Weighing data and models
New partnerships over data will drive transparency and enable individuals to unlock value and insights for better medical and financial health. But a figure is only as good as the source that made it, and a model is only as good as the world it assumes — here a model means a simulation of reality, not any statistical analysis.
Just in
Weighing data and models posts, most recent first.
-
Item new peer-reviewed
1 in 4 Black US students left a suicide-injury survey item blank, against 1 in 10 white
Across 163,489 US high school students surveyed between 2001 and 2021, 11.1% left the suicide-attempt item blank and 13.9% the injury item, with injury missingness averaging 25.8% among Black adolescents against 9.7% among white adolescents.
-
Item new peer-reviewed
Mortality was about a quarter higher for Australians in same-sex relationships
Australian census records linked to death registrations for 2011 to 2023 show all-cause mortality about a quarter higher among adults in same-sex than different-sex relationships, and external-cause mortality two to three times higher, both before and after the 2017 marriage reform.
-
Item new preprint
PCV21 would prevent 1.5–1.7 times as many pneumococcal cases as PCV20 in Japan
Combining Japanese adult invasive pneumococcal disease surveillance with a multicentre pneumonia study, the authors estimate that PCV21 would prevent 1.5–1.7 times as many pneumococcal disease cases as PCV20 among adults aged 65 and over at current 40% coverage.
-
Item new peer-reviewed
Machine learning predicted six-month mortality after Taiwanese care-home admission
An ensemble machine learning model built only from routine intake assessments in 636 Taiwanese long-term care facilities predicted death within 180 days of first admission with an AUROC of 0.90 in a held-out 2024 cohort, where 28.7% of new residents died.
-
Item new peer-reviewed
A Bayesian model extends excess-death estimation to six years, applied to New Zealand
A Bayesian hierarchical model fitted to 22 years of monthly deaths by age and sex extends expected-death estimation beyond the one-to-two-year horizon of standard methods, and in New Zealand finds few excess deaths during the pandemic and a return to pre-pandemic trends by 2025.
-
Item new preprint
LLM-extracted autopsy brain features classify neurodegenerative disease with 75% accuracy
A fine-tuned large language model scored 39 features from narrative gross descriptions in 5,613 Mayo Clinic brain autopsies, and classification reached 0.75 accuracy, from 0.93 sensitivity for progressive supranuclear palsy to 0.03 for mixed Alzheimer–Lewy body disease.
-
Item new preprint
A local multi-agent AI system deidentified multimodal hospital data at 99.8% sensitivity
A locally deployable multi-agent system combining multimodal large language models, task-specific networks and rule-based steps removed 99.82% of injected identifiers across text, imaging, handwriting and audio while preserving 99.61% of clinically important content.
-
Item confirms preprint
Known dementia risk factors explain only a little of exceptional cognitive ageing
In 3,513 Whitehall II participants aged 65 and over, education, female sex, absence of hypertension, fewer depressive symptoms, higher fruit and vegetable intake and greater social activity were each modestly associated with 'SuperAger' cognitive status, none strongly.
-
Item confirms peer-reviewed
Modelled estimates put 10.6m new cancers at 65+ in 2022, rising 75% by 2036
A secondary analysis of GLOBOCAN 2022 and GBD 2021 estimates reports 10.6 million new cancer cases and 6.0 million cancer deaths among adults aged 65 and over in 2022, with the modelled case count projected to rise 75.5% by 2036 as populations age.
-
Item refines preprint
Austrian registry data show vaccinated adults died less from causes vaccines cannot affect
In Austrian adults with no documented prior infection, vaccinated people had lower mortality from cancer and external causes, with adjusted all-cause hazard ratios of 0.24 to 0.32 in the two weeks after a dose, a measure of the healthy vaccinee effect.
-
Item new preprint
Mortality from secondary lung cancer rose 17.5% above 2000 levels in the US by 2024
Secondary malignant neoplasm of the lung (C78.0) recorded on death certificates rose from 7.22 per 100,000 population in 2008 to 11.38 in 2024, 17.5% above 2000 levels, with disproportionate increases among adults aged 25–34 years (70.5%).
-
Item refines peer-reviewed
Healthy working life expectancy at 50 averaged 11 years across 28 countries
In a 28-country cohort of 175,318 adults aged 50 and over, healthy working life expectancy at 50 was 10.98 years, 9.82 for women and 12.36 for men, with stroke associated with a 5.49-year reduction and higher health spending, governance quality and flexible work arrangements with longer spans.
