MATRIX Discovery
Electrome's operating internal AI-native software supports its discovery and R&D pipelines. It helps scientists prioritize candidate signals and design controlled tests, but no model has authority to advance work independently: every advancement passes a human scientific release gate.
System Limitations
MATRIX Discovery is internal R&D software, not an FDA-cleared medical device or a product or service available for patient use. It does not examine, diagnose, monitor, treat, or directly dictate care for a person. Machine learning models generate hypotheses that must be empirically tested in wet-lab environments and subjected to formal clinical and regulatory scrutiny before they can be applied as therapies.
Computational & Experimental Synergy
The search space for bioelectric parameters, including frequencies, amplitudes, duty cycles, and waveforms, is large. Exhaustive manual testing is impractical.
MATRIX Discovery uses model-driven methods to rank candidate waveforms against available evidence and experimental data. This approach mirrors modern computational drug discovery paradigms, where neural networks narrow vast chemical spaces down to testable candidates. AI acts as a high-speed filtration and patterning tool; human scientists design the experiments and interpret the outcomes.
Evidence graph and candidate ranking
A research knowledge graph organizes relationships among signals, tissues, indications, and evidence. Models use that context to rank candidates for scientist review, not to diagnose a person or prescribe a protocol.
Controlled experimental validation
Scientists translate selected candidates into controlled laboratory experiments with review gates, controls, replication, and statistical analysis. Results, positive or negative, inform the next research cycle.
Separate from MATRIX System
MATRIX Discovery is software for laboratory R&D. MATRIX System is the separate, proposed whole-body mat hardware concept for a future Scan → Diagnose → Compute → Treat → Measure → Adapt → Treat again loop. The System, closed-loop sensing and control, and diagnostic front end remain at concept-and-design stage. Diagnose is a future research goal, not an existing diagnostic capability. MATRIX System is not FDA-cleared, not an approved therapy, and not available for clinical or consumer use.
Computational Paradigms in Discovery
Stokes et al. (Cell, 2020) demonstrated the power of deep learning to rapidly identify novel therapeutic candidates from massive datasets. Electrome applies a parallel philosophy to the bioelectric signal space, using AI to direct empirical research rather than replacing it.
Cell 2020 (PMID 32084340)Validation Gates & Regulatory Context
Generating a prediction is only the first step. Translating computational output into clinical value requires adherence to strict verification frameworks.
Good Machine Learning Practice (GMLP)
We align our internal development with the guiding principles established by global regulators for AI in medical development. This includes maintaining separation between training and testing datasets, ensuring models perform consistently across intended populations, and prioritizing transparency in how predictions are derived.
FDA GMLP Guiding PrinciplesAI in Therapeutic Development
The use of artificial intelligence in R&D must be properly contextualized. AI assists in recognizing patterns and optimizing parameters, but it is not a substitute for safety and efficacy trials. Regulatory bodies evaluate the final therapeutic intervention, not the algorithm that suggested it.
FDA AI for Drug DevelopmentCollaboration Context
We build tools to empower researchers, not replace them. Our AI-native infrastructure is designed to interface with academic institutions, clinical partners, and biological engineers who share our commitment to rigorous, evidence-based exploration of the electrome.