidden. “I have a list of thousands of things, but if I’m hearing from the girls that they really want live streaming, I’m going to put that up at the top.” Meanwhile, her core crew of gen Z assistants – Drew, Chloe and Naomi, who once ran her OnlyFans and now act as her “angels” – weigh in on everything from marketing strategy to her sex tape’s final cut. Years of navigating porn sites and the minefield of social media have given them an instinct for what will resonate.
So far, Hidden has registered over 113,000 users who have spent on average $53 each, and has enrolled more than 2,100 active creators – most of them gen Z women.
At this moment public sentiment toward porn is souring. That’s in part due to the rise of “rage bait” porn, the kind of deliberately provocative content that first launched Barey into viral fame, when her TikTok about sleeping with her father’s fiftysomething best friend broke the internet. Gen Z creators such as Bonnie Blue and Lily Phillips, whose exhibitionist gangbang spectacles were engineered to be detested, thrive in an ecosystem warped by burnout, censorship, algorithmic whiplash, “slop” churn, and audiences with ever-shortening attention spans. For critics, they are proof the industry has lost its grip.
The backlash also mirrors cultural anxieties about sex, gender and power among gen Z. By age 13, most US teens have already encountered pornography, often by accident. Gen Z is the first cohort to grow up with porn not just available but ambient, algorithmically unavoidable.
The American Survey Center’s 2025 report found that nearly two-thirds of men under 25 now support making online pornography harder to access – a sharp increase from previous generations. This shift could be tied to growing discomfort with porn’s ubiquity, as well as a broader conservative turn among young people (as you can see in the “NoFap” and abstinence trends spreading across TikTok). At the same time, feminist critics, gen Z or otherwise, are calling out the damaging effects of some porn, from normalized choking to transactional “porn-script” sex bleeding into dating culture.
Interestingly, survey data shows gen Z reports less sexual activity than earlier generations, suggesting a more cautious cohort. And yet, new Kinsey Institute research finds gen Z to be the most kink-friendly generation on record.
Internet porn historian Noelle Perdue argues that the contradictions themselves are the story. “There is among younger generations this resentment towards the concept of mainstream pornography,” she says, “but they are also genuinely curious about their sexuality.” What gen Z is open to, she adds, is ethically produced porn that matches their sensitivities and desires. Recent Pornhub data also shows a broader cultural shift toward authenticity in porn: searches for “ethical porn” rose 92% in 2024, while “authentic sex” climbed 43%, meaning viewers are increasingly drawn to user-generated and amateur content over scripted, studio productions that can feel unrealistic.
“Ethical porn”, a recent buzzword in the industry, generally refers to erotic content that is transparently and legally produced, fairly paid and filmed with mutual pleasure in mind; feminist porn filmmakers such as Erika Lust are often held up as the gold standard. Hidden is, in that sense, as ethical as a platform can reasonably claim to be: its content is self-filmed by age-verified performers who own their work and keep most of what they earn. However, one can never be entirely sure what is or is not ethical without being in the room where the sex is happening.
Between her time on TikTok and an in-person erotic philosophy reading series she started last February, Barey has drawn in a wide circle of college-aged women. “I know there’s a rise in conservatism amongst gen Z, but I see this generation as the most accepting of porn of any generation yet,” she says. “They’re so supportive of sex work and understanding it as a legitimate job.”
By the time young people have worked out what they want – or do not want – from porn, it might be a moot point. The Heritage Foundation’s Project 2025 went so far as to openly call for pornography to be outlawed and its producers imprisoned, indicating a rightwing thirst for a national ban on the industry. A less existential but more immediate threat are new age-verification laws across the US and UK that require users to upload government IDs or biometric data before accessing porn. Lawmakers frame these bills as child-protection measures, but in practice they penalize the very platforms trying to comply and slash the income of sex workers – especially queer and trans creators already working on razor-thin margins. When Louisiana’s Act 440 went into effect in 2023, Pornhub reported an 80% drop in traffic from the state, while VPN searches spiked. As Perdue notes, minors will always find adult content; what these laws actually do is punish compliant platforms like Hidden, OnlyFans and Pornhub, and funnel users toward sketchier sites rife with pirated or nonconsensual material.
As of now, porn still accounts for more than a third of internet data transfers. Its future could depend on creators finally taking control of the industry they built. “Tech companies have a long history of establishing financial sustainability by hosting explicit content and then suddenly abandoning it,” Perdue says. “It would be amazing to have this pattern disrupted by a company that is truly aligned with sex workers, instead of just seeing adult content as a means to a financial end.”
Three dogs pant out of open windows as Barey’s pickup barrels down the freeway, away from the hiking trail. She will miss the beginning of her 2.30pm call, a meeting with her chief technology officer to review Hidden’s next software update. There is no signal on this stretch, so for a few more minutes she can remain in her favorite role – just another horny girl on the internet with a camera roll filled with nudes.
“Even though porn has been around forever, this version of online sex work is so brand new,” Barey says. In this way, Hidden may be less a product than a provocation – an argument that an industry dismissed as slop can still be reinvented.
