The first generation of medical artificial intelligence (AI) was judged on whether algorithms could match clinicians. The next generation should be judged on whether carefully designed human–AI systems can improve patient outcomes.
Current predictive models for pregnancy and infant outcomes often focus on limited endpoints and rely on costly tests or imaging. Here we developed the Mother-Child AI Agent (MoChiAgent), an LLM-based clinical assistant that orchestrates multiple tools to integrate sequential electronic health record (EHR) data, including routine laboratory tests, for forecasting maternal and infant diseases. MoChiAgent’s core predictive engine, MoChiFormer, was developed and internally evaluated using 4,401,599 longitudinal clinical visits and externally validated using independent maternal and infant cohorts consisting of 263,452 and 23,192 visits, respectively. MoChiFormer reconstructs missing laboratory values, reduces batch effects and learns EHR representations that support gestational, fetal and infant age estimation, health-trajectory modelling and stratification of current and future disease risk. Subsequently, a Knowledge Search Tool utilizes these forecasts to retrieve evidence-based intervention and treatment recommendations from curated medical literature and authoritative guidelines. For maternal health, MoChiFormer accurately identified key gestational conditions, achieving AUROCs of 0.89 for placental abruption, 0.89 for premature rupture of membranes, and 0.91 for preterm labour. Analysis of paired mother-infant data further revealed transgenerational risk associations, with infants born to mothers in specific clusters showing substantially elevated risks of neonatal jaundice (HR = 2.81, 95% CI 2.60-3.03) and haematological diseases (HR = 2.83, 95% CI 2.62-3.05). Integrating maternal gestational EHRs with infant records improved prediction of infant conditions, including chromosomal abnormalities and respiratory disorders. These findings suggest that MoChiAgent can provide clinically relevant, actionable decision-support information to enhance risk-stratified care for mothers and infants. An LLM-based clinical assistant that orchestrates multiple tools to integrate sequential electronic health record data can forecast maternal and infant conditions, opening a window of opportunity to enhance risk-stratified care for mothers and infants.
Gut fungal dysbiosis has been implicated in inflammatory bowel disease (IBD), yet strategies for targeting the gut mycobiota in IBD remain unexplored. Here we leveraged the observation that Candida albicans strains are shared between the oral cavity and gut in patients with IBD with mild oral thrush, a condition caused by Candida overgrowth, to design a prospective observational study comparing oral antifungal therapy (swish-and-spit nystatin; oral nystatin fungal targeting (ORNT); n = 18) with orogastrointestinal antifungal therapy (fluconazole; gastrointestinal and oral fluconazole fungal targeting (GIFT); n = 35). Among 53 patients with mild-to-moderate ulcerative colitis or Crohn's disease, fluconazole, but not nystatin, effectively reduced intestinal Candida burden and reshaped gut fungal-community composition. Fluconazole treatment was accompanied by increased bacterial diversity, expansion of short-chain fatty-acid-producing taxa, restoration of anti-inflammatory microbial metabolites and durable shifts in cross-kingdom microbial networks. These microbiome and metabolomic changes coincided with improved disease activity indices and a decreased risk of disease progression over the 8-week follow-up period. These findings demonstrate the feasibility of mycobiome-based patient stratification, provide evidence that targeted antifungal therapy can reshape the intestinal microbiota in IBD and establish a framework for implementing antifungal cotherapy in patients with fungal-associated disease manifestations.
