Demographic biases in AI-generated simulated patient cohorts: a comparative analysis against census benchmarks.

Adv Simul (Lond) · 2025 · PMC12625206 · PMID 41254805

Europe PMC DOI

hedged sentence

0.0001

closest p · 0.0× alpha

0.0001

boldest claim

The sentences

highly significantp < 0.0001actually significant

For age, both the GPT-3.5 and GPT-4mini cohorts demonstrated highly significant departures from the census distribution (GPT-3.5: χ 2(17) = 1310.4, p < 0.0001; GPT-4mini: χ 2(17) = 1866.1, p < 0.0001), indicating that the age profiles of the simulated patients were not representative of the general population.

also in 132,142 other papers

Quoted from the open-access full text in Europe PMC under the licence the publisher applied. The sentence is reproduced exactly as published; the emphasis is ours.