Nym Health

The engine is the product. Nym sells autonomous medical coding and nothing else: its Clinical Language Understanding technology combines proprietary machine learning models with rules based clinical ontologies to reconstruct the encounter narrative and assign ICD-10-CM/PCS and CPT codes, routing confidently coded encounters to billing with no human step. There is no coding bureau underneath it.

What the system may do and the threshold that governs it are both published. Nym's technology page states encounters are routed directly to billing when coding confidence exceeds 95 percent, and that the engine otherwise runs in the background as the first coder inside the existing workflow; every code carries an audit trail with its reasoning, and guideline updates are implemented by Nym's compliance and clinical teams.

The path for encounters below the threshold (return to the customer's coders) follows from that rule. Confidence is not accuracy. Nym publishes neither audited accuracy above the threshold nor the share of charts that fall below it, by specialty.

The approach is named in detail without the training data behind it. Nym describes Clinical Language Understanding: proprietary machine learning models combined with rules based clinical ontologies encoding AMA, CMS and WHO guidelines, in named stages (semantic representation, context extrapolation, ontological linking, narrative construction), handling negation, temporality and subjectivity; guideline updates are implemented by its compliance and clinical teams. The training data for the machine learning components is not characterized, and no version or change log is published.

A structurally short chain. Nym's models are proprietary and paired with its own rules based ontologies, built by its own team, and its 2021 post names AWS as the host; no third party language model is described. No subprocessor list is published.

Named customers without outcome figures. Nym names Inova (a case study on emergency department coding, with its health information management lead saying coding accuracy was maintained while efficiency improved) and Ochsner Health (a press release on adopting the engine for emergency department coding), and reports 30+ health systems and physician groups. A 2021 post cites 98 percent accuracy with no method. No dated outcome figure with a baseline is published on the home, engine or technology pages.

General assurances that do not answer the questions AI raises. Nym's 2021 post on its AWS environment says all PHI is encrypted with AES-256, that retention follows purge mechanisms such as storage expiration without stating periods, and that autonomous coding means charts are 'never having to be seen by other people for coding purposes'; encounters below the 95 percent confidence threshold still go to human coders, so that is true only of the charts the engine codes. Nothing published states whether customer charts train or refine the models, or any de identification step. Safety engineering is not published.

Compliance is claimed without the document behind it. Nym's 2021 post says HIPAA compliance is built into the product, that it chose AWS as its cloud platform and HIPAA compliance partner, and that PHI never leaves the United States; its privacy policy covers the websites and EEA data transfers, not protected health information. No business associate statement or agreement is published.

Controls are asserted with nothing independent behind them. Nym's 2021 post describes AES-256 encryption of PHI and hosting on AWS; AWS's attestations are the host's. No SOC 2 report, HITRUST certification, ISO 27001 certificate, penetration testing statement or trust center appears on the home, engine, technology or privacy pages, and a search found none on Nym's own sites.

No device claim is made and none applies: coding is an administrative determination. The regime that governs is coding compliance, and Nym's design addresses it: codes follow rules based ontologies built from AMA, CMS and WHO guidelines, its compliance and clinical teams implement guideline updates, and each code carries an audit trail. Liability for an incorrect autonomously coded claim still sits with the billing provider under the False Claims Act where a government payer is involved.

A named process governs the rules, with no evaluation of outcomes published. Nym assigns codes through rules based ontologies encoding AMA, CMS and WHO coding guidelines rather than learning a customer's historical coding intensity, which anchors the engine to published guidance and limits drift toward a health system's past habits; its technology page says Nym's compliance and clinical teams implement every coding guideline update into the engine. No bias or subgroup evaluation, audit of code distribution or accuracy by specialty is published.

Mechanisms exist and nothing stands behind them. Every coded encounter carries an audit trail showing which clinical findings informed each code, which guidelines were applied and the reasoning, which lets a provider defend or correct a code on audit or appeal. Nym publishes the rule that sends encounters to billing (coding confidence above 95 percent), which tells a buyer which charts no person sees.

What is missing is anything standing behind the output: a 2021 post cites 98 percent accuracy without method or denominator, and no warranty, indemnity or correction commitment is published, while the billing provider carries the False Claims Act exposure for an autonomously coded claim.

Integration is claimed with no system named. Nym says its engine layers onto the existing enterprise stack and runs in the background, coding first inside the current revenue cycle workflow and routing coded encounters to billing. The company names no EHR, billing system, standard or listing.

The hosting and residency are stated, with isolation left open. Nym's 2021 post on its AWS environment states the engine runs on AWS, that patient health information never leaves the United States, and that PHI is encrypted at rest with AES-256. No tenancy model, region detail or current statement is published. Nym's research and development is based in Israel, and nothing says whether any support or engineering access to PHI happens from outside the United States.

No price is published and the posture is discoverable in part. Nym sells an autonomous coding engine to health systems and physician groups across six specialties, integrated into the existing workflow. Charging per chart processed, per chart autocoded, or as a platform fee gives very different economics when the autonomous rate varies by specialty and documentation quality, and none is stated.

Coverage is named with validation behind part of it: six specialties and service lines across outpatient and inpatient settings (emergency medicine, radiology, outpatient surgery, outpatient visits, inpatient professional services and urgent care) for health systems and physician groups. Named deployments support emergency department coding (Inova, Ochsner Health); no autonomous rate or accuracy is published per specialty, which is where performance will differ.