Clinical AI in acute and critical illness
Development, evaluation, and critical appraisal of clinical AI for high acuity care, with emphasis on sepsis, severe infection, model discrimination, interpretability, bias, generalizability, and bedside clinical utility.
Selective testing and label bias
Study of how surveillance intensity, selective diagnostic testing, missing outcomes, and workflow can shape the labels used to train prediction models. The central question is whether a model is learning disease biology or patterns of observation.
Read the full insight: Outcome Ascertainment Bias in Clinical AIFrom prediction to treatment decisions
Research connecting clinical prediction with causal inference, treatment effect heterogeneity, and individualized decision support. The goal is to move beyond asking who is at risk toward determining which patient may benefit from a particular intervention.
Sepsis and cardiovascular outcomes
Clinical outcomes research in sepsis, severe infection, and acute illness associated atrial fibrillation, including competing risks, treatment selection, and longitudinal outcomes.
Transplantation, kidney disease, pulmonary medicine, and surgery
Collaborative research in complex clinical populations where surveillance, selection, longitudinal outcomes, and treatment decisions are especially important.