Genomind

The assets are a laboratory and a curated knowledge base, not a model. Genotyping is by quantitative polymerase chain reaction across a defined gene panel, and the software layer maps that genotype against a medication list using interpretation rules drawn from published pharmacogenetic guidance, flagging gene to drug and drug to drug interactions and proposing alternatives.

The company describes its medical affairs team holding monthly literature reviews to keep the report content current, which is human curation performed well and is the opposite of a learned system. No machine learning or artificial intelligence claim was located in the company's own material, and the restraint is credited: a laboratory that describes itself as a laboratory gives a buyer a more accurate picture than several vendors in this category manage. Graded on the mechanism, which is the moat is the dataset pattern combined with a regulated laboratory operation.

Results reach a prescriber who decides, and the company states the limitation of its own product more plainly than most vendors in this index manage. Its provider material says directly that the field cannot determine the perfect drug, and that pharmacogenetics should be used alongside clinical judgement, current symptoms, past treatment response, family history and treatment goals rather than in place of them.

That is a vendor telling a clinician not to over rely on its output, in a segment where over reliance is the obvious failure mode, and it is worth crediting explicitly. The laboratory also states a considered policy for uncertainty, listing novel variants as indeterminate on the report and following up by direct communication rather than forcing an interpretation. Held at B because no data is published on how often clinicians change a prescription after a report, which is the measure that would show whether the caution is heeded.

The laboratory half is transparent and the interpretation half is not. Method, gene list, analytical accuracy above 99 percent and participation in a formal proficiency testing programme run by the College of American Pathologists are all publicly registered, which lets an outside party check the assay. The evidence sources behind the guidance are also named, including the Food and Drug Administration's tables of pharmacogenetic associations and of biomarkers in drug labelling.

What is not published is the interpretation logic itself, the step where a genotype becomes a recommendation to prefer or avoid a particular drug. That step matters more than it might appear, because independent work comparing commercial pharmacogenetic decision support tools has found that different products can produce different medication recommendations from the same genotype, which means the interpretation layer, not the assay, is where products actually differ.

The inputs are named and the operating chain is not. Evidence sources behind the guidance are identified specifically, including the regulator's tables of pharmacogenetic associations and of biomarkers in drug labelling, so a buyer can see which corpus the recommendations are built from and read it themselves, which is a real answer to the data half of this axis and better than most vendors manage. The assay method and gene list are registered publicly alongside it.

What is absent is everything about who operates the system: no model or model family, no hosting arrangement, no sub processor list, and no description of the interpretation software itself, including whether any part of it is licensed rather than built. The content type raises the stakes above the ordinary case and deserves its own questions.

Genetic data is permanent, cannot be revoked once disclosed, and carries implications for biological relatives who consented to nothing, and this company holds it as the laboratory of record rather than as an interpreter of someone else's result, with collection extending into the patient's home. No retention schedule, no statement of whether genotype data is retained after reporting, and no description of secondary use for research or product development was located. The company also licenses its reporting suite to other laboratories, so ask whose data flows where in that arrangement, and for a retention and secondary use position in writing.

This grade reflects what the evidence shows rather than how well it was gathered, and those point in opposite directions here. The company sponsored and co authored a multicentre randomised controlled trial of its own test in outpatients with non psychotic major depressive disorder, blinded to both participants and raters, randomising 304 patients across two arms over eight weeks, published in Depression and Anxiety in 2020 with investigators from Massachusetts General Hospital and Harvard alongside two company employees.

The conclusion was that pharmacogenomic testing using a panel of pharmacokinetic and pharmacodynamic variants was not associated with significant improvement in the primary efficacy outcome when providers were unconstrained by the assay results. A later review of the field records that all three large controlled studies of pharmacogenomically guided antidepressant selection failed to show significant benefit.

Supporting evidence is weaker in design and more favourable in result: a company sponsored case control analysis of healthcare utilisation and cost, and citation of a third party meta analysis. Commissioning the trial was the strongest evidentiary act available to any vendor in this category, and the trial did not support the product's central claim.

The company holds genetic data, which is permanent, cannot be revoked once disclosed and carries implications for biological relatives who consented to nothing, and it holds that data as the laboratory of record rather than as an interpreter of someone else's result. Sample collection extends to the patient's home as well as the clinician's office, and the company publishes patient facing reports, so the data path reaches consumers directly.

Retrieval located a privacy policy and a consumer privacy notice offering an opt out of information sale, which indicates a real privacy programme, and located no retention schedule, no statement of whether genotype data is retained after reporting, and no description of secondary use for research or product development. A company that licenses its reporting suite to other laboratories should also state clearly whose data flows where in that arrangement.

Two retrieval passes located no privacy rule statement, no business associate agreement terms, no execution path and no compliance page, though a general privacy policy and a consumer privacy notice are published. The scope of this grade should be read carefully.

As a clinical laboratory the company is a covered entity in its own right rather than only a business associate, so its obligations arise directly rather than through a customer contract, and its electronic health record integration and health plan and long term care relationships all imply executed agreements. What is absent is any published expression of the terms, which is what a prospective institutional customer needs before contact. For a business holding genetic results this is the highest value document it could publish.

