Glytec

Independent peer reviewed reviews classify this system, alongside its direct competitors, as relying on rule based or proportional integral derivative algorithms with static decision rules rather than adaptive machine learning models, and describe the intravenous algorithm concretely: hourly insulin dose equals blood glucose minus a floor value multiplied by a factor beginning at 0.02 and stepping up until glucose reaches target, with the underlying formula traced to a 1982 publication.

That is genuine closed loop control that individualises to the patient through feedback, and it is not learning in any technical sense. The company's own material describes the algorithm as leveraging real time and historical data by learning each patient's insulin sensitivity and anticipating future needs.

Graded on the mechanism rather than the marketing, and the gap between the two is recorded because a buyer comparing this against products that do use learned models will otherwise mistake a control system for one. None of this speaks to whether the software works, which is a separate axis where it does considerably better.

The chain is tightly bounded: a provider orders the protocol, the software calculates each dose, and a nurse administers and confirms it, so the algorithm never reaches the patient without a licensed human executing. Two implementation details strengthen the position. Order set initiation lets the software receive its starting parameters directly from the provider's order rather than from manual re entry, removing a transcription step that is a known error source.

And dose confirmation closes the loop in the other direction: the calculated dose is written to the medication administration record for the nurse to view and chart, and the dose actually given is sent back so the algorithm computes its next recommendation from what happened rather than from what it proposed. Held at B because no override rate, no acceptance figure and no account of what the system does when the confirmed dose diverges materially from the recommended one were located.

The algorithm is knowable, which is unusual, but not because the vendor published it. Independent reviews in the clinical literature set out the control mechanism, the linear formula, the starting multiplier and the escalation behaviour in enough detail that a pharmacist or intensivist can understand exactly how a dose arises, and can trace the approach to its 1982 origin. That is a real transparency benefit and it belongs to the academic literature rather than to the company.

From the vendor a reader gets a functional description that overstates the mechanism, with no published thresholds, no calibration data, no account of how the multiplier escalation was chosen or validated, and no statement of the conditions under which the algorithm should not be used. Graded C: the information exists and a buyer must go outside the vendor to get it.

This is the deepest evidence base in the glycemic segment and among the deeper ones in this category. Published work spans reductions in hyperglycaemia and hypoglycaemia, time to target glucose, length of stay, thirty day readmissions, postoperative complications, nursing time and cost, including a comparison against provider managed subcutaneous basal bolus therapy in the Journal of Diabetes Science and Technology.

Headline claims include a 99.8 percent reduction in intravenous patients experiencing severe hypoglycaemia, readmission reductions of 36 to 68 percent and length of stay reductions up to 3.2 days. Two caveats hold this at B rather than higher and both are visible in the literature itself. Author lists on the pivotal comparative work include multiple company employees alongside the academic investigators, so the most cited studies are not independent.

And a retrospective comparison against a competitor at one health system changed the target glucose range at the same time as the software, from below 120 to below 140, which means the reported difference in hyperglycaemia and hypoglycaemia cannot be attributed to the software alone.

The data footprint is narrow and clinically precise, consisting of glucose values, insulin doses, weight, and the parameters of the ordered protocol, which is far less than most vendors in this category ingest. Delivery is cloud based with integration into the electronic health record and, through the Roche partnership, onto the bedside glucose meter itself, which extends the data path to a point of care device and adds a hardware manufacturer to the chain of custody.

Retrieval located no retention schedule, no de identification statement and no description of whether pooled glucose and dosing data across the install base informs product development. The last question is worth asking given that the company also sells surveillance and analytics on the same data.

Two retrieval passes located no HIPAA statement, no business associate agreement terms, no execution path, no subprocessor list and no compliance page. As elsewhere in this lane the grade describes published posture rather than contractual reality, and here the presumption of a robust underlying compliance programme is stronger than usual, since the product has held device clearance since 2006, operates in more than 400 hospitals and has passed the vendor review of a major diagnostics manufacturer to reach a partnership on its bedside hardware.

