Medication Safety & Prescribing
AI applied to the medication itself: catching a prescribing error before it reaches the patient, computing a dose from that patient's own pharmacokinetics, screening a regimen for interactions and cumulative risk, scoring a patient for controlled substance risk, and surveilling an inpatient population for adverse drug events and antimicrobial misuse. The category separates from clinical decision support on what the model decides. Here the decision is about a drug, which drug, what dose, and whether this patient should receive it, and the output is either administered or used to withhold something. That distinction has a regulatory expression as well as an editorial one. Several products in this category are cleared by the FDA as Class II software precisely because they compute a dose, while others assert the Cures Act exclusion for non device clinical decision support on the ground that a clinician can independently review the basis of their output, a claim that is currently the subject of a citizen petition asking the agency to decide the question. Establishing which side of that line a product sits on, and on whose reading, is the first thing to ask a vendor here and it is rarely on the datasheet. Three failure modes define the category and they are not the same failure. The first is the wrong quantity. A dose produced by a pharmacokinetic model fitted to two blood levels cannot be reproduced in a clinician's head, so the ordinary safeguard of independent review is unavailable exactly where the consequence is kidney injury or hypoglycaemia. The second is the ignored alert. This is the segment that produced the override problem in the first place, and legacy interaction checking is routinely overridden at rates that would be treated as a defect anywhere else, so the useful metric is not sensitivity but alert burden per order and the proportion of alerts that change a prescription. The third is the withheld medication, and it exists nowhere else in this index. A risk score attached to a patient's controlled substance history can result in that patient being refused treatment, so a false positive here is not wasted time but denied care, and the published false positive rate for the most widely deployed such score is high enough that roughly a third of scored patients may be misclassified. Buyers should ask four things. What is the alert burden per order and what proportion of alerts result in a changed prescription, since a vendor who cannot answer is selling into the problem it claims to solve. Whether the model is the vendor's own or licensed, because several incumbents in this category embed another vendor's dosing engine and sell it as a feature. Whether the algorithm and its validation are published in a form an outside party can check, since the most consequential products here are also the most opaque. And whether performance has been reported by subgroup, which matters more than usual in a category where the scored behaviours, pharmacy counts, prescriber counts and distance travelled, track access to care at least as well as they track risk.
Vendors in this category
16 indexed
| Vendor | Category | AI Centrality | Website |
|---|---|---|---|
|
F
FDB (First Databank)
FDB, formerly First Databank, supplies the drug knowledge that sits underneath most medication decisions made in the United States. Its databases are embedded in the majority of hospitals, physician practices and pharmacies, which means that when a prescribing system warns about an interaction, a dose ceiling or a contraindication, the content behind that warning is frequently this company's rather than the software vendor's. It is owned by Hearst and part of Hearst Health, the same group that owns Zynx Health. The asset is human curated: clinical editors maintain the content and the company's stated credibility rests on rigorous quality control and clinical expertise rather than on any model. What makes the record current rather than historical is a deliberate repositioning. In October 2025 FDB opened pilot integrations of a Model Context Protocol server, and on 31 March 2026 it made FDB MedProof MCP generally available, described as the first such server purpose built for agent driven medication decision support, letting other companies' AI agents and language models query clinically validated drug knowledge in patient specific context rather than answering from training data. Two further products were previewed in March 2026 and are not yet generally available: FDB Script Agent, which turns a spoken clinical conversation into a structured prescription for physician review, and FDB VerifyAssist, an inpatient pharmacy order verification assistant. The company is positioning itself as the grounding layer beneath other people's medication AI.
|
Medication Safety & Prescribing | D | fdbhealth.com |
|
G
Genomind
Genomind pairs a pharmacogenetic laboratory test with software that turns the result into prescribing guidance, with a centre of gravity in mental health. The assay is a clinician ordered buccal swab genotyped by quantitative PCR across 26 to 27 genes, split between 11 that affect how a drug is metabolised and 15 that affect how the body responds to it, returning results in three to five days and covering more than 700 medications. The Precision Health Platform then integrates that genotype with the patient's current medication list to flag gene to drug and drug to drug interactions and propose alternatives. The company was founded in King of Prussia, Pennsylvania by Dr Ronald Dozoretz, sells to individuals, prescribers, long term care facilities, accountable care organisations, employers and health plans, holds a contract supporting testing for United States active duty service members, and licenses its reporting suite to other laboratories so that its interpretation layer reaches patients whose samples it never handled. What distinguishes this record from everything else in the category is what the company did with its own evidence. It sponsored and co authored a multicentre randomised controlled trial of its own test in major depressive disorder, blinded to both participants and raters, and the trial found no significant improvement in the primary efficacy outcome. The company continues to cite that trial on its own science page.