Most Wall Street analysts remained divided on Tesla ‘s future after the electric vehicle maker’s shareholders voted to pass CEO Elon Musk’s nearly $1 trillion pay plan. The pay plan, introduced in September, garnered 75% support among voting shares. Results were announced at the company’s annual shareholders meeting on Thursday in Austin, Texas. Specifically, the pay package consists of 12 tranches of shares that Musk would be granted if Tesla hits certain milestones over the next decade. It would also increase Musk’s ownership and voting power in the company. Analysts highlighted pros and cons from the passage. On the former, they noted that it likely ensures Musk stays as Tesla’s chief exec. On the latter, however, they raised questions over how Musk will execute to reach the lofty goals set. Here’s what they said. UBS: sell rating, $247 Analyst Joseph Spak’s target implies about 44% downside from Tesla’s Thursday close. “This award clears the way for ~$1T in awards to Elon Musk if a series of 12 market capitalization and 12 operational milestones are met. We expected this to pass and based on conversations with investors believe it was widely expected to pass. Tesla announced that preliminary results show 75% voted for the proposal. This likely ensures Musk stays at Tesla, removing a potential overhang, and allows him to focus on his AI vision autonomous, humanoids towards achieving the goals.” Barclays: equal weight, $350 Barclays’ forecast corresponds to downside of around 22%. “While there were few surprises at today’s Tesla AGM, the event broadly reminded us of the excitement investors face ahead on Tesla’s growth prospects. Yet we believe the key question for the stock now remains on the execution path for Tesla’s growth initiatives.” Goldman Sachs: neutral, $400 Goldman Sachs’ target calls for 10% downside going forward. “Given that the 2025 CEO incentive award was preliminarily approved, we believe investor focus will now shift to the potential for Tesla to achieve these objectives, as well as on key milestones and datapoints including: 1) Tesla’s plan to remove safety observers from its robotaxis in Austin before year-end; 2) The timing for personal FSD to become unsupervised; 3) 4Q automotive deliveries (likely reported in early January); and 4) the unveil of Optimus V3 (which the company suggested on its 3Q earnings call could occur in late 1Q).” Bank of America: neutral, $471 Analyst Federico Merendi’s forecast is 6% above Tesla’s Thursday closing price. “The shareholder vote for Musk’s compensation package was overwhelmingly positive with a 75% approval rate. Potential investment in xAI was approved as well. All the other proposals in the proxy were voted on in-line with the suggestions from Tesla’s Board of Directors, excluding the proposal for the annual re-election of board members.” Baird: outperform, $548 Baird’s price target was approximately 23% higher than Tesla’s closing price on Thursday. “TSLA’s CEO performance award was approved by shareholders with 75% voting in favor. We expect this to be a modest positive for the stock, but more importantly, avoids what we believe would’ve been a more drastic negative impact. There are still questions that remain such as what will come of the legal dispute pertaining to the previous pay package. That said, we do not view this as an overhang for investors and believe the focus will now pivot to rolling out new products and TSLA’s AI ambitions.”
Simply sign up to the Technology sector myFT Digest — delivered directly to your inbox.
US tech stocks are on course for the worst week since President Donald Trump’s “liberation day” tariffs rocked global financial markets in April, as concerns about elevated valuations fuelled a $900bn rout in companies linked to the artificial intelligence boom.
The market value of eight of the most valuable AI-related stocks — including Nvidia, Meta, Palantir and Oracle — has fallen by $911bn since the end of last week.
Trading on Friday morning deepened tech investors’ losses for the week, with Nvidia falling 2.6 per cent in early trading. Other big tech companies whose share prices declined this week included Microsoft, Amazon and Broadcom.
The declines have left the tech-heavy Nasdaq Composite on course for a weekly loss of 4 per cent, its worst five-day run since the index fell 10 per cent after Trump launched his trade war with a flurry of tariff announcements in April.
Nvidia, the world’s most valuable company, has fallen the most in dollar terms over the week, losing more than $430bn in market capitalisation.
Nvidia’s chief executive Jensen Huang told the Financial Times this week that he expected China was ultimately “going to win the AI race” against the US.
He subsequently tried to row back on the comments, saying that China was “nanoseconds behind America in AI”, but the remarks came as the Silicon Valley chipmaker’s hopes were waning that the US government would allow it to sell a version of its latest Blackwell AI processor to Chinese customers.
Chinese competitors are already narrowing the technical lead held by OpenAI, Microsoft, Google, Meta and Anthropic, which are all making huge investments in AI infrastructure, much of it based on Nvidia’s chips.
This week’s debut of Beijing-based Moonshot AI’s new Kimi K2 Thinking model was hailed as the latest breakthrough by Chinese developers, with reports suggesting it cost less than $5mn to train.
“Is this another DeepSeek moment?” Thomas Wolf, co-founder of AI developer platform Hugging Face, said in a social media post about Kimi. The release of DeepSeek’s low-cost R1 model sparked a Wall Street panic in January that wiped $589bn from Nvidia’s market value in a single day.
Comments this week by OpenAI’s finance chief Sarah Friar that the $500bn AI group might look to the US government to provide a funding “backstop” also triggered speculation about its finances.
OpenAI’s chief executive Sam Altman sought to calm anxiety in a social media post on Thursday, saying: “We do not have or want government guarantees for OpenAI data centres.”
He predicted that OpenAI’s revenues would “grow to hundreds of billion[s] by 2030”, though that figure may fall below its AI infrastructure commitments, which he said totalled $1.4tn over the next eight years.
AHA 2025 officially begins this weekend in New Orleans, featuring key science simultaneously published across ACC’s JACC Journals, including JACC, JACC: CardioOncology, JACC: Advances, JACC: Case Reports, JACC: Heart Failure and JACC: Cardiovascular Imaging. Highlights include:
JACC
MESA, UK Biobank Data: Lp(a) and IL-6 Predict CHD, ASCVD Risk
Lipoprotein a (Lp[a]) and interleukin-6 (IL-6) are independent predictors of coronary heart disease (CHD), defined as myocardial infarction or resuscitated cardiac arrest, and when combined the two biomarkers identified patients at the highest risk of CHD, according to research in two primary prevention cohorts, the Multi-Ethnic Study of Atherosclerosis (MESA) (n=6,514) and the UK Biobank (n=26,574), published in JACC.
The hazard ratios (HRs) for CHD were 1.13 and 1.22 for Lp(a) and IL-6, respectively, in MESA and 1.11 and 1.19 in UK Biobank cohort. Similar results were found for atherosclerotic cardiovascular disease and peripheral vascular disease. The analysis conducted by Harpreet Bhatia, MD, MAS, FACC, et al., found the HRs for CHD for LP(a) plus IL-6 were 1.72 in MESA and 1.39 in the UK Biobank cohort.
“The findings of our study support the potential incorporation of IL-6 in a primary prevention setting in addition to Lp(a); at the minimum, IL-6 may be considered a risk enhancer as it identifies individuals with increased cardiovascular risk independent of traditional risk factors, hsCRP and Lp(a),” write Bhatia and colleagues.