Retrieval located no service organisation controls report, no HITRUST certification, no ISO 27001, no trust centre, no penetration testing cadence and no vulnerability disclosure programme. A distinction needs drawing because the company holds accreditations that are easy to mistake for security assurance.

Laboratory accreditation and participation in a formal proficiency testing programme run by the College of American Pathologists demonstrate that the assay produces accurate results, and say nothing whatever about how the resulting data is protected once it leaves the bench. Those are different questions assessed by different bodies against different standards, and a buyer should not accept one as evidence of the other.

The test operates as a laboratory developed test under clinical laboratory regulation rather than through device clearance, which is the established route for pharmacogenetic panels and is a legitimate position rather than a gap.

The interesting feature is that the company grounds its interpretation in the Food and Drug Administration's own tables of pharmacogenetic associations and of biomarkers in drug labelling, so the agency's published positions supply the evidence base for a product the agency has not reviewed. A buyer should understand the boundary that creates.

The agency's tables describe associations it considers supported for specific gene and drug pairs; a commercial panel typically reports on more genes and more drugs than those tables cover, and the report format does not necessarily distinguish guidance that rests on regulatory consensus from guidance that rests on weaker literature.

Retrieval located no performance reporting by ancestry, age or sex, no bias assessment and no coverage statement. Two exposures are specific rather than generic. The frequency of metabolising enzyme variants differs substantially between ancestral populations while pharmacogenetic reference data has historically been derived disproportionately from people of European ancestry, so the completeness of any panel's guidance varies by whose genome it reads, and nothing in a report distinguishes a well characterised result from a thinly evidenced one.

Separately, this panel reports on pharmacodynamic genes affecting drug response as well as pharmacokinetic genes affecting metabolism, and the published evidence supporting pharmacodynamic gene associations with treatment response is substantially weaker and less consistent than the metabolism evidence, a distinction a uniform report format tends to flatten. Publishing which gene and drug pairs rest on strong evidence and which do not would be the most useful single disclosure available.

The externally verified half of this product is verified properly and it is not the half where products differ, which is the whole assessment. The laboratory side is checkable by an outside party: the method and gene list are registered, analytical accuracy above 99 percent is published, and the laboratory participates in a formal proficiency testing programme run by a professional pathology body, which is ongoing external examination by a third party rather than a self report.

That is a real accountability structure and few vendors in this index have anything comparable. The interpretation layer has none of it. The step where a genotype becomes a recommendation to prefer or avoid a particular drug is not published, and independent work comparing commercial pharmacogenetic decision support tools has found that different products can produce different medication recommendations from the same genotype.

That is the finding a buyer should sit with, because it means the assay is not where the risk lives. Two laboratories can both be analytically correct and still tell a prescriber different things, so the accuracy figure that is published describes the part that varies least between vendors while the part that varies most is undisclosed and unexamined.

Ask which evidence hierarchy the interpretation applies, how conflicts between sources are resolved, how often guidance changes when the sources update, and what the company commits to when a recommendation is wrong.

Ordering and result review are available inside Epic, so a prescriber can request the test and read the interpretation without leaving the record system, which for a laboratory service is the integration that determines whether it is used at the moment of prescribing or filed and forgotten.

Distribution extends through a diagnostics partner network that resells access to the testing, and through a reporting suite licensed to other laboratories, which places this company's interpretation layer inside competitors' workflows and is an unusual structural position worth understanding.

Held at B because no FHIR conformance statement, marketplace certification detail or public interface documentation was located, and because integration beyond the single named record system is not described.

Delivery combines a physical laboratory operation with a cloud platform for ordering, tracking and result interpretation, and sample collection can occur in a clinician's office or at the patient's home, so the operational footprint spans logistics as well as software.

Retrieval located no named hosting provider, no cloud region, no data residency commitment and no description of how long physical samples are retained after analysis, which for genetic material is a distinct question from data retention and one a patient may reasonably care about. The company describes international partners without describing how data crosses borders in those arrangements.

This vendor publishes more about the commercial arrangement than most in the category, without publishing a price. Patient facing terms are stated concretely: eligibility for health savings and flexible spending accounts, flexible payment plans and financial assistance, which for a test a patient may part fund is the information that actually determines access.

Institutional material states that a genetics based medication management approach is available to nursing facilities at no additional cost to the facility, which discloses the billing model even though it leaves the payer arrangement implicit. What is missing is list price, the reimbursement position across payer types, and the commercial terms of the reporting suite licensed to other laboratories. Graded C for publishing terms that matter to the person paying, while the institutional economics stay opaque.

The clinical centre of gravity is mental health, where the panel's pharmacodynamic genes and the company's founding purpose both sit, and the registered test lists conditions including major depressive disorder, anxiety and attention deficit hyperactivity disorder. Coverage extends past that focus, with the panel addressing more than 700 medications across therapeutic areas.

Buyer coverage is unusually wide for a laboratory: individuals, prescribers, long term care facilities, accountable care organisations, employers, health plans, other laboratories and international partners, plus a contract supporting testing for United States active duty service members, which is a demanding procurement environment.

Held at B because deployment is essentially United States based, because the evidence concentrates in the one indication where the controlled trial was negative, and because coverage claims for medication counts vary between the company's own pages and its partner material.