That a company with those credentials publishes nothing readable is the finding. A device manufacturer of twenty years standing is exactly the vendor best placed to publish its compliance posture and least excused for not doing so.

Retrieval located no service organisation controls report, no HITRUST certification, no ISO 27001, no medical device software lifecycle standard, no trust centre, no penetration testing cadence and no vulnerability disclosure policy. The absence of the device software standards is the more surprising half.

This is regulated Class II software computing a therapeutic dose, so a quality management system and a hazard analysis process necessarily exist to hold the clearance, and a direct competitor in this category publishes conformance to exactly those standards. Naming them would cost nothing and would tell a hospital more about this product's safety engineering than any security attestation could.

The strongest regulatory position in this category and the clearest illustration of why the category exists. Glucommander is cleared as Class II software as a medical device, for intravenous insulin dosing since 2006 and for subcutaneous dosing since 2010, in hospitalised adult and paediatric patients aged two and over, and the company states it has had no recalls in that period.

Twenty years of maintained clearance across two indications and a paediatric population is a substantive regulatory record rather than a badge. It also demonstrates the boundary this index draws: computing an insulin dose is treated by the agency as a device function requiring clearance, while advisory medication alerting elsewhere in this same category operates under the statutory exclusion for non device decision support. Same category, same patients, opposite regulatory status, and the difference is whether the software produces a quantity that goes into the patient.

Retrieval located no performance reporting by age, sex, race, ethnicity, body habitus, renal function or diabetes type, no bias assessment and no post deployment monitoring statement. Two exposures are specific to this product rather than generic.

Insulin requirements vary substantially with body composition, renal function and insulin resistance, and a control algorithm tuned on one case mix may reach target more slowly or overshoot in patients unlike it, with hypoglycaemia as the failure mode.

And the clearance extends to children from two years old, where dosing tolerances are far tighter and the published evidence base is far thinner than for adults, so paediatric performance is the specific stratification a buyer should request.

Integration is bidirectional and clinically deep rather than a data feed. The software integrates with major electronic health record systems including Epic and Oracle Health, receives its initiating parameters directly from the provider's order set, and writes the calculated dose into the medication administration record where the nurse charts against it, with the administered dose returning to the algorithm.

Reach extends beyond the record system through a partnership placing the application on Roche's cobas pulse hospital glucose meter, so the dosing recommendation can surface on the device that produced the reading. Held at B because no marketplace certification, FHIR conformance statement or public interface documentation was located, and because the integration described is specific to a handful of major systems rather than broadly standards based.

The platform is cloud based and integrates directly into hospital record systems, with a command centre component giving system wide visibility across sites, which implies a multi facility hosted architecture. Retrieval located no named hosting provider, no cloud region, no data residency commitment and no on premise option, and no description of how the bedside device integration handles data when network connectivity to the hospital's systems is interrupted.

That last question matters more here than for most software in this index, because an insulin infusion continues whether or not the dosing system is reachable, and what the workflow falls back to is a patient safety question rather than an availability one.

No price, unit, pricing basis, contract shape or implementation fee was located. The company publishes extensive return on investment material built on reduced length of stay, readmissions and complications, so the value side is quantified in detail while the cost side is entirely absent, which is the recurring asymmetry across this category.

One commercial context worth noting for a buyer: the company markets actively against a 2026 CMS glycemic quality mandate, so purchasing pressure in this segment is currently regulatory rather than discretionary, and a vendor selling into a compliance deadline has less incentive to compete on published price than one selling into a discretionary budget.

Coverage within glycemic management is the most complete in the segment. Both intravenous and subcutaneous dosing carry clearance, the indication spans hospitalised adults and children from age two, and the platform extends past the inpatient episode through an outpatient product that titrates basal bolus regimens using continuous glucose monitoring data, giving a continuum from critical care to the clinic. Install base is reported at more than 400 hospitals and health systems.

Held at B because the function is confined to one drug class in one disease, there is no evidence of use outside the United States, and the outpatient component has a far thinner published record than the inpatient one that carries the company's reputation.