AlertWatch:OB

The mechanism is a maternal early warning score grounded in national obstetric standards, a proprietary filtering layer over the resulting alerts, and real time integration and display of several hundred data elements. Nothing retrieved describes machine learning. The grade describes mechanism, not quality. HONEST NON INFLATION CREDIT IS DUE AND IS RARE IN THIS LANE: the company describes itself as clinical decision support software rather than reaching for the AI label, in a category where nearly every competitor does the opposite. Same treatment this index gives Elaborate and Candid Health. What the product actually contributes is information architecture and alert discipline, which is a legitimate and, on this record, deliberately chosen engineering position.

No autonomous action. The system scores, alerts and pages, and clinicians decide. There is no low risk or de escalation output of the kind that creates the invisible failure mode on CLEW, Etiometry and Healthplus.ai. THE SHARPEST QUESTION ON THIS RECORD IS THE FILTER, AND IT IS A NEW FORM OF AN OLD PATTERN. Proprietary filtering to prevent alarm fatigue necessarily means the system is deciding which alerts a clinician does not see. That is a suppression decision, and it belongs in the same family as Affineon auto handling normal labs and CLEW labelling patients low risk: something is being removed from human attention, and the removal is invisible to the person it is removed from. The difference is that here the suppression is the headline feature rather than a side effect. ASK DIRECTLY: what does the filter suppress, on what criteria, and what proportion of suppressed alerts preceded a deterioration. Held at B rather than C because the alerting logic itself is grounded in published national standards, the filter reduces alert volume rather than removing a patient from monitoring, and clinicians retain the underlying data display.

The clinical grounding is traceable and that is credited: the score is stated to be based on national obstetric standards, and the postpartum hemorrhage assessment is explicitly tied to ACOG risk criteria, so a clinician can locate the underlying clinical logic in published guidance. THE PROBLEM IS THAT THE TWO ELEMENTS THAT DIFFERENTIATE THE PRODUCT ARE BOTH DESCRIBED AS PROPRIETARY AND NEITHER IS SPECIFIED. The maternal early warning score is a proprietary modification of the standards, and the alert filtering is proprietary. The filtering matters most, because reducing alarm fatigue without missing deterioration IS the product claim, and nothing published describes what the filter suppresses, on what basis, or how often. No operating characteristics, sensitivity, specificity or positive predictive value are published for either the score or the filter.

Usage volume is documented, clinical outcome evidence is not. More than 10,000 births assisted at commercial launch in January 2020, rising to more than 15,000 in later reporting, with early sites at the University of Michigan, University of Tennessee and University of Vermont. The company's research page lists a study evaluating AlertWatch:OB alongside Maternal Early Warning Criteria for detecting severely morbid postpartum hemorrhage, and a 2019 SOAP conference abstract reporting a provider survey in which most respondents felt the system improved safety, communication and response time, but findings from the former were not retrievable and the latter measures perception rather than outcome. The comparative data from AlertWatch:OR covering more than 17,000 surgeries was a 2014 conference presentation on a different product. INDEPENDENT CONTEXT BELONGS ON THIS RECORD, because it applies to the whole approach rather than to this vendor alone: a 2025 systematic review appraising modified obstetric early warning scores against PROBAST and TRIPOD found them methodologically weak and at high risk of bias, with several developed by clinical consensus rather than statistically, and an earlier systematic review found obstetric early warning systems achieve high sensitivity and specificity but a low median positive predictive value of around 41 percent. Separately, a University of Chicago cohort of 19,611 obstetric patients found a deterioration rate of 0.2 percent, which is why positive predictive value is low across every score in this space.

No privacy policy, data processing statement, retention period or training use disclosure was located. Not Rated means no public evidence was found rather than evidence of weakness. One question is now specific to the ownership change: following the 2022 acquisition by BioIntelliSense, whose platform combines inpatient vitals, EMR and laboratory data with its own medical grade wearables, a buyer should establish whether obstetric data flows into that wider platform and on what basis.

No public statement on business associate agreements, execution terms or subprocessor disclosure was located. Not Rated means no public evidence was found. Named academic health system deployments mean agreements exist, but nothing about them is published, and the 2022 change of ownership makes the identity of the contracting entity a question worth asking directly.

