Arine

The learned layer is the entire argument against the incumbent alternative. Medication therapy management has existed as a plan service for over a decade delivered by pharmacists working manual worklists, and the company's stated proposition is that automation, predictive analytics and machine learning make that practice scalable in a way manual targeting cannot.

The platform synthesises clinical, socioeconomic and behavioural data to prioritise members, generate personalised care plans and identify prescribing aberrations, and the company describes a feedback loop that links risk to intervention to measured impact and feeds that back into targeting. A named engine sits behind it and the company employs a head of artificial intelligence who publishes on the subject.

Remove the model and what remains is a worklist somebody has to prioritise by hand, which is the product this one was built to displace. One scoping note for diligence rather than for the grade: with several hundred employees, a buyer should establish how much of the delivered value is platform and how much is clinical operations staffing.

Two distinct oversight questions arise here because the platform points in two directions. Member facing recommendations reach a clinical pharmacist who reviews them and a prescriber who must act, which is the conventional and sound chain. The prescriber facing product is different and is worth a buyer's attention.

It identifies prescribing aberrations and enables a health plan to send targeted education and recommendations to the clinicians who wrote the prescriptions, so the model's output is aimed at a professional's practice pattern rather than at a patient's regimen, and it is delivered by an organisation that also pays that professional's claims.

Nothing published describes how a prescriber contests a finding, what threshold defines an aberration, or whether the analysis is ever reviewed before it is sent. That is not an accusation, since reducing behavioural health polypharmacy and high dose opioid exposure are legitimate aims, but the power asymmetry between payer and prescriber is real and the appeal path should exist and be published.

The company talks about its technology more openly than most payer side vendors and still publishes nothing checkable. Predictive analytics and machine learning are named as the methods, a named engine is referenced in third party company records, and the head of artificial intelligence co authors public material on how plans should evaluate these systems, which is a more serious posture than a marketing page.

What is absent is the substance: no model family, no feature set, no risk threshold, no calibration, no description of how the socioeconomic and behavioural inputs are weighted, and no account of how the continuous feedback loop retrains or what guards against it amplifying its own targeting decisions over time. That last question is specific to a system that learns from the outcomes of interventions it selected.

Nothing identifies any party in the chain: no model or model family, no hosting arrangement and no sub processor list was located in two passes, and no retention schedule or de identification statement was found. Handling controls are certified rather than asserted, with an independent risk based certification covering the platform, which is worth recording and answers a different question from this one. The footprint is the broadest in this category and deserves naming precisely.

The platform combines clinical data, socioeconomic data and behavioural data across an entire plan membership, so it holds determinants a member disclosed for some other purpose, or never disclosed at all and which were inferred or purchased, and it holds them about people who have had no contact with the product.

Socioeconomic and behavioural inputs are also the categories where provenance matters most, because they frequently originate outside healthcare entirely and arrive through data brokers whose own sourcing is opaque. Nothing states where those inputs come from. The pooling question follows: the commercial value of a payer platform grows with the breadth of populations it has learned from, and nothing states whether one plan's member data informs models used for another. A plan should establish contractually whether its members' data improves a competitor's targeting. Ask that, ask where the non clinical inputs originate, and ask for a sub processor list.

Results are extensive, named to real clients, and entirely self reported. The company publishes savings above 1,500 dollars per engaged member per year with reductions above 40 percent in inpatient admissions and above 20 percent in emergency department visits in an Oklahoma Medicaid population, savings up to 4,300 dollars per engaged member per year alongside a halving of behavioural health polypharmacy and a 20 percent reduction in average daily high dose opioid use with Magellan Health, and five star performance across triple weighted adherence measures for a Medicare quality programme.

One genuine methodological credit belongs on the record: the company states that its outcome and cost savings figures rest on matched pair statistical analyses, which names a method rather than asserting a number, and matched pair designs are a reasonable approach to a non randomised population intervention.

Held at C because no peer reviewed publication, independent evaluation or external replication of any of it was located, and because the entity reporting the improvement is also the entity that selected which members to engage.

Handling controls are certified rather than asserted, with a HITRUST risk based two year certification covering the platform and a stated commitment to safe and responsible use of health data. The data footprint is the broadest in this category and deserves naming precisely: the platform combines clinical data, socioeconomic data and behavioural data across an entire plan membership, which goes well beyond the medication record and into determinants a member never disclosed for this purpose.

Retrieval located no retention schedule, no de identification statement and no description of whether member data informs models used for other plans, which matters more here than elsewhere because the commercial value of a payer platform grows with the breadth of populations it has learned from. A plan should establish contractually whether its members' data improves a competitor's targeting.

The HITRUST risk based two year certification is the substantive element here, since that framework incorporates federal and state regulatory requirements including the HIPAA security rule into an audited control set, and the certification is announced with its tier specified rather than as a generic claim.

The company operates exclusively with covered entities, since health plans are its only customer type, so the business associate relationship is the entire commercial basis of the business. Held at B for the pattern this index applies consistently: the agreement itself is not published, so breach notification timelines, subcontractor flow down and data return obligations on termination are not readable before contact. For a vendor holding socioeconomic and behavioural data on entire plan memberships, publishing those terms would be a meaningful differentiator.

