npj Digital Medicine, Published online: 14 September 2026; doi:10.1038/s41746-026-03214-z

While clinical trials are the gold standard for determining causative side effects from treatment, some trials are too logistically or ethically challenging to complete. Many side effects of treatments are learned from retrospective analyses; however, these can be confounded by other random variables. Causal and explainable machine learning tools are promising for teasing apart confounders from real treatment side effects in observational data.

npj Digital Medicine, Published online: 07 April 2026; doi:10.1038/s41746-026-02615-4

Yakdan et al. demonstrate that foundation models (FMs) trained to predict cervical spondylotic myelopathy from electronic health record data outperform traditional models on internal datasets but lose their advantage during external validation. This suggests that the feature-dense patterns learned by FMs may reduce their portability across settings, particularly for rare outcomes. As FMs approach clinical deployment, local validation, subgroup analysis, and attention to implementation burden are essential to inform health system planning and stewardship.