|
Medication Safety & Prescribing | D | genomind.com |
|
T
TheraDoc
TheraDoc is one of the two incumbent clinical surveillance platforms in American hospitals, monitoring live feeds of laboratory, microbiology, pharmacy and admission data to raise alerts for infection prevention, antimicrobial stewardship, pharmacy surveillance and anticoagulation management, and to produce mandatory infection reporting. It was founded in Salt Lake City in 1999, sold to Hospira, and acquired by Premier in 2014 for 117 million dollars in cash, and is now marketed as Premier's Clinical Surveillance powered by TheraDoc. Its mechanism is a rules engine rather than a model: pharmacists and infection preventionists author their own alert logic, and the system evaluates those written queries continuously against incoming data. Premier makes no artificial intelligence claim for the product on its own pages, which is more accurate than what its principal competitor publishes. The one genuine model inside the product is licensed rather than built, since Premier's own material states that the integrated precision vancomycin dosing tool in TheraDoc comes from InsightRX. Two ownership facts matter for procurement. Premier completed a 2.6 billion dollar take private by Patient Square Capital, having reported an eleven percent revenue decline in fiscal 2025 and having explored strategic alternatives since 2023. And Premier separately owns Stanson Health, so a single supplier now sits behind both this surveillance platform and a point of care decision support product.
|
Medication Safety & Prescribing | D | solutions.premierinc.com |
|
V
VigiLanz
VigiLanz is the other incumbent clinical surveillance platform in American hospitals and the direct competitor to TheraDoc. Founded in Minneapolis in 2001 by David Goldsteen and acquired by Inovalon in February 2024, it ingests near real time data from any electronic health record, including admission feeds, medication administration records, laboratory results, microbiology and medication history, across acute, post acute and ambulatory settings, and converts it into alerts for pharmacy surveillance, infection prevention, antimicrobial stewardship, quality improvement and patient safety event reporting. The mechanism is configurable rules, and the company's own product page describes it in a phrase worth quoting because it captures the entire problem this index exists to address: flexible, rule based AI. Independent product listings describe the same engine plainly as configurable and rule based. Alerts surface directly inside the electronic health record for faster acknowledgement, and the platform supports intravenous to oral conversion review, antibiogram trending and transitions of care. As with its competitor, the genuine model reaching clinicians through the product is licensed from elsewhere: Inovalon announced a collaboration in October 2025 bringing DoseMeRx precision dosing into these pharmacy surveillance workflows. The company cites eight consecutive years as the leading pharmacy surveillance solution in one analyst firm's rankings.
|
Medication Safety & Prescribing | D | inovalon.com |
|
E
EndoTool
EndoTool is an FDA cleared inpatient insulin dosing system and the closest direct competitor to Glucommander, covering intravenous dosing, the transition from intravenous to subcutaneous insulin, and subcutaneous dosing for adult and paediatric patients. Its algorithm is described in independent literature as model predictive control using a nonlinear dosing equation that is individualised and then optimised over time against how that patient's glucose actually responded to the doses already given, drawing on diabetes type, height, weight, creatinine, age and sex. That is a more elaborate approach than the proportional control its main competitor uses, and it is still deterministic mathematics rather than machine learning: the same peer reviewed reviews that describe it group it with rule based systems carrying static decision rules rather than adaptive learning models. The product was built by Monarch Medical Technologies of Charlotte, North Carolina, which also appears in the clinical literature under the name MD Scientific LLC, and Glooko acquired Monarch in September 2025 to extend its outpatient diabetes platform into the hospital, describing the combination as a hospital to home continuum. Glooko has stated that both product lines continue to operate and be supported. The vendor characterises the algorithm as patented and patient specific, and markets against the same 2026 CMS glycemic quality measures its competitor does.
|
Medication Safety & Prescribing | C | monarchmedtech.com |
|
G
Glytec
Glytec's Glucommander is the oldest and most widely deployed computerised insulin dosing system in American hospitals, cleared by the Food and Drug Administration as Class II software for intravenous insulin since 2006 and for subcutaneous insulin since 2010, in patients aged two and over. It sits inside the GlytecOne platform alongside surveillance, alerting and analytics, and the company reports use across more than 400 hospitals and health systems and dosing for over 867,000 patients since 2015. The mechanism deserves stating plainly because the marketing does not. Independent reviews describe Glucommander as a proportional integral derivative controller applying a linear formula derived from a 1982 paper, where the hourly insulin dose is a function of blood glucose above a floor multiplied by a factor that steps up until glucose reaches target. It adapts to the individual patient through that feedback loop, which is real closed loop control and is not machine learning, and the same reviews explicitly classify these systems as rule based with static decision rules rather than adaptive learning models. The company nonetheless describes the algorithm as learning each patient's insulin sensitivity. Distribution now extends to the bedside device itself through a partnership making Glucommander the first software application to run on Roche's cobas pulse hospital glucose meter, and the company markets heavily against a 2026 CMS glycemic quality mandate.
