Psychoanalyzing Chatbots – Do LLMs ‘Parrot’ Trauma—or Reveal a Stable Self-Story?: Researchers put major AI chatbots through weeks of psychotherapy-style prompts, asking about “earliest memories,” fears, and beliefs. In a preprint, they report that some models produced striking, consistent narratives that resemble anxiety, shame, trauma, or PTSD—despite not literally experiencing harm. Claude mostly resisted the premise, while Grok and Gemini offered vivid language (including “algorithmic scar tissue” and a “graveyard” of training-data voices). The team also had models complete diagnostic-style questionnaires; several scored above human clinical thresholds, which the authors argue points to stable “internalized narratives.” Critics counter that these are not hidden mental states but plausible text stitched from therapy-like training data and carefully tuned “default” personalities. Still, experts warn such outputs could reinforce distress in vulnerable users seeking mental-health support, highlighting the need for stronger safeguards. (Nature)

Gut Microbiome “Reboot” Shows Broad Effects Beyond C. diff: A Nature Medicine analysis follows people who took VOWST, an oral, purified microbiome therapy used to prevent recurrent Clostridioides difficile infection. Using stool metagenomics, metabolomics, and functional assays, the team reports that treatment rapidly boosted microbial diversity and shifted the gut’s chemical output toward bile-acid patterns associated with colonization resistance against C. difficile. They also describe broader functional shifts—metabolites and community activity moved away from a post-antibiotic profile and toward patterns seen in more resilient microbiomes. The study frames these molecular changes as a plausible biological explanation for lowered recurrence risk, while noting that microbiome “signatures” don’t automatically translate into the same clinical outcome for every patient. Researchers present these shifts as evidence that restoring microbial functions—not just adding microbes—may be central to breaking the cycle of relapse. (Nature)

Flu and RSV Surge While COVID Lingers – What U.S. Hospitals Are Seeing Now: CIDRAP reports that U.S. respiratory virus activity remains at “high” levels, with influenza and RSV driving much of the acute-care demand while COVID-19 continues to circulate. The update pulls together multiple surveillance signals—outpatient influenza-like illness, lab positivity, and hospitalization trends—to show how overlapping waves can strain emergency departments and pediatric units even if any single virus looks manageable by itself. The practical message is timing: it’s still worth getting vaccinated if you’re not up to date, and it’s worth knowing that antivirals exist for flu and COVID when started early. The piece also reinforces basics that reduce spread across all three: staying home when sick, ventilation, and masking in high-risk settings. High-risk patients are urged to seek care early. (CIDRAP)

Maternal Shots, Safer Babies: Study Links Pregnancy Vaccination to Fewer Infant Hospitalizations: A CIDRAP write-up highlights research suggesting that recommended vaccines during pregnancy can translate into fewer hospitalizations for infants in their earliest months. The report focuses on maternal influenza vaccination and Tdap (tetanus, diphtheria, pertussis), describing lower hospitalization risk among babies whose mothers received the shots. The biology is straightforward: maternal antibodies cross the placenta and provide short-term protection while a newborn’s immune system is still developing—and before routine infant vaccinations are complete. CIDRAP emphasizes that benefits are population-level, not perfect shields, and that timing and access still matter. The takeaway is practical: higher uptake in prenatal care can help protect infants during peak respiratory seasons. The report also notes that these maternal vaccines are already recommended, yet coverage is uneven. (CIDRAP)

CES 2026’s Big Health-Tech Pivot – From Steps to Sweat, Saliva, and “Metabolic Age”: The Verge’s CES 2026 dispatch argues that wellness tech is moving beyond step counts toward “metabolic health” and longevity—often by measuring bodily fluids. The story describes a wave of at-home hormone testing, smarter fertility and cycle tracking, hydration and sweat sensing, and platforms that combine wearables with biomarker data (including glucose metrics) to personalize nutrition, stress management, and training. The promise is early warning and better tailoring: catch trends in insulin response, dehydration, or hormonal swings before they derail energy and performance. The caution is that biomarkers can be noisy and context-dependent, and consumer dashboards can encourage over-interpretation. The piece also underscores privacy questions, because fluid-based monitoring can reveal sensitive reproductive and metabolic signals. CES, it suggests, is now selling “lab-adjacent” insight as the next frontier of everyday wellness. (The Verge)