ECV’s Prognostic Ability in ATTR
Extracellular volume (ECV) identified on cardiovascular magnetic resonance (CMR) imaging provides a quantitative framework for staging and therapeutic planning in transthyretin amyloidosis (ATTR) by directly quantifying myocardial amyloid load and defining reproducible thresholds to stratify burden and refine risk prediction, according to findings published in JACC. Study authors Awais Sheikh, MBChB, et al., analyzed all-cause mortality in 1,541 patients undergoing CMR for ATTR, classified as TTR-variant carriers, extra-cardiac ATTR, early stage ATTR-CM or overt ATTR-CM. At a median 2.8-year follow-up, results showed the ECV had a strong prognostic ability with <30% excluded and ≥40% confirmed cardiac involvement, independently predicting mortality across categories: <30% none; 30-39% mild, 40-49% moderate, 50-59% moderate-severe and ≥60% severe (HR, 1.22 per 10% increase; p<0.001). “ECV outperforms traditional staging tools,” write the authors. “These data establish ECV as a quantitative biomarker that bridges clinical practice and trial design, translating amyloid biology into actionable measurement.”
JACC: CardioOncology
ALLSTAR: Gamification For Breast, Prostate Cancer Survivors
Gamification increased physical activity among Black and Hispanic cancer survivors compared with attention control alone, according to research published in JACC: CardioOncology. The clinical trial randomized 150 Black and Hispanic breast and prostate cancer survivors treated with cardiotoxic therapy and with at least one risk factor for cardiovascular disease (mean age, 64 years; 81% women; 64% Black; 35% Hispanic) to either attention control (n=76) or gamification (n=74), both featuring a wearable device to track daily steps and self-determined daily step goals. Gamification participants were awarded points each week, with points lost if they did not reach their goals. Those who kept most of their points each week were moved up a level, with participants at the highest level receiving a trophy at the end of the intervention.
Results showed that gamification participants had a greater mean daily step average and spent more time on weekly moderate-vigorous physical activity than control in both the 24-week intervention period (+759 steps, p=0.007; +16 minutes, p=0.010) and 12-week follow-up period (+581 steps, p=0.070; +11 minutes, p=0.048). Noting the greater health barriers historically marginalized communities can face within health systems, “by leveraging principles from behavioral economics, including precommitment, status quo bias, the endowment effect, goal gradients, the fresh start effect, and social accountability, the intervention addresses common barriers to behavior change,” write study authors Alexander C. Fanaroff, MD,et al.
JACC: Advances
FH Clinical, Genetic Signs in CAD Risk Assessment
Clinical and genetic signatures of familial hypercholesterolemia (FH) were useful in determining the risk of coronary artery disease (CAD) among patients with severe hypercholesterolemia, beyond the incorporation of LDL-C alone, according to original research published in JACC: Advances. In the retrospective study, investigators Hayato Tada, MD, et al., evaluated data from 1,273 patients (mean age, 49; 50% women) with LDL-C ≥180 mg/dL. Results showed that at a median follow-up of 12.4 years, 144 CAD events occurred. Patients with either an FH-variant (n=298) or clinical signs of FH (n=199) had a higher risk of developing CAD events (HR, 1.44, p=0.007; HR, 2.27; p<0.001, respectively). Those participants with both FH-variant and clinical signs (n=548) had a five-fold higher risk for CAD (p<0.001) after adjustment for known risk factors and LDL-C year score. “We strongly suggest assessing these clinical and genetic signatures of FH and then treating them differently based on their risk strata,” write the investigators. “These procedures may be reasonable not only for accurate diagnosis but also for personalized medicine.”
JACC: Case Reports
Quality Improvement Project: Optimizing Telemetry Use
In a quality improvement project presented in JACC: Case Reports, Hera Jamal, DO, et al., outline a low-cost, resident-led initiative that improved telemetry ordering practice at a tertiary care center. The initiative offered an educational lecture to internal medicine residents and teaching hospitalists regarding AHA best practice guidelines and posted physical reminders near telemetry stations. Results from 156 preintervention admissions and 110 postintervention showed an increase from 61% to 74% on the proportion of appropriate telemetry orders upon admission (p=0.03), although no significant change was observed at 48 hours (38% vs. 39%, p=1.00). Jamal and colleagues write on their initiative: “This is important given the established safety of discontinuing unnecessary telemetry in nonintensive care settings and the burden that continued use places on both patients and healthcare staff.”
JACC: Heart Failure
Survival in Patients with Advanced Symptoms of oHCM Undergoing Surgical Myectomy
Patients with advanced symptoms of obstructive hypertrophic cardiomyopathy (oHCM) undergoing surgery myectomy (SM) had worse long-term survival compared to those with lesser symptoms – despite earlier time to surgery from initial presentation, according to research published in JACC: Heart Failure. The observational study stratified 3,546 oHCM patients by NYHA class upon initial presentation, and at a mean follow-up of 12 years recorded 698 composite events, including 661 deaths, 11 cardiac transplants and 51 appropriate ICD discharges. Even with time to initial surgery shorter in the higher classes, the number of events grew from Class I (70 events, 19% of patients) to II (232, 16%) to III (396, 23%; p<0.001). “Patients undergoing SM at an earlier symptomatic stage demonstrate superior long-term outcomes, underscoring the importance of timely referral before advanced disease occurs,” write study authors Shada Jadam, MD, et al.
JACC: Cardiovascular Imaging
Efficacy of Treadmill Stress Echocardiography in HCM
Treadmill stress echocardiography (TSE) has diagnostic and prognostic value in patients with asymptomatic HCM, according to a study by Sana Sultana, MD, et al., published in JACC: Cardiovascular Imaging. Of 1,299 NYHA Class I asymptomatic HCM patients undergoing TSE, 22% were found to have latent dynamic left ventricular outflow tract obstruction (LVOTO), only observed during the testing. Furthermore, 37% failed to achieve age-sex predicted metabolic equivalents (ASP-METs) ≥85%. At a mean follow-up of 12.1 years, ASP-METs <85%, undetected without TSE, was associated with increased mortality (17% vs. 12%) as well as lower survival rates at one, 10, 15 and 20 years. “The findings challenge the reliability of patient-reported symptom status and underscore the limitations of subjective self-reported assessments in clinical decision-making,” write the authors. “Whenever feasible, it is important to objectively assess functional capacity and record latent dynamic LVOTO using exercise echocardiography.”
Click here to read all the articles simultaneously published in the JACC Journals.