|
Medication Safety & Prescribing | D | glytec.com |
|
S
Synapse Medicine
Synapse Medicine builds what it calls a medication intelligence platform, and the distinctive thing about it is where the machine learning sits. Rather than learning to make a prescribing recommendation, the company uses natural language processing to read and classify medication information from three kinds of unstructured source, manufacturer product documentation, official recommendations from health authorities, and published research, and to assemble that into a structured knowledge base that then drives interaction checking, prescription support and medication reconciliation. A separate product applies the same approach to pharmacovigilance itself, automating the coding and prioritisation of adverse drug reaction reports for the centres that receive them. The company was founded in Bordeaux in 2017 by Clement Goehrs and Louis Letinier, both physicians, with Bruno Thiao-Layel, and has raised roughly 39 million dollars across three rounds through a Series B in March 2022 led by Korelya Capital, with backing from a healthcare professionals' insurer and public investment banks rather than from pharmaceutical money. Its institutional position in France is unusual: users include the national medicines agency and the digital health agency alongside Assistance Publique Hopitaux de Paris, the Hospices Civils de Lyon and the Bordeaux teaching hospital, plus more than a hundred hospitals and dozens of telemedicine companies. The company states that it is completely independent of the pharmaceutical industry.
|
Medication Safety & Prescribing | B | synapse-medicine.com |
|
Y
YouScript
YouScript computes the cumulative effect of a patient's whole medication regimen together with their pharmacogenomic results, rather than checking drug pairs one at a time. Its founder, Kristine Ashcraft, makes the argument directly: American practice has relied on binary interaction alerting for decades, and binary alerting breaks down once a patient is on five or more medications, which now describes tens of millions of people. The patented engine models drug to drug interactions, drug to gene interactions and phenoconversion, where one drug alters how a patient metabolises another and effectively changes their genotype in practice, then returns severity coded guidance across four levels from change to no change alongside safer alternatives within the same drug class or indication. It draws on more than 19,000 high evidence references covering over 2,000 medications and integrates with Epic, Cerner, Allscripts and Telus Health as well as through standalone applications and APIs. The product was incubated for over a decade inside Genelex, a pharmacogenetic testing laboratory in Seattle, established as YouScript in 2016, acquired by Invitae in 2020, and sold to Aranscia in an all cash transaction announced in November 2023. It holds something rare in this category: a prospective randomised controlled trial with a rehospitalisation endpoint. It also markets itself as laboratory agnostic while having been owned by a testing business for essentially its entire existence.
|
Medication Safety & Prescribing | C | youscript.com |
|
A
Arine
Arine sells medication intelligence to health plans rather than to hospitals, which makes it the clearest payer side record in this category. The platform ingests clinical, socioeconomic and behavioural data across a plan's membership, identifies members at risk of medication related problems before they produce an event, generates personalised care plans for clinical pharmacists to act on, and separately analyses prescriber behaviour so a plan can direct education and recommendations at the clinicians writing the prescriptions. A third product line targets the Medicare Advantage and Part D quality measures, particularly the triple weighted adherence measures that carry disproportionate weight in the CMS Star Ratings that determine a plan's bonus payments. The company was founded in San Francisco by Yoona Kim, a doctor of pharmacy with a doctorate who serves as chief executive, alongside Penjit Moorhead as chief technology officer and David de Vries as chief operating officer, and it launched under the Arine name in 2019 after operating as Akeso. It has raised roughly 95 million dollars across six rounds through a Series C in June 2025, and holds a HITRUST risk based two year certification obtained in July 2024. Named clients span Medicaid, Medicare Advantage and behavioural health, including the Oklahoma Health Care Authority, Magellan Health, VNS Medicare and MDX Hawaii. Every outcome figure the company publishes is its own, though it names the analytic method behind them, which is more than most.
