FusionCloud™ / Fusion AI

Actionable Intelligence for hospital metagenomics.

FusionCloud™ is the interpretation layer of ONETest™: transforming microbial sequence data into ranked, interpretable outputs for organism relevance, AMR-marker context, and interpretation support.

Beyond detection, Fusion AI applies organism-specific dynamic ML thresholds calibrated using validation and control data, turning genomic signal into structured reporting that can support laboratory review and clinical-context interpretation.

Explore the Engine View Reporting Layer

Why FusionCloud™ Matters

The hard part is not sequencing. It is interpretation.

Clinical metagenomics can generate many microbial signals from pathogens, colonizers, normal flora, environmental background, and reagent contamination. FusionCloud™ is designed to convert that complexity into a structured evidence layer for laboratory review and clinician interpretation.

Separate signal from noise

Control-aware filtering distinguishes sample-derived microbial evidence from background, contaminants, and normal flora.

Calibrate per organism

Organism-specific baselines and dynamic thresholds support detection rules that are not one-size-fits-all.

Rank what matters

FusionCloud™ organizes results around organism prioritization, AMR-marker context, and interpretation support and not an undifferentiated organism list.

Actionable Intelligence Engine

From raw sequence files to qualified species reports.

FusionCloud™ connects bioinformatics, empirical modeling, negative-control learning, contamination modeling, species qualification, and structured reporting into one analysis pipeline.

Input

DNA sequencer file enters the cloud workflow.

QC + host depletion

Quality filtering and host-read handling.

Taxonomic assignment

Reference mapping and organism calls.

Quantification

FPKM, coverage, abundance, and mapping metrics.

Modeling

Training sets, thresholds, controls, and contamination models.

Qualification

Pathogen vs. normal flora and background classification.

Reporting

Structured species output, sample distribution, and interpretation support.

How Fusion AI Interprets Signal

A dynamic, control-informed model for each analysis.

FusionCloud™ uses empirical thresholds and multi-parameter evidence to determine whether a microbial signal should be qualified, categorized, and surfaced for review. This is the software layer that turns targeted metagenomics into laboratory-reviewable workflow.

Not black-box AI

The engine is built around interpretable evidence: mapped reads, abundance, coverage, species-specific background distributions, controls, and structured output categories.

Species-specific background baselines

Thresholds are calibrated against observed background distributions rather than applying one universal cutoff across all organisms.

Negative-control learning

Run controls and historical control behavior help identify recurrent reagent, sequencing, and process-associated background.

Multi-parameter qualification

FPKM, relative abundance, breadth of coverage, mapped reads, and other metrics are combined to support a qualified result.

Actionable reporting

Final outputs are organized to support laboratory review, clinical-context interpretation, antimicrobial-stewardship review, and follow-up testing considerations.

Reporting Layer

From organism list to structured clarity.

FusionCloud™ is designed to make reports useful in the critical-care window: which organisms are detected, which signals are prioritized, what AMR-marker evidence is available, and what evidence should be reviewed in context.

Ranked

Species-level findings

AMR

Resistance-marker context

Evidence-Informed Software

Built from real BAL validation data, not synthetic examples.

The FusionCloud™ model reflects the realities of lower-respiratory specimens: host-rich samples, commensal flora, antibiotic exposure, polymicrobial findings, and organism-specific background behavior.

No-template controls

Used in the clinical cohort to model background signatures.

Human-cell controls

Helped define recurring species-level background patterns.

99%

Isolate-level specificity

Micro-averaged across culture-benchmarked BAL organisms.

21%

Additive yield

Additional microbiologic findings beyond routine culture in the BAL cohort.

Interpretation Principle

Additional detections require clinical context.

FusionCloud™ is built to support contamination-aware interpretation: OT-positive / culture-negative signals are treated as additional microbiologic findings that should be interpreted alongside abundance, controls, specimen type, antibiotic exposure, and orthogonal evidence where available.

Pathogen

Qualified signal with support from organism-specific thresholds and metrics.

Normal flora

Contextual classification for organisms commonly present in respiratory specimens.

Background

Signals suppressed or downgraded based on control-informed background behavior.

ONETest™ System Fit

The fourth innovation that makes in-hospital tNGS interpretable.

ONETest™ combines QuantumProbes™ capture, UniPrep™ chemistry, fluidics automation, and FusionCloud™ analytics. FusionCloud™ is where raw genomic signal becomes structured analytical evidence.

Capture

QuantumProbes™ enrich microbial DNA.

Prepare

UniPrep™ converts DNA into a sequenceable library.

Automate

Fluidics workflow reduces manual complexity.

Interpret

FusionCloud™ ranks organism evidence and AMR-marker context.

From Dx To Rx in <24 Hours

Actionable Intelligence is what makes metagenomics usable.

FusionCloud™ closes the gap between sequence data and interpretation by transforming complex microbial signals into ranked, analytical outputs for organism relevance, AMR-marker context, and interpretation support.

Explore ONETest™ Platform Contact Fusion Genomics