01
CPIC Pharmacogenomics
ScreenMyGene's pharmacogenomics logic is built on the peer-reviewed guidelines of the Clinical Pharmacogenetics Implementation Consortium (CPIC). Rather than flagging a single gene in isolation, the engine reads the patient's active medication list and identifies the drug–gene interactions that actually matter for that person — a clopidogrel–CYP2C19 or an SSRI–CYP2D6 relationship, for example. A pharmacogenomic panel is only surfaced when there are enough clinically actionable drug–gene pairs to justify it, which keeps recommendations defensible and avoids low-yield single-gene orders. The result is guidance a prescriber can act on: which panel to order, and precisely why the patient's regimen supports it — the way a clinical pharmacologist would reason, but in seconds and across the whole chart.
02
ICD-10-CM Validation
Every recommendation ScreenMyGene makes is anchored to a real, documented diagnosis code. The platform validates ICD-10-CM codes to the character, distinguishing specific, billable codes from vague or unspecified ones, and it only qualifies a panel when the chart carries a documented, matching, non-generic code. This closes one of the most common and costly gaps in genetic testing: ordering a test a payer later denies because the medical-necessity diagnosis was missing or too broad. By tying each panel to its supporting code up front, ScreenMyGene turns a recommendation into something order-ready and audit-ready — a defensible link between the patient's documented condition and the test being requested, visible to the clinician before anything is sent.
03
Gene–Disease Validity
Not every gene belongs on every panel. ScreenMyGene curates its panel content against established gene–disease validity frameworks — the kind maintained by ClinGen, GenCC and Genomics England's PanelApp — so the genes it recommends testing reflect relationships with real, evaluated evidence behind them. This prevents the twin failure modes of modern genomics: bloated panels full of genes with weak or disputed disease associations, and narrow panels that miss well-established ones. When ScreenMyGene surfaces a hereditary-cancer, neuromuscular or cardiac panel, the underlying gene list is grounded in curated validity rather than marketing. For a medical director that means the recommendation stands up to scrutiny; for the patient it means a test scoped to genes that can actually explain their presentation.
04
Coverage & Medical Necessity
A genetically appropriate test is only useful if it can actually be ordered and covered. ScreenMyGene aligns its logic to payer coverage policy — Medicare local and national coverage determinations (LCDs and NCDs) and comparable medical-necessity criteria — so recommendations arrive with reimbursement in mind. The engine considers whether the documented indication meets the coverage conditions for a given panel, flagging what qualifies now versus what needs additional documentation. This shifts the coverage conversation from a downstream denial to an upstream decision, saving labs and clinics the rework of appeals and the patient the surprise of an unexpected bill. It is the difference between a clinically interesting suggestion and an order that is ready to be placed and paid.
05
Deterministic Engine
ScreenMyGene is built to be reproducible. Given the same chart, it returns the same recommendations every time — no random variation, no drift between runs. This determinism is essential in a clinical setting: it means results can be validated, audited and trusted, and that two clinicians reviewing the same patient see the same reasoning. It also makes the system testable. ScreenMyGene ships with a regression suite of clinical vignettes that runs the full pipeline and checks output against known-correct answers before any release, so changes that would alter a recommendation are caught before they reach a user. Reproducibility is not a nice-to-have here; it is the foundation that lets a decision-support tool be relied upon in real patient care.
06
Safety Gates
Reasoning about eligibility is not only about when to recommend a test — it is also about when not to. ScreenMyGene applies clinical safety gates that suppress inappropriate recommendations. For epilepsy with a documented structural cause such as hydrocephalus, for presentations better explained by a reversible cause, or where age constraints make a hereditary panel inappropriate, the engine holds back rather than over-calling. These gates encode the judgment a specialist applies almost automatically, protecting patients from unnecessary testing and clinicians from recommendations that would not withstand review. The effect is a system that is appropriately conservative where it should be — surfacing tests when the clinical picture supports them, and staying quiet when it does not.
07
Clinical NLP
Charts are written for humans, not machines. ScreenMyGene's clinical language layer reads free-text notes the way a clinician does, resolving them into structured data: diagnoses (both explicitly coded and inferred from the narrative), medications mapped to their drug classes and gene interactions, family history, phenotype and prior workup. Crucially, it trusts physician-written codes verbatim while applying anti-fabrication safeguards to anything it infers, so the structured chart it builds is faithful to what the clinician actually documented. This extraction is what lets everything downstream work: you can paste a messy note and get an accurate, machine-readable representation of the patient in seconds, without a coder in the loop and without losing the nuance buried in the prose.
08
OCR Ingestion
Real clinical records rarely arrive as clean text. Referrals, outside records and requisition forms come as scanned PDFs and images. ScreenMyGene's optical character recognition lifts the text out of those documents so they can be analyzed like any other note — no retyping, no manual transcription. A clinician can upload a scanned consult letter or an outside genetics report and have it read, structured and reasoned over automatically. Combined with the batch workflow, this means an entire folder of mixed-format patient records can move through the pipeline with minimal manual handling. OCR removes the single most tedious barrier to using unstructured records, turning documents that would otherwise sit in a queue into structured, actionable input.
09
Batch Workflow
Clinics do not see one patient at a time. ScreenMyGene lets a user bring in up to five patients at once — pasted notes or uploaded records — and process them together. Before anything runs, each extracted chart can be reviewed and edited, with labs added and details corrected, so the analysis reflects the real patient rather than an OCR approximation. The batch then runs with live progress and accurate per-patient counts, and every result lands in the analysis history, a single click apart. For a busy lab or genetics service, this turns test triage from a one-by-one chore into a reviewable, repeatable pass over a whole caseload — the same rigor applied to each patient, at the throughput a clinic actually needs.
