Precision Analgesia is designed from the ground up to meet all four FDA criteria for Non-Device Clinical Decision Support — a self-assessed classification, pending confirmatory review by regulatory counsel, aimed at avoiding 510(k) clearance requirements while maintaining the highest standard of clinical transparency.
Non-Device CDS (self-assessed)
21st Century Cures Act — pending counsel review
Not Anticipated
Pending confirmatory classification memo
Yale — Planned, July 2026
Addiction Medicine Fellowship
Software is exempt from device regulation if it meets all four criteria. A caution finding requires additional counsel review; it does not automatically disqualify classification.
The software does not acquire, process, or analyze medical images (X-ray, MRI, CT), in vitro diagnostic signals, or physiological signals from a connected device. All inputs are entered manually by the clinician or pulled from structured EHR fields.
No image processing. No waveform analysis. No device connectivity. All inputs are discretized clinical parameters.
The product is explicitly designed for, and marketed exclusively to, licensed healthcare professionals — attending physicians, nurse practitioners, physician assistants, and clinical pharmacists operating within institutional EHR environments.
HCP-only design. Mandatory clinical responsibility acknowledgment on every launch. No consumer-facing features.
Every recommendation is accompanied by the full formula, data inputs, intermediate calculations, and the source guideline (CDC 2022, NCCN 2025, AGS Beers Criteria). The "Algorithm Transparency" and "Citations" interfaces show the complete derivation — not just the output.
Full derivation displayed. Every recommendation hyperlinked to published source. Zero black-box outputs.
The primary classification criterion. The statute requires that the software "enables such health care professional to independently review the basis for such recommendations so that it is not the intent that such health care professional rely primarily on any such recommendations to make a clinical diagnosis or treatment decision."
No blocking behavior remains in the engine: every safety gate surfaces a specific, actionable, guideline-cited recommendation that the clinician reviews and can accept or override, with the override logged. The only non-overridable rule is a pediatric gate blocking codeine/tramadol prescribing under age 12, per FDA Black Box warnings — regulator-originated, not engine judgment. A confirmatory classification memo from specialized health IT counsel is planned as Stage 3; it has not yet been commissioned or obtained, and the classification below should be read as self-assessed, not settled.
Why deterministic, traceable logic enables Non-Device classification while AI/ML approaches cannot.
Every MME conversion uses the published equianalgesic ratio from the CDC 2022 guideline. The formula, input values, and intermediate steps are displayed inline — not hidden behind a result card. A clinician can perform the same calculation on paper and arrive at the same answer.
Every clinical recommendation is hyperlinked to its source guideline. CDC 2022 for opioid dosing. NCCN 2025 for cancer pain. AGS Beers Criteria 2023 for geriatric thresholds. ASAM 2020 for OUD protocols. A clinician can independently verify any recommendation against primary literature in under 30 seconds.
No machine learning. No neural networks. No probabilistic inference. Every output is fully deterministic — given the same inputs, the same calculation runs every time. This eliminates the "black box" problem that prevents AI/ML CDS tools from achieving Non-Device classification.
The software surfaces a specific, actionable, guideline-cited recommendation with full reasoning at the moment of prescribing; the clinician reviews it and can accept, adjust, or override it. The mandatory disclaimer on every session launch reinforces this. Every override is logged, recorded in the generated clinical note and the zero-PHI safety log. Only the pediatric gate (codeine/tramadol under 12, per FDA Black Box warnings) is non-overridable.
The classification decision is not just regulatory — it is strategic.
| Dimension | Non-Device CDS (Our Path) | SaMD 510(k) |
|---|---|---|
| Time to Market | Pending confirmatory counsel review | 12–24 months |
| Regulatory Cost | $10–25K (counsel + memo) — illustrative estimate | $800K–$2M |
| Clinical Changes | Rapid update on evidence | New 510(k) submission per change |
| Ongoing Compliance | Internal QMS documentation | FDA post-market surveillance |
| AI/ML Features | Not available (by design) | Permitted under 510(k) |
Three features were identified as Criterion D risks and resolved by design in June 2026 — converted from blocking behavior to advisory recommendations with documented overrides. They are kept here as the record of how each risk was retired, verified across the product's 250+ iOS clinical validation scenarios and 120+ backend pytest tests.