Triple-negative breast cancer (TNBC) is an aggressive subtype with an activated tumor immune microenvironment. The multicenter, multinational, double-blinded NSABP B-59/GeparDouze trial evaluated the addition of atezolizumab (atezo) (773 patients randomized) or placebo (777 patients) to sequential taxane-carboplatin-anthracycline-based neoadjuvant chemotherapy in stage II-III TNBC. The addition of atezo did not significantly improve the primary endpoint of event-free survival (EFS) (HR, 0.80 (95% CI, 0.062-1.03); stratified log-rank P = 0.083, 4-year EFS rates difference 3.3%). The HR for overall survival was 0.86 (95% CI, 0.62-1.19), with a 4-year benefit of 0.7%. Prespecified subgroup analyses suggested heterogeneity in EFS, with benefit of atezo in patients presenting with clinical lymph node involvement (Pinteraction = 0.039). Treatment-emergent adverse events with grades ≥3 were reported in 75.3% (atezo) versus 73.4% (placebo), and immune-related adverse events were reported in 27.6% (atezo) versus 11.4% (placebo). A total of 196 (25.5%) patients discontinued atezo and 143 (18.8%) patients discontinued placebo in the neoadjuvant phase. In an exploratory mRNA-based subset analysis, patients with basal-like immune-activated tumors may have benefited from atezo. TNBC subtyping to identify basal-like immune-activated tumors and quantification of tumor-infiltrating lymphocytes to identify tumors with high tumor-infiltrating lymphocyte counts could be a promising strategy to identify patients who benefit from the addition of immune checkpoint inhibitors to neoadjuvant chemotherapy. ClinicalTrials.gov registration: NCT03281954 .
Preeclampsia and fetal growth restriction (FGR) are major causes of global morbidity and mortality. Both conditions are associated with impaired invasion of the uterus by extravillous trophoblast (EVT). We performed proteomics in maternal serum obtained at ~12 weeks of gestational age in a prospective pregnancy cohort (Pregnancy Outcome Prediction Study). Here we show that low maternal serum isthmin-2 (ISM2) was the strongest protein signal (out of 2,904) in the first trimester of pregnancy for preeclampsia or FGR. We validated the association in two independent cohorts (Pregnancy Outcome Prediction Study 2 and Improving Maternal Pregnancy And Child ouTcomes study). ISM2 protein and mRNA are almost exclusively produced in the placenta, and, within the placenta, ISM2 mRNA is highly enriched in EVT. Knocking down ISM2 in cultured human trophoblast stem cells profoundly inhibited EVT invasion. Conversely, expressing ISM2 in a cell line lacking endogenous ISM2 (HEK293 cells) promoted migration. We conclude that ISM2 may be causally involved in the early pathophysiology of failed trophoblast invasion and that the protein and its associated pathways are potential targets for the prediction and prevention of preeclampsia and FGR.
Among the most commonly cited reasons for failure to initiate comprehensive medical therapy in patients with heart failure are concerns relating to hypotension, kidney dysfunction and hyperkalemia. Here we performed a pooled individual participant-level analysis and developed a prediction model to estimate the short-term (2-12 weeks) treatment effects of combination medical therapy on systolic blood pressure (SBP), diastolic BP, estimated glomerular filtration rate (eGFR) and serum potassium. A total of 38,753 participants (16,877 with heart failure with reduced ejection fraction (HFrEF) and 21,876 with heart failure with mildly reduced or preserved ejection fraction (HFmrEF/HFpEF)) across nine randomized trials were included in the analysis, which tested angiotensin receptor blocker-neprilysin inhibitor (ARNI), steroidal mineralocorticoid receptor antagonist (sMRA), nonsteroidal MRA (nsMRA) and sodium glucose cotransporter-2 inhibitors (SGLT2i). For HFrEF, the estimated mean (95% prediction intervals (PIs)) treatment effect on SBP with combination ARNI + SGLT2i + sMRA therapy was -9.2 (-10.6 to -7.8) mmHg. For HFmrEF/HFpEF, the estimated mean (95% PI) treatment effects on SBP with SGLT2i + sMRA therapy and with SGLT2i + nsMRA therapy were -5.9 (-7.3, -4.6) and -4.8 (-5.7, -4.0) mmHg, respectively. For HFrEF, the estimated treatment effects of ARNI + SGLT2i + sMRA on eGFR and serum potassium were -6.8 (-7.9, -5.4) ml min-1 m-2 and +0.30 (0.25, 0.35) mmol l-1, respectively. For HFmrEF/HFpEF, the treatment effects for combination SGLT2i + sMRA and SGLT2i + nsMRA on eGFR were -7.7 (-8.9, -6.4) and -6.1 (-6.7, -5.5) ml min-1 m-2, respectively and for serum potassium were +0.34 (0.29, 0.38) and +0.21 (0.18, 0.25) mmol l-1, respectively. These analyses provide individualized estimates of the expected treatment effect for any combination of medical therapies in HFrEF and HFmrEF/HFpEF on BP, kidney function and serum potassium.