No SOC 2 of any type, HITRUST, ISO 27001, trust centre or report request path was located. Not Rated means no public evidence was found. As recorded elsewhere in this category, FDA 510(k) clearance is a safety and effectiveness determination and is not an information security attestation.

Genuine and repeated clearance history across three products: AlertWatch:OR cleared in 2014 for the operating room, AlertWatch:AC for acute care, and AlertWatch:OB, which the company describes as the first and only FDA 510(k) cleared software system dedicated solely to maternal safety in labor and delivery. That is a specific and checkable competitive claim rather than a vague one, and clearing a dedicated maternal safety indication is not trivial. Held at B rather than A because no clearance numbers, indications for use text or performance basis were retrieved for the OB product, so the scope of what was actually cleared could not be read directly, and because no recent regulatory activity was identified. Per this index's standing method, the 510(k) summary should be pulled and read on the refresh pass, since the device description and indications routinely differ from marketing in ways that matter.

No subgroup performance, calibration or fairness analysis is published, in the clinical domain where that gap costs most. The company's own framing invokes the disparity directly, citing a US maternal death rate roughly three times that of other industrialised countries and 70 near misses for every death, and arguing that many could have been prevented by a timelier response. Timeliness of response is exactly where the documented racial disparity in US maternal mortality operates, so a product aimed at that mechanism has both the clearest opportunity to reduce it and the clearest obligation to measure whether it does. Nothing published addresses it. The independent systematic review evidence that modified obstetric early warning scores are methodologically weak and frequently consensus derived sharpens the point, because a score built on expert consensus inherits whatever the consensus encodes. Same ask as PeriGen and Birth Model: publish subgroup performance, and better, subgroup outcomes.

Ingests broadly rather than from a single source, integrating hundreds of data elements across physiological monitors, the EMR, laboratory systems and medical history, which is what allows the postpartum hemorrhage assessment to check blood availability and intravenous access alongside vital signs. Distribution and integration have run partly through Clinical Computer Systems, maker of the OBIX electronic fetal monitoring platform, as both reseller and co marketing partner, which is a practical route into labor and delivery units already running that system. Held at B because no EHR vendor is named, no marketplace listing or partner certification was located, and no FHIR or SMART on FHIR capability is described.

THE ISSUE HERE IS CURRENCY RATHER THAN ABSENCE, and it is stated plainly because a buyer needs to know. Named early sites are the University of Michigan, University of Tennessee and University of Vermont, and usage was reported at more than 15,000 births, but that reporting dates from 2020 and no current deployment, customer count or installed base was identified. AlertWatch was acquired by BioIntelliSense in 2022, whose stated direction combines the AlertWatch display and analysis layer with its own medical grade wearables across care settings, and it was not possible to verify from public sources whether AlertWatch:OB remains actively sold and developed as a standalone obstetric product. No implementation timeline, hosting architecture or data residency commitment is published. Establish current product status and roadmap directly before evaluating.

No pricing published at any level: no rate card, no unit of pricing such as per delivery or per labor and delivery bed, no band, and no implementation fee. Distribution through a reseller adds a question a buyer should raise early, since a product bought through the OBIX channel may carry different commercial terms from one bought direct, and the 2022 change of ownership makes it worth confirming who the contracting party now is.

Obstetric only, but the span within obstetrics is the widest in this category and it is the right axis to compare on. Coverage runs from triage through labor and delivery into operating rooms and post anesthesia care units and onward through the postpartum period, which matters because a substantial share of severe maternal morbidity and mortality occurs after delivery rather than during it, and postpartum is precisely where continuous attention lapses as the focus shifts to the newborn. THE CLEAN CONTRAST WITH PERIGEN IS ON WHAT IS MONITORED, NOT ON QUALITY: AlertWatch monitors the MOTHER across the whole perioperative and postpartum journey and does not address the fetal signal at all, while PeriGen monitors both mother and fetus but is centred on the intrapartum period. Held at B because the specialty is singular and no gestational age or population limits were retrieved.