The risk: for morphine in dialysis, the engine replaced the dose with a terminal AVOID — evidence-based (M3G/M6G accumulation toxicity), but blocking behavior that invited Criterion D scrutiny.
AVOID remains the default recommendation with full evidence display (NKF guidelines; M3G/M6G neurotoxicity). A two-step "Review Contraindication & Override" pathway reveals the unadjusted equianalgesic reference, explicitly labeled OVERRIDE — CONTRAINDICATED. The override is recorded in the generated clinical note, logged (zero-PHI), and resets whenever renal status changes. This mirrors how EHR-native advisories handle renally-cleared drugs.
The risk: for patients under 18, the calculator locked its entire interface — a constraint that went beyond the underlying FDA mandates and constituted engine-originated blocking.
The pediatric advisory now offers a two-step documented override into adult-reference mode: calculations proceed with a persistent warning that conversion factors are adult-validated (CDC 2022) and weight-based formulary verification is required. The override resets when the age changes. The narrow FDA-mandated contraindication — codeine/tramadol under age 12 (FDA Drug Safety Communication 2017) — remains enforced without override, because it is regulator-originated, not engine judgment.
The risk: the OUD Consult pathway routes by COWS score (ASAM 2020 + SAMHSA thresholds) to prevent precipitated withdrawal. The engine never blocked — it always produced an actionable plan — but the routing could be characterized as restricting clinical judgment.
Every below-threshold plan now states explicitly that the routing is an evidence-based recommendation, names the alternative (standard/high-dose induction per clinical judgment, with re-scoring and precipitated-withdrawal management guidance), and cites its sources inline (ASAM 2020; Herring 2021; SAMHSA 2021). "Wait and Assess" lists the Low-Dose Initiation alternative rather than a bare instruction to wait.
Before any EHR BPA integration with a hospital partner, Lifeline Medical Technologies will commission a formal FDA classification memo from specialized health IT counsel. With the advisory-override design in place, this is a confirmatory memo on a clean Non-Device CDS profile — not a defense of ambiguous blocking behavior — which materially reduces its cost and timeline. Hospital compliance teams and Epic certification reviewers expect counsel-signed classification documentation; this memo is that artifact.
Five stages from internal development to live hospital deployment. The founder's Yale addiction medicine fellowship (began July 2026) anchors the academic validation pathway.
iOS app: 28,955 lines of clinical logic across 94 files (Swift 6, iOS 17+). 250+ automated clinical validation scenarios in the iOS validation suite, plus 120+ EHR backend pytest tests. A 10-rule deterministic safety engine, verified across both suites, plus OIRD risk scoring (PRODIGY), MME threshold logic, CDC naloxone co-prescribing prompts, and pharmacogenomic guidance. AGS Beers Criteria 2023 geriatric thresholds implemented.
Founder is in Yale School of Medicine's addiction medicine fellowship (began July 2026) — the app's core clinical domain (OUD consult, MOUD induction, COWS, buprenorphine protocols). Yale's addiction medicine faculty are targeted as potential co-investigators for a quality improvement validation study, expected/anticipated to qualify for IRB-exempt status, subject to Yale's own IRB determination. Clinical logic review against institutional protocol standards.
Commission a confirmatory FDA classification memo from specialized health IT counsel on the advisory-override architecture (all former hard stops resolved by design, June 2026). Establish Quality Management System documentation and pre-deployment checklist for EHR partners.
Planned 90-day observational quality improvement study at a target academic medical center — no site or partner is confirmed as of this writing. Proposed safety metrics: OIRD rate, high-MME prescribing rate, safety-gate trigger rate, documented override rate. Anticipated to qualify for an IRB-exempt QI designation (no PHI retained, no experimental intervention), subject to that institution's own IRB review and determination.
Full SMART on FHIR + CDS Hooks live deployment. Epic sandbox validation → hospital staging environment → limited floor pilot → system-wide rollout. Epic Connection Hub listing application for broader institutional discovery.
Whether you're evaluating integration from a compliance, legal, or clinical informatics perspective — we have detailed documentation on our classification rationale, HIPAA architecture, and quality management approach.