The whole lifecycle of O. tholozani ticks was completed three times in consecutive experiments using the artificial feeding system in this study. In total, 5788 ticks of different life stages were tested, including 1605 larvae, 601 first-stage, 364 second-stage, and 269 third-stage nymphs, 30 females and 34 males, and 2885 eggs. The first cycle was completed in 351 days (standard error, SE = 17.35), the second cycle in 546 days (SE = 13.8), and the third cycle in 238 days (SE = 6.15) (Additional file 1: Additional Table 1). The difference between the lengths of the cycles was due to a technical reason related to the feeding system. In the second cycle, we changed the dog from which we produced the lipid hair extract since the dog used in the first cycle was no longer available. The time between the end of one lifecycle and the next, i.e., from the first oviposition of a certain tick generation until the feeding of the first batch of larvae of the next generation, ranged between 45 and 63 days [average (AV): 51.5, SE = 5.7].
Table 1 Borrelia persica acquisition by third-stage nymphs at defined time points post infection. Presence of B. persica DNA and quantification of B. persica loads in tick organs at different time points post infection
Reproductive capacity parameters, engorgement, and molting rates were measured for the different tick generations (Additional file 1, Table 1). The average number of eggs laid per female was 56.8, 183.7, 184.2, and 189.0, for the parent stock, first, second, and third generation, respectively. The percentage of larvae hatching was 60.2%, 91%, and 94% for the first, second, and third generations, respectively. The period of days between the female tick engorgement and oviposition was 44.2, 21.2, 28.7, and 26.4 for the parent stock, first, second, and third generation, respectively. Average percentage rates of engorgement were 81.9% (SE = 5.1), 58.0% (SE = 10.3), and 68.2% (SE = 2.3) for the first, second, and third generation, respectively; average molting rates were 56.9% (SE = 11), 85.6% (SE = 3.5), and 85.5% (SE = 3.9), for the first, second, and third generation, respectively. The average mortality rates of ticks in all three lifecycles were 7.1% (SE = 3.1), 4.5% (SE = 2.3), 2.4% (SE = 2.4), 12.2% (SE = 3.9), and 0%, for larvae, first-, second-, and third-stage nymph, and adults, respectively. Females were fed only once in their adult life prior to oviposition. Females, except for those of the parent stock and of the third generation, which were analyzed after the first oviposition, were allowed to continue with additional cycles of folliculogenesis for up to 5 months. Out of nine females belonging to the first and second generations, two females had four ovipositions, two had three ovipositions, three had two ovipositions, and two had only one oviposition. The other tick life stages were also fed only once in each life stage prior to molting, except for an additional feeding attempt that was carried out for those ticks that refused to feed in the first try, which was done a week after the first one. This part of the study showed that the whole lifecycle of O. tholozani can be reproduced under laboratory conditions exclusively by artificial feeding with no need for feeding on experimental animal hosts.
Borrelia persica survival in bovine blood
Follow-up of spirochete motility in the bovine blood was continued over 8 days. Motile spirochetes were observed until 96 h post-inoculation and not thereafter, supporting the idea that B. persica requires being in a tick or vertebrate host to continue its long-term survival.
Borrelia persica acquisition by third-stage nymphs at defined time points post-infection
Ticks raised in the laboratory, which originated from the parent stock, were brought from the egg stage to the third-stage nymphs by artificial feeding on uninfected bovine blood. At that stage, 42 third-stage nymphs were fed with heparinized bovine blood containing 10⁶ B. persica spirochetes per ml, and were divided into five groups according to the time point of analysis post-infection. The groups included ticks examined by PCR for B. persica in their guts and salivary glands, 1, 2, 3, 4, and 5 weeks after infection (Table 1). All 16 (100%) of the third-stage nymphs tested at 1 and 2 weeks after infection were found to be positive for B. persica by PCR only in their guts. At week 3, seven out of eight (87.5%) ticks tested were positive for B. persica in their guts, and one was negative in both guts and salivary glands. In the fourth week, two out of eight third-stage nymphs had molted to females and one to a male, while the remaining five stayed as third-stage nymphs. Six (75%) of these ticks were infected, including five third-stage nymphs and one female. When relating to all tick stages, B. persica DNA was found in the guts of four ticks and salivary glands from three. Two third-stage nymphs were positive in both their guts and salivary glands, two were positive only in their guts, and one was positive only in its salivary gland. The positive female was infected in its guts and its salivary glands. Of the ten ticks evaluated after 5 weeks of infection, three nymphs had molted to fourth-stage nymphs, and one had molted to a male. Infection was detected in five ticks, including five guts and one salivary gland. Three third-stage nymphs were positive only in their guts, one fourth-stage nymph was infected in its gut, and another fourth-stage nymph was infected in its gut and salivary glands. The male was negative.
In total, out of the 42 infected third-stage nymphs in all the groups, 34 (81%) ticks (31 third-stage nymphs, two fourth-stage nymphs, and one female) were positive for B. persica by the flaB PCR, including 33 guts and 5 salivary glands, with infection of salivary glands detected only in ticks analyzed at week 4 and 5 after infection (Table 1). The average B. persica loads in the guts, expressed as Borrelia flaB copies per tick 16SrRNA gene copy, ranged between 4.7 and 67.4 in the different weeks (Table 1). The copy number of B. persica in the guts at week 1 (AV = 67.4) was significantly higher than at week 3 (AV = 5.5) and 4 (AV = 48.7) (Kruskal–Wallis test H = 8.25, df = 3, P = 0.041; Table 1). However, when comparing the B. persica loads in the guts of ticks at week 5 of infection (AV = 28.5) with those found at the other time points, only a nonsignificant trend was observed (Kruskal–Wallis test H = 8.79, df = 4, P = 0.066). The average B. persica loads in the salivary glands at weeks 4 and 5 post-infection were 7.9 flaB-copies per tick 16SrRNA gene copies (SE = 7.5), and in the guts analyzed at week 4 and 5 it was 37.46 (SE = 23.43) with no significant differences between loads in the salivary glands and in the guts with those found at the other time points (Mann–Whitney U-test: U(13) = 16.00, Z = −0.309, P = 0.825). The results described in this section portray the dynamics of B. persica within experimentally infected ticks, with movement of infection from the tick gut to the salivary glands noted from 4 weeks post-infection.