|
Medication Safety & Prescribing | A | arine.io |
|
D
DoseMeRx
DoseMeRx is a Bayesian precision dosing platform for high risk parenteral medications, computing an individualised regimen from clinically validated pharmacokinetic models, patient characteristics, measured drug concentrations and, where available, genotype. It covers 42 drugs and reaches more than a thousand clinicians across more than a hundred locations spanning the United States, Europe, Asia Pacific, Africa and South America, which is the widest geographic footprint in this category. The product originated in Brisbane, Australia, and its corporate history is unusually eventful for software a hospital pharmacy depends on daily: Tabula Rasa HealthCare acquired it in a deal reported at up to 30 million dollars, completing in January 2019, and sold it four years later to an affiliate of Fairlong Capital on 20 January 2023, with customer contracts transferring to DoseMe Operations Inc. and Paul Edwards installed as chief executive. It now operates from Houston. Three owners in six years is a supplier continuity question a procurement committee should ask about directly rather than a criticism. What sets the record apart is disclosure. This is the only vendor in the category so far to publish where United States patient data physically resides, to state plainly that the customer owns the data it enters, and to describe in public the licence it takes back over de identified data for model improvement. It also holds HITRUST certification and claims to be the only Bayesian dosing platform that does.
|
Medication Safety & Prescribing | C | doseme-rx.com |
|
P
PrecisePK
PrecisePK is Bayesian therapeutic drug monitoring software from Healthware, Inc. in San Diego, and it has the longest continuous clinical lineage of any product in this category. It began as T.D.M.S. 2000, developed by Philip Anderson, a doctor of pharmacy, with pharmacokinetic consultants, and programmed by Anjum Gupta, a computer scientist, both professors at the University of California, San Diego, and beta tested by the Applied Pharmacokinetics Service at UCSD Medical Center and Rady Children's Hospital. It became a desktop product and later a web platform, and the company marks more than thirty years of partnership with UC San Diego Health across nine University of California sites. The software takes a patient's characteristics, laboratory results and dosing history, assigns them to a population pharmacokinetic model, and updates that model against measured serum concentrations to compute a regimen against an area under the curve, peak or trough target, with the Bayesian method able to work from a single randomly drawn level rather than requiring steady state trough sampling. It carries a genuine distinction: an independent group comparing five Bayesian dose optimising programs in critically ill patients found PrecisePK the least biased of the five. The company markets that result heavily, and a buyer should know the study included nineteen patients and also found this product among the harder ones to use. `founded` is left blank because the product lineage predates the current company branding and no founding year is published.
|
Medication Safety & Prescribing | C | precisepk.com |
|
F
FeelBetter
FeelBetter applies machine learning to polypharmacy, the problem of an older patient accumulating enough medications that the regimen itself becomes the hazard. The platform synthesises longitudinal pharmacy and clinical data across a whole population, identifies the patients at highest risk of near term deterioration or hospitalisation attributable to their medication regimen, recommends specific regimen changes to a clinical pharmacist, and monitors what happens after the intervention. It is organised as four tools: Insight for risk stratification, Navigator for workflow, Action for the pharmacist facing recommendation engine, and Impact for post intervention monitoring. The company was founded in 2018 by Liat Primor, previously VP of global portfolio and chief of staff at Teva Pharmaceuticals, and Yoram Hordan, and operates from Boston and Tel Aviv. It raised 5.9 million dollars in July 2023 bringing total funding to roughly 8 million, which is modest relative to the health systems it serves. Named customers span three distinct buyers: Atlantic Health System in New Jersey across its accountable care organisations, Americare in long term care, and Leumit Health Services, an Israeli health maintenance organisation. The evidence position is unusual for a company this size, since investigators at Brigham and Women's Hospital published a retrospective cohort of 108,817 patients in the American Journal of Managed Care evaluating the platform's recommendations. The company markets its approach under a coined term, Pharmaco-Clinical Intelligence, which describes a market position rather than a method, and almost no technical, security or compliance detail is published alongside it.
|
Medication Safety & Prescribing | A | feelbetter.healthcare |
|
I
InsightRX
InsightRX builds model informed precision dosing software, which computes an individualised dose for a drug whose therapeutic window is narrow enough that getting it wrong causes harm. The core product, Nova, combines published population pharmacokinetic models with maximum a posteriori Bayesian forecasting: it starts from what is known about how a drug behaves in a population, updates that against the concentrations actually measured in this patient, and returns a regimen. A companion product, Apollo, aggregates the resulting data so an institution can examine dosing effectiveness across its own population. The company was founded in 2015 in San Francisco and its platforms are used by more than a hundred institutions and life science organisations across the United States and Europe, covering more than a hundred drugs and therapeutic areas including infectious disease, oncology and transplant. Vancomycin dominates the published record. What distinguishes InsightRX in this category is inspectability. The pharmacokinetic models it runs are named published models rather than proprietary ones, the company publishes which model it selects as default for which patients and the analysis behind that choice, and it has published its machine learning approach to model selection in a peer reviewed pharmacometrics journal. It is also one of the vendors whose software reaches clinicians through a competitor's product, since the precision vancomycin dosing tool inside Premier's TheraDoc clinical surveillance platform is InsightRX. The regulatory position deserves care from any buyer: the software carries a European CE mark obtained by self certification as a Class I device under the superseded Medical Devices Directive, and it holds no United States clearance of any kind.