10
Phenotype Extraction
Diagnosis codes tell only part of the story; the phenotype often carries the rest. ScreenMyGene extracts the clinical features documented in the chart — the signs, symptoms and exam findings that point toward a genetic etiology — and factors them into which panels it surfaces. A note describing pes cavus and distal weakness supports a hereditary neuropathy pathway; Kayser–Fleischer rings and low ceruloplasmin point toward a metabolic workup. By reading phenotype alongside coded diagnoses, the engine reasons more like a geneticist, catching eligible patients whose presentation is documented in the narrative even when the single right code is absent. It is a richer, more faithful reading of the patient than codes alone can provide.
11
Family-History Logic
In hereditary disease, family history is often the trigger for testing — but it has to be weighed correctly. ScreenMyGene treats family history as a genuine clinical signal while respecting a firm rule: it is a supportive factor, not a stand-alone qualifier. A documented personal diagnosis code is required to qualify a panel; a strong family history then strengthens and contextualizes that recommendation rather than manufacturing one on its own. This keeps the platform aligned with how professional-society criteria actually work and prevents over-calling on family history alone, while still ensuring a patient with a meaningful hereditary pattern is not overlooked. The result is recommendations that honor family history without letting it override the need for documented medical necessity.
12
Audit Trail
Trust in a clinical tool comes from being able to see its work. Every ScreenMyGene recommendation is traceable: the panel it surfaced, the documented diagnosis that supports it, the drug–gene interactions or phenotype that drove it, and the safety checks it passed. Nothing is a black box. A medical director can review exactly why a panel was recommended and reproduce that reasoning, and the same transparency underpins compliance and quality review. Because the engine is deterministic, that trail is stable over time — the explanation you see today is the explanation you will see tomorrow for the same chart. This auditability is what lets ScreenMyGene function as decision support a clinician can defend, rather than an opaque recommendation they must take on faith.
13
Consent Generation
ScreenMyGene doesn't stop at deciding which test to order — it helps you act on it. For an eligible panel, the platform can generate a patient-ready informed-consent document tailored to the specific test and clinical context, capturing the purpose of testing, the conditions covered, the possible results and the standard consent language a genetics program requires. Instead of hunting for a template and filling it in by hand for every order, the clinician gets a clean, consistent consent form built from the same structured chart that drove the recommendation. This shortens the path from decision to sample, keeps documentation uniform across a practice, and ensures the consent actually matches the panel being ordered — one less manual, error-prone step between identifying the right test and collecting it.
14
Letter of Medical Necessity
Coverage often hinges on a well-written letter of medical necessity, and writing one for every order is slow. ScreenMyGene generates a real-time letter of medical necessity for a recommended panel, assembling the patient's documented diagnosis, clinical findings, family history and rationale into the structured argument payers expect. Because it draws on the same reasoning that qualified the panel — the matching ICD-10, the drug–gene interactions or phenotype, the applicable coverage criteria — the letter is specific to the patient rather than boilerplate. The clinician reviews and signs rather than drafts from scratch. For labs and practices this removes one of the biggest bottlenecks and denial drivers in genetic testing, turning a task that used to consume a clinician's time into a document that's ready the moment the recommendation is made.
15
NCBI Evidence
Every recommendation should be defensible against the literature, and ScreenMyGene surfaces the evidence directly. For each panel it can provide supporting references drawn from NCBI's biomedical resources — the peer-reviewed and curated sources that underpin the gene–disease and gene–drug relationships behind the recommendation. Rather than asking a clinician to trust an opaque suggestion, the platform points to the published basis for it, so a medical director or genetic counselor can verify the rationale and cite it in documentation or appeals. This keeps ScreenMyGene grounded in current science rather than a static internal list, and gives the recommendation the evidentiary weight that clinical and payer review demand. Evidence isn't an afterthought here — it's attached to the panel.
16
CMS Coverage (LCD/NCD)
Knowing a test is clinically right isn't enough — you need to know whether Medicare will cover it. For each recommended panel, ScreenMyGene checks CMS coverage and tells you whether a Local Coverage Determination (LCD) or National Coverage Determination (NCD) applies, and how the patient's documentation lines up with it. Instead of manually searching the Medicare Coverage Database policy by policy, the clinician sees, per panel, whether coverage exists, what the governing determination is, and what's needed to satisfy it. This turns coverage from a downstream surprise into an upfront fact, letting practices order confidently, document to the right criteria, and avoid predictable denials — the reimbursement reality check attached to every clinical recommendation, at the point of decision.
17
Audio Output
Clinicians don't always have time to read. ScreenMyGene can speak each recommendation aloud, delivering the reasoning in a clear, geneticist-style narration so a provider can absorb why a patient qualifies for a panel while reviewing the chart, moving between rooms, or preparing to counsel the patient. The audio explains the clinical picture — the findings, the eligibility, the rationale — in plain language, without reciting codes or gene lists, so it plays like a colleague summarizing the case. It makes the platform faster to act on and more accessible, and doubles as a way to communicate the rationale to patients and staff. The same defensible reasoning that appears on the card is available as a spoken briefing, on demand.
18
AI Narrative
Behind every analysis, ScreenMyGene produces a written AI narrative — a plain-language summary that ties the whole case together. It explains, for this specific patient, which panels are qualified and potentially qualified, what in the chart supports each, and how the reasoning was reached, mirroring exactly what the recommendation cards show. This gives clinicians a readable story instead of a bare list, makes the logic easy to review and share, and slots directly into documentation. Critically, it is reconciled to the cards, so it never claims something the engine didn't surface — the narrative and the recommendations always agree. For a medical director it's an auditable account of the analysis; for an ordering clinician it's the fastest way to grasp a patient's genetic-testing picture at a glance.