Transstadial and transovarial transmission of B. persica
Transstadial transmission was analyzed by infecting 161 larvae with 1.2 × 10⁷ B. persica spirochetes per ml of blood and testing 20 individuals from each tick life stage, e.g., larvae, first-, second-, and third-stage nymphs, and adult ticks (ten females and ten males). Larvae were tested 1 week after feeding on infected blood, whereas first-, second-, and third-stage nymphs were analyzed a week after molting; adults were analyzed 2 weeks after molting. Except for larvae, which were analyzed after engorgement, the remaining stages were analyzed unengorged. The rate of infection decreased from 100% (20/20, 95% CI 83.16–100%) in larvae to 55% (11/20, 95% CI 31.53–76.94%), 20% (4/20, 95% CI 5.73–43.66%), and 25% (5/20, 95% CI 8.66–49.10%) in first-, second- and third-stage nymphs, respectively. Adult ticks showed an infection rate of 20% (4/20, three male ticks and one female tick).
Average B. persica loads in ticks were 243.8 (SE = 42.1), 21.6 (SE = 7.7), 38.1 (SE = 4.9), and 61.2 (SE = 21.9) flaB-copies per tick 16SrRNA gene copies in larvae and first-, second-, and third-stage nymphs, respectively. Average B. persica loads in adults were 199.2 flaB-copies per tick 16SrRNA gene (SE = 25.7) [244.4 (SE = 50.2) and 63.5 in males and females, respectively] (Table 2). There was a significant difference in parasite load between tick stages (Kruskal–Wallis test H = 20.18, df = 4, P < 0.001). Post hoc pairwise comparisons using Dunn’s test with Bonferroni correction indicated that the larvae had a significantly higher parasite load than first-stage nymphs (Z = 18.344, P = 0.002), but no other pairwise differences were significant.
Table 2 Transstadial transmission of Borrelia persica infected at the larval stage
Transovarial transmission was tested by mating eight infected females with eight uninfected males and analyzing 20 eggs and 20 larvae laid and hatched from each of the eight potentially infected females. Because only eight eggs were analyzed from one of the females, a total of 148 eggs and 160 larvae were analyzed. The number of days from feeding and engorgement of the females to the beginning of the first oviposition ranged between 21 and 44 days (average = 26.6, SE = 2.6), and the number of eggs laid per female ranged between 70 and 190 (average = 129.6, SE = 13). Larval hatching rate ranged between 67% and 85% (average = 77%, SE = 2.3). Dissection and molecular analysis of the females and males, which were infected at third-stage nymph, was done after the first oviposition between 12 and 16 weeks post-infection, and real-time PCR targeting the Borrelia flaB gene [21] was done on the salivary glands, guts, and gonads. Of the eight females, five were infected (62.5%). Three of them were positive in their guts, five in their salivary glands, and three in their ovaries. Of those, three females were positive in their guts, salivary glands, and ovaries, and two females only in their salivary glands. All eight uninfected males that were mated with the females were negative for B. persica DNA. Regarding B. persica infection in eggs, 4 of 148 eggs (2.7%) analyzed were positive by PCR. The positive eggs belonged to three positive females. Considering only the eggs laid by infected females (n = 5), the rate of egg infection was 4 of 100 (4%), ranging from 5% to 10% per infected female (Table 3). Regarding B. persica infection in larvae that hatched from the eggs of the experimentally infected ticks, 2 of 160 larvae tested were positive (1.3%), and both belonged to larvae hatched from eggs laid by one positive female. Considering only larvae derived from positive females, the rate of larvae infection was 2 of 100 (2%) and 10% per the single positive female, which gave rise to the infected larvae (Table 3).
Table 3 Transovarial transmission of Borrelia persica to eggs and larvae from infected Ornithodoros tholozani female ticks
Five uninfected females were paired with five uninfected males as controls. The period of days from feeding and engorgement of the adults until the beginning of the first oviposition ranged between 21 and 29 days (AV = 26.4, SE = 1.6), and the number of eggs laid per female ranged between 84 and 263 (AV = 189, SE = 29.3). The larvae hatching rate ranged between 74% and 94% (AV = 88%, SE = 3.9%). The hatching rate of the larvae originating from the control group of uninfected females (88%) was close to being significantly higher than the hatching rate of the larvae derived from B. persica-infected females (75%) (Welch’s t-test, t(5.477) = −2.49, P = 0.051).
The outcomes of the experiments on transstadial and transovarial transmission indicated that both modes of B. persica transmission could be demonstrated by experimental infection using artificial tick feeding, and that the rate of transovarial transmission was considerably lower than that of its transstadial counterpart.
Horizontal transmission of B. persica between ticks by hyperparasitism
Twenty-two third-stage nymphs infected with heparinized bovine blood containing 10⁶ spirochetes per ml were divided into two groups. In the first group, each infected third-stage nymph was individually paired immediately after infection with an uninfected unfed male for 3 h to study possible direct acquisition of B. persica by the males from the nymphs. Biting of nymphs and engorgement of the males with laceration of nymph dorsal plates caused by male mouthparts were observed in three nymph-male pairs (Figs. 2A, B). Five weeks after pairing, three out of ten potentially infected and engorged nymphs molted into fourth-stage nymphs, one molted into a male, and the rest remained as third-stage nymphs. Molecular screening by real-time PCR targeting the flaB gene [21] revealed that five out of these ten ticks were positive for B. persica, including three third-stage and one fourth-stage nymph. All positive ticks showed infection in the guts, and one fourth-stage nymph showed B. persica DNA also in the salivary glands. Out of ten uninfected, unfed males before the pairing, three showed blood in their gut during their dissection, which was performed 5 weeks after pairing with the nymphs. These three males had been paired with third-stage nymphs, which showed laceration lesions in their dorsal plates after the pairing, and two of these males were positive for B. persica flaB DNA in their guts at week 5 of the experiment. These two positive males had been paired 5 weeks earlier with two nymphs that were PCR-positive, demonstrating transmission of B. persica by hyperparasitism. The third engorged male, which was negative by PCR, was paired with a nymph that turned out to be PCR-negative for B. persica when tested.