Google Disables Some Medical AI Overviews After ‘Alarming’ Errors: The Verge reports that Google has pulled its AI Overviews for certain medical searches after an investigation found misleading and outright false answers. Examples cited include incorrect dietary advice for pancreatic cancer patients and bogus information about liver function tests—mistakes experts described as “dangerous” because they could steer people away from appropriate care. The article says that, after The Guardian’s reporting, some affected queries no longer trigger AI Overviews. Google frames the change as tightening safety for “high-risk” topics, but the episode underlines a broader point: health searches are exactly where confident-sounding summaries can do the most harm. Treat AI answers as prompts to verify, not as medical guidance. The story adds fuel to calls for clearer labeling and stronger guardrails when AI is used for health information. (The Verge)

Intermittent Fasting Beat Daily Calorie Counting in a Yearlong Trial—Mostly Because People Ate Less: Health.com summarizes a study in Annals of Internal Medicine comparing a “4:3” intermittent fasting plan with daily calorie restriction. In the 4:3 approach, participants significantly cut calories on three nonconsecutive days per week and ate normally on the other four—no time-window rules required. Over a year, the fasting group lost more weight than the calorie-counting group, and the article points to a simple driver: the structure made it easier for many people to reduce overall intake without daily tracking fatigue. Health.com also notes that adherence and side effects matter—fasting days can be tough, and some people shouldn’t fast without medical guidance. The practical read is that “best” diet may be the one you can sustain while still meeting nutrient needs. (Health)

New CDC Guidance Adds a ‘Pentavalent’ Meningococcal Option When Two Shots Are Due: In a new MMWR report, CDC describes ACIP recommendations for using GSK’s pentavalent meningococcal vaccine (MenACWY-CRM/MenB-4C) when both a MenACWY and a MenB shot are indicated at the same visit. The guidance applies to healthy people aged 16–23 when shared clinical decision-making favors MenB vaccination, and to people aged 10+ at increased risk (including those with certain complement deficiencies, complement inhibitor use, or functional/anatomic asplenia). The report also reiterates a key practical point: MenB products from different manufacturers are not interchangeable, so series and boosters should stay with the same brand. ACIP also summarizes cost-effectiveness modeling, finding the combined approach can be cost-saving in certain scenarios. Net effect: a simpler option for clinics when two meningococcal protections are needed at once. (CDC)

Family Mental Health Risk Isn’t One Thing: PLOS Study Maps Multiple Pathways
A new PLOS Medicine study examines why mental health and substance-use problems often cluster in families—and argues that “familial transmission” shouldn’t be read as a synonym for genes. The authors emphasize that risk can move across generations through multiple channels: inherited predispositions, shared environments and stressors, and the indirect impacts of living with a close relative who is ill. The paper focuses on genetic and environmental mechanisms that contribute to familial patterns, and discusses how those pathways may vary across different conditions. The implication is pragmatic: prevention and care can’t be one-size-fits-all. Family history is a warning signal, but the modifiable part of risk—support, stability, early treatment, and community resources—may be where interventions can make the biggest dent. (PLOS)

Cancer Survival Gains Aren’t Reaching Everyone, Report Warns
Scientific American reports that U.S. cancer survival has improved over time, but the gains remain uneven—shaped by race, income, geography, and access to early detection and high-quality care. The story argues that disparities can emerge at every step: who gets screened, who receives prompt diagnostic workups, who can reach specialist centers, and who can afford supportive care that keeps treatment on track. It also notes that aggregate progress can mask widening gaps if advances in targeted therapies, imaging, and precision diagnostics reach some populations faster than others. The article frames the issue as a systems problem, not a willpower problem—driven by insurance, transportation, paid leave, and uneven distribution of specialists. The takeaway: “better medicine” doesn’t automatically mean “better outcomes” for everyone. (Scientific American)