|
Medication Safety & Prescribing | B | insight-rx.com |
|
N
NarxCare
NarxCare is a set of scores and visualisations layered on top of state Prescription Drug Monitoring Program data, built by Bamboo Health, formerly Appriss Health, which also supplies the underlying platform for more than 40 state and territory monitoring programmes. It is almost certainly the most widely deployed algorithmic risk score in American medicine, reaching hundreds of healthcare facilities and, by the company's account, five of the six largest pharmacy retailers. The report presents three Narx Scores covering narcotics, sedatives and stimulants, a set of configurable state indicators, and an Overdose Risk Score, a three digit value the company states is produced by a logistic regression model evaluating ten inputs drawn from dispensation history such as prescriber counts, pharmacy counts and morphine milligram equivalents over varying windows. Bamboo Health is explicit that the score does not predict whether a patient will experience an overdose, that it is a correlative summary against the histories of people who died of unintentional overdose, and that it must never be the sole justification for providing or refusing medication. The record here is unusual in the index because the company's security and compliance engineering is among the strongest of any vendor graded, while its algorithmic accountability is among the weakest, and the gap between those two is a choice about what to publish rather than a failure of capability. Peer reviewed informatics literature records that the score has been validated only on a subset of three states, has never been evaluated clinically, and has no published subgroup analysis, and in 2025 clinicians petitioned the Food and Drug Administration to determine whether it is a regulated device at all.
|
Medication Safety & Prescribing | B | bamboohealth.com |
|
M
MedAware
MedAware is an Israeli company whose platform monitors prescribing for medication related risk using machine learning rather than the rule libraries that conventional medication decision support runs on. It was founded in 2012 after a nine year old asthma patient died because his physician selected the wrong drug from an electronic prescribing dropdown, an error no interaction rule was written to catch because the prescription was internally consistent and simply belonged to the wrong patient. The engine applies outlier detection over longitudinal and real time patient data, an approach the company compares to fraud detection in financial services, and flags prescriptions that are anomalous for this particular patient rather than prescriptions that violate a predefined rule. Products span prescribing error detection, evolving adverse drug event monitoring, and an algorithm that identifies patients at risk of opioid use disorder. The platform runs continuously rather than only at the moment of ordering, and delivers into existing systems through vendor agnostic APIs, an athenahealth Marketplace listing and, at Ballad Health from 2024, natively inside an Epic pharmacy workflow. MedAware is unusual in this index for the independence of its evidence base, which was produced largely by investigators at Brigham and Women's Hospital and Harvard Medical School and published in JAMIA and the Joint Commission Journal on Quality and Patient Safety, and equally unusual for how little it publishes about its own compliance, security and governance posture.
|
Medication Safety & Prescribing | A | medaware.com |
|
M
Micromedex
Drug and toxicology reference from Merative, the company formed in June 2022 when Francisco Partners acquired the Watson Health assets from IBM. Headquartered in Ann Arbor, Michigan; Merative chief executive Gerry McCarthy, with Sonika Mathur as general manager for Micromedex. Content is curated by clinical experts from primary literature with in line referencing and daily updates, and is used in more than 80 countries by providers, poison control centres, government departments, universities, pharmaceutical organisations and health plans. The artificial intelligence layer is search. Released 23 September 2025, it lets a clinician ask natural language questions about drug information including interactions and intravenous administration, and returns answers drawn from Micromedex content with a citation on every result that opens the underlying source in one click. Interface affordances include suggested searches and follow up prompts. The company describes the capability as clinically validated. Scoping note, because the boundary matters for this index. DynaMedex is a separate joint product combining this drug content with EBSCO's DynaMed disease content, and it is graded on the DynaMed record rather than here; a December 2025 agreement additionally routes Micromedex dosing and medication safety content into that product's generative layer. This record covers Micromedex itself. Two facts sit outside what a reference product normally carries and both are on the record. Micromedex is recognised as a drug compendium under United States federal law, which makes its content a determinant of coverage for certain off label uses rather than merely a guide to them. And in a separate arrangement the company has licensed a subset of its drug content into a consumer artificial intelligence answer engine as a premium source. Named number one for point of care drug reference in the 2026 Best in KLAS report, for the second time.
|
Clinical Reference & Evidence | C | merative.com |