Fig. 2
Laceration lesion in a nymphal dorsal plate (black arrow) immediately after hyperparasitism by a male (A), and healed laceration lesion in the dorsal plate (black arrow) of the same nymph 5 weeks after the breach caused by a male (B)
In the second group, 12 uninfected, engorged third-stage nymphs were paired individually immediately after feeding with 12 potentially infected unfed males for 3 h. After the pairing, five nymphs showed laceration of the dorsal plates and one on the ventral exoskeleton, caused by male mouthparts. Five weeks after pairing, 3 out of 12 engorged uninfected third-stage nymphs molted into males, 1 into a female and 1 into a fourth-stage nymph, and the remaining 7 remained as third-stage nymphs. Borrelia-specific flaB DNA real-time PCR screening revealed that all of the previously engorged uninfected ticks were negative for B. persica DNA, while two of the potentially infected males, which fed on the uninfected ticks, harbored B. persica DNA in their guts. In essence, B. persica was not transmitted during hyperparasitism from infected males to engorged nymphs. The results of this part of the study indicate that B. persica can be transmitted between O. tholozani ticks by feeding directly on each other.
Molecular analysis
All positive samples were sequenced. In total, 101 flaB DNA sequences of B. persica were obtained. From these, 38 sequences were from 38 ticks infected during the transstadial infection experiment, 38 sequences were from positive organs of 34 ticks infected weekly during 5 weeks, 17 sequences were from positive organs and offspring of infected females during the transovarial infection experiment, and 8 sequences were from organs of infected nymphs and males that participated in the hyperparasitism infection experiment. All DNA sequences were 100% identical to each other. Four DNA sequences were submitted to GenBank (accession nos. MW284983-86). Uninfected control ticks of all life stages and all DNA extractions from negative controls were PCR-negative for B. persica.
Pre-agreed parametric triggers reached for full payout based on storm’s central pressure and path
WASHINGTON, NOVEMBER 7, 2025 — Following Hurricane Melissa, the Government of Jamaica will receive a full payout of $150 million under its catastrophe insurance coverage with the World Bank, backed by a catastrophe bond issued in 2024 by the World Bank (International Bank for Reconstruction and Development, or IBRD, AAA/Aaa).
Analysis carried out by third-party calculation agent, AIR Worldwide Corporation, concluded that Hurricane Melissa reached pre-agreed parametric triggers qualifying for a full redemption of the World Bank Catastrophe Bond, which offers Jamaica financial protection against specified natural disasters. The analysis was based on the storm’s central pressure and path, as reported by the National Hurricane Center.
As one of the most exposed countries to natural disasters, Jamaica has a well-developed disaster risk financing strategy. Jamaica initially received insurance coverage against named storm events from the World Bank through a World Bank-issued catastrophe bond in 2021 and three years later renewed its coverage with the 2024 catastrophe bond. Further information related to the bond’s structure can be found here.
Catastrophe bonds transfer financial risks from natural disasters to global capital markets and are one of many financial instruments available to support countries in the aftermath of natural disasters such as hurricanes and earthquakes. Catastrophe bonds are recognized by credit rating agencies as beneficial to countries and form part of their disaster risk management toolkit.
“Our thoughts are with the people of Jamaica as they recover and rebuild from this tragedy. Jamaica’s comprehensive disaster risk management strategy and proactive approach serve as a model for countries facing similar threats and seeking to strengthen their financial resilience to natural disasters,”said Jorge Familiar, World Bank Vice President and Treasurer.“The payout underscores the role of catastrophe bonds in effective risk management strategies and their efficiency in transferring disaster risks to capital markets.”
In addition to the forthcoming full payout of the catastrophe bond, a broad package of World Bank Group assistance is ready to be mobilized to support Jamaica — combining quick-disbursing emergency finance, the redeployment of existing project funds, and targeted private-sector support through the International Finance Corporation, the World Bank Group’s private sector development arm.
“As Jamaica confronts the aftermath of Hurricane Melissa, its strong commitment to preparedness is proving its worth — allowing the country to move swiftly from relief to reconstruction and to use this moment not just to rebuild, but to leapfrog toward more resilient infrastructure”, said Susana Cordeiro Guerra, World Bank Vice President for Latin America and the Caribbean. “The World Bank Group stands with the Government and people of Jamaica to help rebuild stronger, restore livelihoods, and set a new benchmark for resilience across the Caribbean.”
Catastrophe insurance backed by catastrophe bonds are part of the World Bank’s Crisis Preparedness and Response toolkit which provides developing countries with an innovative suite of tools to better respond to crises and prepare for future shocks. This includes fast access to cash for emergency response, expanded catastrophe insurance and the option to pause debt service payments in the aftermath of a natural disaster.
Contact:
World Bank Media Relations: +1 (202) 473 7660, press@worldbank.org
“On behalf of Renault Group, I would like to pay tribute to Louis Schweitzer, a visionary and bold leader whose commitment and high standards contributed to the modernization and internationalization of the Group with iconic launches such as Twingo and Megane, the acquisition of Dacia, and the creation of the Renault-Nissan strategic Alliance. He also championed a humanistic vision of business, combining economic performance with social responsibility. We extend our sincere condolences to his family and loved ones.” saidJean-Dominique Senard, Chairman of Renault Group.
“It is with deep sadness that I learned of the passing of Louis Schweitzer, who led our company with vision and determination. Under his leadership, Renault Group underwent major transformations. The company’s evolution bears the mark of the strategic vision he instilled. On behalf of all our employees, I would like to pay tribute to the memory of an exceptional man and offer our sincere condolences to his family and loved ones.” said François Provost, CEO of Renault Group.
Louis Schweitzer was born on July 8, 1942 in Geneva. He studied at Sciences Po Paris, then at the Ecole nationale d’administration (ENA), from which he graduated in 1970. He began his career as an inspector of finance, a prestigious position in the French senior civil service.
He became chief of staff to Laurent Fabius who was then Prime Minister, from 1984 to 1986. This political experience placed him at the heart of the French executive power and opened the doors to the industrial world.
In 1986, he joined Renault Group, first as Chief Financial Officer, then as Deputy Chief Executive Officer.
In 1992, he became Chairman and Chief Executive Officer of the Group, succeeding Raymond Levy.
Under his leadership, Renault Group was privatized in 1996, and he orchestrated the strategic Alliance with Nissan in 1999, a major turning point for the Group. Louis Schweitzer was also the architect of the Dacia takeover in 1999, with a bold strategic vision: to create a reliable car at a very low cost for emerging markets.
In 2005, he stepped down as Chairman of the Group.
Louis Schweitzer is Grand Cross of the Legion of Honour.