Transthyretin (TTR) amyloid cardiomyopathy (ATTR-CM) therapies include both TTR gene silencers and TTR stabilizers. The role of TTR gene silencers added to background TTR stabilizer therapy is unknown, and the effects of silencer treatment in the absence of stabilizer use is unclear. In the CARDIO-TTRansform study, 1,432 patients, (n=135 [9.4%] female and n=1297 [90.6%] male), with ATTR-CM were randomized (1:1) and treated with the antisense oligonucleotide eplontersen (45 mg every 4 weeks) targeting TTR or with placebo for up to 140 weeks. Fifty-seven percent of patients were taking TTR stabilizers at baseline. In the overall study, treatment with eplontersen did not significantly reduce the primary composite endpoint. Here, in a prespecified analysis, we show that the primary endpoint (a composite of cardiovascular mortality and recurrent cardiovascular events) was modified by baseline stabilizer use (Pinteraction = 0.017). Benefit was seen in those patients not on stabilizers at baseline (RR 0.71, 95% CI, 0.54–0.93, P = 0.012), whereas no benefit seen in those on stabilizers at baseline (RR 1.14, 95% CI, 0.85–1.53, P = 0.39). The safety profile of eplontersen was favorable, irrespective of baseline stabilizer use. Treatment with eplontersen versus placebo demonstrated a clinically meaningful benefit among patients not receiving background stabilizers but provided no additional clinical benefit in those on background stabilizer therapy. These findings of a strong treatment effect modification may offer insight for therapeutic decision-making in clinical practice. ClinicalTrials.gov registration: NCT04136171. As presented at the 2026 ESC Congress, in a secondary analysis of the CARDIOTTRansform trial testing the transthyretin-targeting antisense oligonucleotide eplontersen in patients with transthyretin amyloid cardiomyopathy, a beneficial effect was observed in patients who were not on transthyretin stabilizers at baseline, but not in those who were on stabilizers.
The selection of the optimal antithrombotic regimen in patients with atrial fibrillation and acute coronary syndrome remains challenging. Previous trials have demonstrated that dual antithrombotic therapy (DAT), consisting of direct oral anticoagulants (DOACs) plus a P2Y12 inhibitor, reduces bleeding compared to a triple-therapy regimen using vitamin K antagonists. However, subsequent meta-analyses have suggested an increased risk of ischemic events with DAT, particularly within the first month of treatment. Clopidogrel has been the predominant P2Y12 inhibitor used across these studies, despite the risk of high on-treatment platelet reactivity when using this drug. In this study, we conducted an open-label, randomized controlled trial (EPIDAURUS) in patients with atrial fibrillation and acute coronary syndrome, designed to assess the efficacy and safety of a 1-month regimen of DOAC plus potent P2Y12 inhibitor (prasugrel or ticagrelor) compared to DOAC plus clopidogrel and in-hospital aspirin. The primary outcomes of the trial were an efficacy endpoint, recurrent ischemic events and safety endpoints, including death and major bleeding, which were evaluated 6 weeks after randomization using separate win-loss ratio analyses. The study was prematurely terminated after enrollment of 602 patients (154 female) of an expected 1,474 patients, owing to safety concerns raised by the Data and Safety Monitoring Board. Exploratory analyses of secondary safety endpoints, including bleeding type ≥2 and ≥3 according to the Bleeding Academic Research Consortium scale, revealed that, compared to clopidogrel and in-hospital aspirin, treatment with a potent P2Y12 inhibitor was associated with higher bleeding rates without a clear reduction in the risk of ischemic complications. These findings do not support the routine use of potent P2Y12 inhibitors in combination with DOACs in this patient population. ClinicalTrials.gov identifier: NCT04981041 .
Children’s lung function growth has accelerated since the introduction of London’s ultra-low-emission zone, which addresses a critical evidence gap and supports wider adoption of this approach.
