Every hour of delay in getting the right antibiotic to a sepsis patient measurably raises their risk of death. That urgency multiplies during mass casualty events, natural disasters, or outbreaks involving bloodstream pathogens, when hospitals may be triaging large numbers of patients simultaneously. Yet the standard test for identifying what’s actually causing a bloodstream infection can take two to seven days, forcing doctors to guess with broad-spectrum antibiotics in the meantime, a practice that can fail patients and fuel antimicrobial resistance, an especially serious concern when large volumes of empiric antibiotics are deployed at once during a surge event.
A new diagnostic platform described this week in Science Advances aims to close that gap, delivering a full diagnosis in as little as 6.75 hours. The work was funded by a National Institutes of Health grant to researchers from Stanford University and Penn State University. Bloodstream infections are notoriously hard to catch quickly because the bacteria causing them are often present in vanishingly small numbers, sometimes fewer than one bacterial cell per milliliter of blood.
The team’s solution, called STREAM (sedimentation-assisted tandem rocking and enrichment for analysis and monitoring), tackles that problem in an unusual way. Rather than trying to fish bacteria directly out of whole blood, a method that tends to miss low-level infections, STREAM gently rocks a blood sample mixed with a custom nutrient broth. Red blood cells clump together and sink, while any bacteria present stay suspended near the top and begin multiplying naturally. This process both clears away most of the blood cells that interfere with detection and gives the bacteria a head start toward detectable numbers, all within a few hours rather than days.
Once enriched, the sample moves through two further steps. The first, a technique the team calls MB-seqFISH, uses fluorescent molecular probes that bind to bacterial genetic material and light up in specific color patterns, allowing the system to identify the exact species present using only a handful of cells. Tested against 104 real patient blood culture samples, this identification step matched standard clinical lab results 96.15 percent of the time. The second step embeds bacteria in a thin gel and tracks how individual cells respond, in shape, growth, and survival, when exposed to various antibiotics, revealing which drugs will actually work and at what dose. Across 219 different drug-dose combinations tested, this approach agreed with conventional lab results in the vast majority of cases.
In practical terms, the researchers showed the full process, from blood sample to actionable results, took between roughly 7 and 17 hours depending on how much bacteria was initially present and how fast a particular species grows. That compares with the 2- to 7-day timeline for conventional blood culture, and the system performed reliably even at bacterial concentrations as low as one-tenth of a cell per milliliter, a level so faint that some existing rapid tests miss it entirely.
The authors are candid about the work that remains before this reaches clinical use. Patients who have already received antibiotics before their blood is drawn, a common scenario, may show reduced bacterial counts that complicate detection. Infections involving multiple organisms at once can also be tricky, since faster-growing bacteria may mask others present in the same sample. And the process currently involves enough manual steps that cost and complexity remain real barriers; the researchers say automation and integration with existing lab equipment will be necessary before the platform can be deployed at scale.
Still, for a field where speed to the right antibiotic is one of the clearest levers on patient survival, and where that same speed becomes a system-wide capacity question during a surge event, a platform that compresses days into hours represents a meaningful step forward for both routine sepsis care and outbreak-scale preparedness.
Sources and further reading:
Chin SM, Epiphaniou E, Kovtunov E, et al. Rapid and robust diagnosis of bloodstream infections by single-cell analysis. Science Advances. August 26, 2026.
This article was researched and sourced by Global Biodefense editors and reported with Claude AI assistance for drafting and editing.