The primary goal of this study was to predict muscle response to FES therapy from baseline sEMG signals for individuals with cervical SCI. sEMG has the potential to be integrated into point-of-care tools to provide biomarkers for clinical decision support and enable precision rehabilitation approaches. By knowing which muscles are likely to benefit, therapists can make informed decisions into the use of the limited therapy time available. We evaluated classification models trained on clinical variables alone, sEMG features alone, and combinations of both. Leave-one-participant-out (LOPO) cross validation was used to assess robustness across participants.
Our findings indicate that sEMG feature sets, particularly the FWD feature set (slope sign changes, mean and median frequencies, and M2), consistently outperform models using clinical variables alone. The FWD feature set combined with a random forest classifier achieved the highest performance across multiple metrics, including MCC (0.41), accuracy (0.76), and macro F1 (0.68). Combining clinical variables with sEMG features did not improve model performance, underscoring the unique predictive values of sEMG features. Additionally, training models on specific AIS subgroups (motor complete and motor incomplete) improved performance, particularly in AIS A-B, compared to models trained on the entire dataset.
Limitations of clinical variables alone
Although the progression of AIS grade from A to D correlates with an increasing percentage of responders (Fig. 1), AIS was not identified as the single most important predictor in logistic regression with forward feature selection. Resulting clinical variables also include NLI, Proximity, and Distance. The logistic regression model trained on these variables demonstrated moderate performance (Table 2), with MCC, macro F1 score, accuracy, precision, and TNR all above expected values by chance. Recall (0.21) from the model is below the expected chance level (0.33), indicating the difficulty in correctly identify true responders. This low recall suggests that some muscles that could benefit from the FES therapy might be overlooked.
When evaluating the logistic regression model’s performance per AIS grade (Fig. 3), the limitations of relying solely on clinical variables become more evident, particularly for AIS A, B, and C, where MCC, precision, and recall are all zero. The poor performance on AIS B is especially notable, as it includes a larger number of participants compared to AIS A and C. Training the model on AIS subgroups slightly improved performance for the motor incomplete group (AIS C-D) but did not enhance results for motor complete group (AIS A-B). In fact, the model predicted all AIS A-B muscles to be non-responders, resulting in 14 (25%) false negatives. With no positive predictions, MCC, precision, and recall remained at zero. By relying on this model with only clinical variables, no muscles from patients with AIS A or B would be selected for FES therapy—or potentially for other treatment as well.
Prediction with sEMG features
In contrast, models using sEMG features alone consistently outperformed those based on solely clinical variables, particularly the FWD set with RF. The FWD set achieved the highest values in all metrics, including MCC (0.41), accuracy (0.76), and macro F1 (0.68), though it did not achieve the highest TNR due to the class imbalance. When trained on all participants, FWD set was the only one to obtain above-chance performance across all AIS grades (Fig. 2), suggesting that it captures essential sEMG characteristics relevant to predicting the response to the FES therapy.
The FWD feature set (SSC, M2, mean and median frequencies) captures diverse aspects of motor unit firing and recruitment patterns by integrating both time- and frequency-domain information. This breadth is likely key to its strong predictive performance, as no single feature alone can fully characterize the neuromuscular output, especially after SCI. For example, M2, a time-domain feature characterizing frequency-domain behavior, quantifies the temporal variability of the signal by computing the squared difference between consecutive time samples. Higher M2 values may reflect more abrupt signal changes and complex activation patterns, which contributes towards predicting positive (responder) class in the SHAP analysis (Appendix C). SSC is related to the frequency of slope changes and indicative of motor unit firing irregularity. Mean and median frequencies, which summarize the distribution of spectral power, also contributed but with greater variability across samples. These findings highlight the physiological relevance of the selected features and support the use of diverse and broad feature sets to improve prediction robustness in the heterogeneous SCI population.
Moreover, Fig. 3 shows that training models specifically on motor completeness subgroups (AIS A-B vs. C-D) leads to further performance improvement for RF with the FWD set. This subgroup-specific training enhances MCC, accuracy, macro F1, precision, and TNR, particularly for AIS A-B, highlighting the advantages of tailoring models to motor completeness levels. This approach appears to capture more distinct sEMG patterns within each subgroup, allowing for improved classification performance. Given the small dataset, we did not separate subgroups further by individual AIS grade, though this may provide further improvements with a larger sample size.
Impact of imbalanced dataset on recall (TPR) and TNR
The consistently high TNR across feature sets and models can be attributed to the dataset’s class imbalance, where non-responders are more frequent than responders. This imbalance leads to models that are effective in identifying non-responders but struggle with recall, particularly in AIS A and C, where MCC scores were close to or below zero (Fig. 2) for most models. This low recall indicates that while models perform well in identifying non-responders, they may overlook true responders, limiting the practical utility. Results from Fig. 4 suggests that subgroup-specific training partially alleviates this issue by improving recall within more homogeneous groups, especially in the motor incomplete subgroup, where the model appears better suited to capturing true responder characteristics.
Variability across muscles and participants
Precision variability across participants from RF on the FWD set (Fig. 5) underscores the challenge of achieving consistent responder classification. Precision had the highest variability, followed by MCC, TNR and recall. This variability suggests that while the FWD feature set with RF generally performs well, individual differences in muscle response create inconsistencies in predicting true responders. Notably, the high variability in precision and MCC indicates that certain participants’ sEMG signals are easier to classify than others, possibly due to differences in baseline after SCI or other individual-specific factors. The results in Fig. 5 should however be interpreted with caution, considering the low number of muscles per participant that may impact the robustness of the metrics in this portion of the analysis.
We recognize that differences in muscle type and size may influence sEMG signal characteristic and classification outcomes. Although this variability was not stratified in the current analysis, future studies with more data may explore muscle-specific stratification approaches.
In the context of existing literature
There exists intensive literature in predicting functional recovery after SCI [23, 24, 49]. However, to the best of our knowledge, no prior study exists to provide prediction of muscle response to FES therapy, a promising intervention for restoring motor function. Our study is the first attempt to address the gap in the literature and focuses on muscle-level prediction of FES therapy outcomes. While clinical variables such as AIS grade and NLI provide general prognostic information [22,23,24], our findings indicate that baseline sEMG features, particularly the FWD feature set, are more effective for predicting responses to FES therapy. Unlike available clinical information, sEMG captures neuromuscular activation patterns that reflects residual motor connectivity. The FWD feature set combined with a random forest classifier consistently outperformed models using clinical variables alone, suggesting that sEMG features capture unique, functionally relevant information at the muscle level. Notably, combining clinical variables with sEMG features did not enhance model performance, reinforcing the unique predictive value of sEMG alone.