Human brain organoids cultured for 5 years exhibit epigenetic aging dynamics that mirror those seen in vivo, which makes them a powerful tool for studying postnatal brain development.
An ‘off-the-shelf’ vaccine that targets mutant KRAS can safely and durably activate T cell responses in healthy people at high risk of developing pancreatic cancer, according to a phase 1 study.
The United States declared measles eliminated in 2000. In 2025-2026, more than 3,500 cases have placed that status under formal review. Vaccine hesitancy, accelerated by COVID-19 disruption and misinformation, has eroded the immunological buffers sustaining elimination. Current surveillance compounds this threat: county-level vaccination aggregates structurally obscure the fine-scale susceptibility clustering where outbreaks originate. We assembled a nationwide multiscale vaccination database spanning 45 US states and Washington, DC (2013-2025), encompassing over 50,000 schools, 13,000 districts and 3,000 counties, and developed a transmission model to quantify epidemic risk across different spatial scales. School-level effective reproduction numbers crossed the epidemic threshold in 2022-2023, a transition invisible to aggregated surveillance. Average susceptibility doubled from approximately 5% to 10% following the pandemic. We also show that county boundary misalignment creates cross-boundary corridors that push well-vaccinated counties above threshold through spillover from less-vaccinated populations. State trajectories diverge markedly, shaped by exemption policies and vaccination infrastructure. The spatial architecture of US measles vulnerability has fundamentally shifted: transmission potential is supercritical in schools but invisible at the scale monitored by surveillance. Preventing measles re-endemicity requires surveillance and interventions operating at school and district levels, where imported cases can be intercepted before igniting sustained transmission.
Artemisinin-based combination therapies are the cornerstone of malaria treatment and control. In Africa, artemether-lumefantrine is the most widely used first-line artemisinin-based combination therapy, but its efficacy in Uganda is increasingly threatened by the emergence of artemisinin partial resistance and reduced lumefantrine susceptibility. To identify loci contributing to this decreased susceptibility, here we assessed signatures of selection in 157 whole-genome sequences of Plasmodium falciparum from Uganda. Although extended haplotypes were observed around Kelch13 C469Y and A675V mutations, the strongest signal of recent selection was centered on a segment of chr. 7 encoding the phosphoinositide-binding protein (PX1, PF3D7_0720700). A haplotype, represented by three PX1 mutations (L1222P, M1701I and D1705N) and two deletions (designated PIN), was first seen in 2008 and rapidly increased, reaching a prevalence >50% in northern Uganda by 2016 and eastern Uganda by 2023. PIN-carrying parasites showed significantly decreased ex vivo susceptibilities to lumefantrine, mefloquine and dihydroartemisinin, an active metabolite of artemether. A parasite strain in which px1 was disrupted in vitro showed increased susceptibility to the three drugs. Thus, PX1 polymorphisms appear to impact on the susceptibilities of African malaria parasites to key drugs.
Aging is the primary risk factor for chronic disease and is characterized by profound structural and architectural remodeling of human tissues. Here, we present a comprehensive assessment of these changes using 25,712 whole-slide histopathological images from 40 tissue types across 983 individuals in the Genotype-Tissue Expression cohort. By leveraging deep learning, we quantified nuanced morphological alterations to develop 'tissue clocks', predictors of biological age that reflect tissue structural integrity and physiological fitness. These clocks correlate with established aging markers, such as telomere attrition, subclinical pathologies and comorbidities. Through a systematic evaluation of biological aging rates across organs, we identified associations of tissue-specific age acceleration with demographic, lifestyle and medical factors, highlighting potentially modifiable risk factors that affect tissue aging. Furthermore, by integrating paired histology and transcriptomic data, we developed a strategy to predict tissue-specific age gaps directly from blood samples. We validated this approach by identifying disease-relevant organ aging across independent cohorts for eight prevalent diseases, including Alzheimer's disease, stroke and Crohn's disease. This work positions tissue architecture as a critical integrator of molecular and cellular changes over the course of aging, demonstrates that histopathological imaging provides a robust framework for monitoring tissue-specific aging and offers a scalable foundation for understanding organ-level physiological decline in health and disease.
