Clinical validation of an AI-based blood testing device for diagnosis and prognosis of acute infection and sepsis.

Summary

In a 1,222-patient ED cohort, a 29-mRNA isothermal amplification test with ML (TriVerity) outperformed CRP/PCT/WBC for bacterial/viral differentiation and predicted 7-day critical care interventions. High rule-in specificity and rule-out sensitivity suggest potential for antibiotic stewardship, pending interventional trials.

Key Findings

• TriVerity achieved AUROC 0.83 for bacterial and 0.91 for viral infection, outperforming CRP, procalcitonin, and WBC.
• Severity score AUROC 0.78 for predicting need for critical care interventions within 7 days, improving over qSOFA.
• Rule-in specificity >92% and rule-out sensitivity >95% across scores; projected 60–70% reduction in inappropriate antibiotic decisions.

Clinical Implications

May enable earlier, more accurate differentiation of bacterial vs viral infections and severity risk stratification in ED workflows, reducing unnecessary antibiotics and optimizing resource allocation.

Limitations

• No interventional testing to show clinical outcome improvement; actionability remains to be proven.
• Device/instrument requirements and costs may affect scalability; real-world impact on antibiotic use needs trial confirmation.

Future Directions

Conduct interventional trials testing TriVerity-guided care pathways on antibiotic stewardship and patient outcomes; assess cost-effectiveness and integration into ED workflows.