Prior studies have shown that stratifying SCI patients into specific subgroups based on motor completeness or baseline neurological impairment can improve prognostic accuracy, such as the Unbiased Recursive Partitioning regression with Conditional Inference Trees (URP-CTREE) model [50]. The URP-CTREE model has been used to stratify patients with acute traumatic SCI into homogeneous subgroups to optimize recovery predictions and enhance the design of clinical trials. We explored training separate models for motor complete (AIS A-B) and motor incomplete (AIS C-D) groups. Our findings similarly suggest that subgroup-specific training improves classification performance, particularly in identifying responders, by allowing models to capture subgroup-specific sEMG patterns related to motor completeness.
Choice of MMT as an outcome measure
MMT is a practical and commonly used clinical assessment for muscle strength and was used as the primary outcome measure. Because of its simplicity and accessibility in a clinical setting, MMT was administered before each FES therapy session to track the target muscle strength, without the need for additional scheduling. To ensure consistency across sessions and raters (therapists), we implemented standardized protocols from ISNCSCI and GRASSP.
While not feasible for frequent longitudinal data collection, other modalities such as imaging or motor evoked potentials could provide quantitative insights into muscle structure and corticospinal connectivity in response to FES therapy. Combining these tools with baseline sEMG could offer a comprehensive evaluation framework, capturing both functional and structural aspects of the recovery. A multi-modal approach with measurements before and after FES therapy or at multiple timepoints throughout the therapy cycle could help refine responder identifications and provide more accurate evaluation, enabling more robust predictive model development.
Limitations and future directions
In this section, we discuss several limitations that should be considered when interpreting the findings and future research directions.
Expanding dataset diversity and demographics
First, there was only one female participant (less than 6%), which does not reflect the proportion seen in the SCI population [51] and restricts the study’s generalizability across sex. A more balanced sample would provide a clearer understanding of potential differences in response to FES therapy between male and female participants.
The dataset also has a higher number of non-responders (67%) than responders. This imbalance likely contributed to the high true negative rate (TNR) observed across models, as well as the relatively low recall, indicating that the models may be better at identifying non-responders than true responders. While we attempted to compensate for this effect by evaluating models with robust metrics such as MCC and F1 score, future studies with more balanced responder and non-responder groups would help validate these findings and improve the model’s sensitivity to true responders.
Overrepresentations of AIS D injuries (45%) and C3–C4 level of injury (74%) are also observed. AIS D often reflects more treatment options and better prognostics. Individuals with motor complete cases (AIS A or B) often face lower expected recovery potential. Our results show that clinical information alone typically predicts all muscles in AIS A cases to be non-responders, effectively closing the door to FES therapy for this subgroup. This exclusion is problematic, as AIS A patients represent a group in dire need of interventions. These imbalances hinder the generalizability of findings to the broader SCI population, which exhibits great demographic and clinical diversity.
While we obtained promising results by training models on specific AIS subgroups, the relatively small sample size prevented further stratification by individual AIS grades. Future studies with larger datasets may benefit from more granular subgroup analysis to capture subtle differences in muscle response within each AIS grade, potentially enhancing predictive accuracy.
Although our primary goal in this study was to explore generalizable predictive patterns in baseline sEMG across target muscles for FES therapy, a larger dataset would allow for investigation of the impact from muscle anatomical variability, including muscle type (e.g., biarticular vs uniarticular) and size. Along with sex and other person- and muscle-level variables, muscle type and size could be explored as predictive variables.
Beyond binary classification
In this study, binary classification results were used to evaluate model performance. With a larger dataset, future work could move beyond binary prediction to estimate changes in MMT scores directly. Predicting both the magnitude and the timing of MMT improvement could provide clinicians with more detailed guidance for treatment planning and help set realistic expectations for patients. Also, the confidence level of the prediction could be investigated to indicate the likelihood of responding, providing additional decision support to treating therapists beyond the current binary classification.
Integration of multi-channel perspectives
The experiment was designed with a specific clinical point-of-care implementation in mind: take sEMG measurements from a potential target muscle during voluntary contractions based on MMT protocols, and predict its responsiveness to FES therapy in real time or a short amount of time. As such, simplicity and clinical feasibility were the top priorities—a simple setup with bipolar electrodes and no posture restrictions, with only one recording session. Analysis was also done on individual muscles, instead of multiple muscles together.
While this approach aligns with the implementation goals, it limits the depth of electrophysiological insights. Incorporating signals from multiple channels to analyze agonist and antagonist interactions or co-activation patterns could be beneficial. In our experiments, firing of non-target muscles are often observed even though only the target muscle was voluntarily contracted. Compared to using information solely on the target muscle, these patterns could potentially provide more information regarding the systemic effect of SCI, leading to a more robust muscle-specific prediction.
Across Asia Pacific, companies are increasingly required to disclose their climate-related data through the CDP as part of growing regulatory and investor expectations. Yet, many organizations struggle with fragmented data systems, resource constraints, and the challenge of aligning internal teams. As a result, they are seeking efficient, accurate, and impactful ways to manage their CDP submissions.
Join CDP, Workiva, and ERM and explore how companies can enhance the quality and value of their CDP submissions while building a strong foundation for broader ESG performance and reporting.
Our experts will discuss:
Insights on evolving requirements and regional trends in Asia.
How to prepare for CDP submissions through structured data management, governance, and readiness support.
Workiva’s new integration with CDP to automate and streamline disclosures.
How ERM + Workiva help streamline CDP submissions and drive continuous improvement in sustainability outcomes
Whether you’re just starting your CDP journey or looking to enhance your disclosure maturity, this session will equip you with practical insights and tools to simplify reporting and unlock value. You’ll hear from regionally experienced practitioners and see a live walkthrough of how Workiva’s platform now connects directly with CDP, enabling more consistent, auditable, and efficient disclosures.
Attendees will leave equipped to reduce effort, improve data quality, mitigate risk, and elevate their CDP submission as a value-creating exercise rather than just a compliance task.
