Substrate MDA simpler way to manage metabolic care
Substrate MD reconciles scattered reports, sorts the work that needs attention, and brings evidence, plans and follow-through into one patient workspace. AI can prepare; a clinician reviews and approves.
Controlled deployments only. Real patient data remains disabled until the clinic and operator complete the activation gates.
Inside neither claim. The clinic target and the reporting laboratory’s interval remain separate and visible.
The coordination gap
Laboratory reports, PNOĒ, body-composition systems, CGM, sleep studies, intake forms and the EHR each hold part of the picture. Reconstructing what is current, comparable and actionable should not consume the visit.
Inside the product
SubstrateMD is not only a report reader or an AI summary. It is the working surface that connects clinic-wide attention, patient-level evidence, accountable decisions and follow-through.
Start with data conflicts, results outside target, overdue follow-ups and plans due for review—ranked across the clinic instead of buried patient by patient.
A ranked queue supports attention; it never replaces clinical priority-setting.Open the priorities, metabolic score, care areas, current plan and open work first. Measurements, source evidence and the full clinical detail stay one deliberate step deeper.
A summary never erases the evidence or the uncertainty behind it.Resolve low-confidence extraction, unmapped measurements, duplicate records and conflicting sources with the context needed to decide—not a warning with nowhere to act.
Unknown or incompatible data is held for a person, not silently normalised.Carry goals, contraindications, reviewed protocols, versioned care plans, patient questions and assigned follow-ups through one accountable coordination path.
Drafts remain drafts until a named clinician approves the decision.The working model
Bring reports and patient information together. SubstrateMD flags missing details and conflicts instead of guessing.
Start with the most important findings, then open the measurements and sources behind them when you need to.
AI can prepare a visit brief and care-plan starting point. A clinician keeps, changes or rejects every suggestion.
Give each question and follow-up an owner and due date so important patient work does not get lost.
Evidence before decoration
Every adopted value retains its reported and canonical form, observation date, source location, parser identity, trust state and lineage. Conflicts remain visible objects; calculated values name their actual inputs.
Clinical safety boundary
Targets, trends, safety checks and treatment context use visible, versioned rules—not an AI guess.
AI may draft a brief or plan. It cannot diagnose, choose between conflicting values, or approve care.
Missing units, unclear dates, old measurements and source conflicts stop for human review instead of being hidden.
Supported input paths
The deterministic pipeline handles supported PDF, CSV, TSV, XML/FHIR, C-CDA, JSON, PNG, JPEG and HEIC inputs. Quest/generic laboratories and published layouts are covered; Jane batch, DEXA, CGM, sleep and exact InBody variants remain “validation in progress” until de-identified real samples pass.
Privacy and security
Per-clinic encrypted databases, MFA-gated clinical capability, bounded break-glass access, tamper-evident audit chains, encrypted backups and strict no-PHI audit detail are implemented controls. “HIPAA compliant,” “tamper-proof” and “fully integrated” are not claims this site makes.
Read the precise security posture →Deployment evidence
The activation workroom turns the current setup into five staff-only evidence packs without patient data. It prepares protocol review, target change control, source and vendor evidence, validation and AI-governance material, and the complete launch register.
A deliberate first conversation
We are working directly with metabolic practices while deployment, workflow and clinical-validation boundaries are proven.
Request a design-partner conversation