A HITRUST risk based two year certification, attained in July 2024, is the most demanding tier of that programme and represents a real and expensive external assessment rather than a self attestation. The company announced it with the tier named and framed cybersecurity as a priority from the outset, which is the correct level of specificity.

Held at B rather than higher because the surrounding apparatus is thin by comparison: no service organisation controls report was located, there is no trust centre where a security team can request or review documents, no penetration testing cadence is published, no subprocessor list is available and no vulnerability disclosure programme exists. The certification is strong and a prospective customer still has to open a sales conversation to see anything behind it.

No clearance was located and none would be expected, since the product stratifies a population and proposes interventions to licensed pharmacists rather than computing a dose or acting on a patient. The regulatory surface that actually matters for this vendor is a different one and it is moving.

Plan side artificial intelligence is under active state legislative attention in the United States, with several states restricting or conditioning the use of algorithms in coverage and utilisation decisions, and while this product supports clinical intervention rather than denial, it is a model operated by a payer that shapes what care a member receives. A plan deploying it inherits whatever obligations that legislation creates. The company publishes no position on its regulatory classification or on how it distinguishes clinical targeting from utilisation management, and both should be established before deployment.

The gap is unusually pointed here because the company engages with the subject publicly. Its chief executive has published on health inequities and the platform explicitly ingests socioeconomic data, so the relationship between social determinants and who receives an intervention is understood inside the business rather than overlooked. What is missing is measurement.

Retrieval located no performance reporting by race, ethnicity, age, language, dual eligibility or geography, no bias assessment, no governance documentation and no post deployment monitoring statement. A model that uses socioeconomic and behavioural data to decide which members a scarce pharmacist calls is making a distributional decision on every run, and publishing thought leadership about equity is not the same as publishing how the model performs across the groups it sorts. The data to answer it exists across multiple named Medicaid and Medicare populations.

The posture is more serious than a marketing page and the substance is absent, which produces the asymmetry this index has recorded elsewhere. The methods are named as predictive analytics and machine learning, a named engine appears in third party company records, and the head of artificial intelligence co authors public material on how health plans should evaluate systems of this kind.

Publishing guidance on how to evaluate these products while publishing nothing evaluable about your own is the same shape found on other records here, and it is worth stating as an asymmetry a buyer can act on rather than as hypocrisy, because a company that authored the test is better placed than most to pass it.

What is missing is everything the guidance would ask for: no model family, no feature set, no risk threshold, no calibration, and no description of how the socioeconomic and behavioural inputs are weighted against clinical ones, which is the question that determines who gets prioritised. One further gap is specific to the architecture and deserves its own answer.

The platform is described as running a continuous feedback loop, and a system that learns from the outcomes of interventions it selected can entrench its own targeting: members it never surfaced generate no outcome data, so the model has no evidence it was wrong about them. Nothing describes what guards against that. Ask for the weighting, calibration by subgroup, and how the retraining loop is protected against reinforcing its own selections.

The integration surface for a payer platform is claims, pharmacy and eligibility feeds rather than a clinical system, and the company describes ingesting clinical, socioeconomic and behavioural data without naming the pipes. Retrieval located no named electronic health record certification, no marketplace listing, no FHIR conformance statement and no public API.

One integration is worth noting because it crosses into the clinical workflow: a partnership with a drug price transparency company brings cost information toward the point of prescribing, and the prescriber facing analytics product delivers to clinicians rather than to plan staff.

Held at C because the mechanism by which recommendations reach a prescriber's workflow is undescribed, and for a product whose value depends on a clinician acting, that path is the one a buyer should interrogate first.

The platform is delivered as configurable hosted software to health plans and is described as scaling across broad populations, which is as much as public material establishes about its architecture. Retrieval located no named hosting provider, no cloud region, no data residency commitment and no statement about environment separation between plan customers, which is the specific question that matters when a single vendor holds membership level data for competing plans in the same market. The HITRUST certification implies controls over these areas were assessed, but assessment is not disclosure and a plan's security review will still start from a blank page.

No price, unit, pricing basis or contract shape is published, and the return figures are expressed per engaged member per year, which is a unit that answers the value question while leaving the cost question untouched. A structural disclosure matters more than the missing price and belongs on the record as structure rather than accusation.

A Medicare Advantage health plan appears among the company's investors, and the product improves the Star Ratings that determine Medicare Advantage bonus payments, so an organisation of the same type as the customer holds equity in a vendor whose output raises that customer type's revenue from a federal programme. That is commercially ordinary and it is the kind of alignment a buyer should know about, particularly a competing plan evaluating the same platform. Ask who else on the cap table participates in the market being sold into.

Coverage is wide across payer types and narrow in every other respect. Named and described customers span national plans, Blue Cross Blue Shield plans, Medicaid, Medicare Advantage and special needs populations, with specific programmes at the Oklahoma Health Care Authority in Medicaid, Magellan Health in behavioural health, VNS Medicare and MDX Hawaii, so the platform has been configured for structurally different memberships, benefit designs and quality frameworks.

Clinically the reach follows the membership rather than a specialty, covering chronic disease, behavioural health and opioid exposure. Held at B because the product is exclusively payer side with no provider or health system deployment described, coverage is United States only and tied to American programme structures such as Part D measures, and the member populations skew heavily to government sponsored programmes.