1 paper2 findings
Amyloid-beta status
upin humans1
Machine learning model predicts amyloid-beta status in humans (AUC 0.888, 95% CI 0.834–0.941).
“a robust Aβ status prediction model (AUC=0.888, 95% CI 0.834 ‐ 0.941) with PSD, APOE, p ‐Tau217, and demographic info was also constructed.”
Humansn = 1,200
AAIC 2025 abstract23 Dec 2025
Cognitive impairment
upin humans1
Machine learning model predicts cognitive impairment in humans (AUC 0.924, 95% CI 0.892–0.955).
“a machine learning CI diagnostic model (AUC=0.924, 95% CI 0.892 ‐ 0.955) incorporating PSD, APOE, GFAP, and demographic info was developed.”
Humansn = 1,200
AAIC 2025 abstract23 Dec 2025
Against the favourable pattern, the poorer pattern had hazard ratios of 1.23, 2.05 and 1.61, comparing cancer risk over time across three cohorts.Against the favourable pattern, the poorer pattern had hazard ratios of 1.23, 2.05 and 1.61, comparing cancer risk over time across three cohorts.
Each 0.151 microgram per cubic metre increase in residential benzene was associated with a 26% higher risk of developing non-alcoholic fatty liver disease.Each 0.151 microgram per cubic metre increase in residential benzene was associated with a 26% higher risk of developing non-alcoholic fatty liver disease.
In killifish experiments, lifelong mating reduced inflammation in old females’ main liver cells, while models based on gene activity predicted “younger” ages for several…In killifish experiments, lifelong mating reduced inflammation in old females’ main liver cells, while models based on gene activity predicted “younger” ages for several liver cell types.
Higher pollution-mixture scores were associated with greater odds of developing overlapping conditions than lower scores, with an odds ratio of 1.25.Higher pollution-mixture scores were associated with greater odds of developing overlapping conditions than lower scores, with an odds ratio of 1.25.
The best model scored 0.917 on a measure of how well it separated patients with and without frailty, in 124 patients from another city.The best model scored 0.917 on a measure of how well it separated patients with and without frailty, in 124 patients from another city.