Researchers reviewed 27 studies on AI in blood coagulation testing

Medical researchers recently published a scoping review in npj Digital Medicine investigating how artificial intelligence handles data from Viscoelastic haemostatic assays (VHA).

VHA tests, such as thromboelastography, measure how blood clots in real-time. Clinicians usually rely on derived thresholds to make decisions, but these tests produce rich data that can be difficult to interpret quickly.

After screening 493 records, the authors identified 27 studies where machine learning helped interpret VHA results. These models were applied across various fields, including trauma, surgery, organ transplants, and critical care.

Most of the research focused on using specific parameters from these tests as predictors for patient outcomes. While some studies showed high performance, the review highlights significant gaps in the current evidence base.

Many of the projects analyzed were limited by small sample sizes and lacked rigorous external testing. Furthermore, while the technology shows promise for quick analysis, sub-minute predictions remain purely theoretical at this stage.