PrecisePK

The model is the product. PrecisePK assigns a patient to a population pharmacokinetic model fitted to published patient data, selecting it automatically by patient type such as critically ill, obese or pediatric or letting the pharmacist choose, and then updates that model against the patient's measured serum concentrations to compute a regimen against an area under the curve, peak or trough target.

A statistical model fitted to population data and then updated on the patient's own measurements is a learned component, and it is the whole of what the software sells: remove it and no dosing calculation remains. The company also describes learning population models from a hospital's own patients over time, and a sibling brand, PreciseRx, is described as combining clinical research, machine learning and cloud computing.

A clinical pharmacist runs the calculation, reviews the output and makes the recommendation, so the software never touches an order. Two design choices strengthen the oversight position. The user can either have the software select the pharmacokinetic model that best fits the patient or select it manually, which keeps the single most consequential modeling decision in human hands rather than automating it silently.

The product also carries a toxicity alerting layer that warns when area under the curve falls outside the recommended range or when a patient approaches an ototoxic exposure, so the software flags risk in the direction of caution rather than only proposing a dose. No acceptance rate, override rate or confidence bound is published, and nothing describes what happens after a dose computed by the software proves wrong.

The technology page walks a reader through the actual computation in sequence: a population pharmacokinetic model drawn from published literature and research, patient specific covariates such as age, sex and weight forming a patient profile, assignment into a cluster with corresponding pharmacokinetic parameters, and then additional drug factors, with two compartment kinetics in the calculation. It is written for a pharmacist rather than a procurement committee.

PrecisePK attributes its accuracy in the Turner evaluation to the breadth of its model library and its two compartment models; that is the company's explanation, not the study's. The specific population models in the library are not named by author and cohort, no thresholds or calibration are published, no versioning or update practice is described, and the machine learning element referenced on the same page has no published method behind it.

The chain is short and published. The dosing models are PrecisePK's own implementation of population pharmacokinetic models from the published literature, so no outside model provider sits between the record and the dose. The subprocessor page, updated 18 March 2026, names 28 parties with each one's role and country, including Amazon Web Services for the data center in the United States, and offers change notices by subscription; the business associate agreement binds every subcontractor that handles PHI and requires return or destruction of PHI at termination.

Training terms are where the disclosure stops. PrecisePK describes learning population models from a hospital's own patients, which is the point at which customer data becomes model input, and nothing published says whether those locally derived models stay with the institution or whether anything learned from them enters the general model library other customers use.

The strongest published result is independent and small, and the company states it without its limits. Turner and colleagues at Pacific University School of Pharmacy, publishing in Pharmacotherapy in 2018, compared five Bayesian dose optimizing programs and two first order equations for estimating vancomycin area under the curve in critically ill patients, and found PrecisePK the least biased of the group at a median of 5.1 percent.

The same study found this product among the more difficult of the five to use, and found that adding a second serum level improved accuracy for two other programs but not for this one. It included nineteen patients. The company's own material describes the product as validated as the most accurate and least biased Bayesian guided software without qualifying the sample size, which is a superlative resting on a very small study.

Set against that, the clinical lineage is real and long: development inside a university applied pharmacokinetics service, beta testing at an academic medical center and a children's hospital, and more than thirty years of continuous use at UC San Diego Health across nine University of California sites.

PrecisePK publishes categorical commitments without the schedule behind them. Its privacy notice states that no patient identifying information is ever accessed or shared by the company's employees or personnel, and its business associate agreement limits use to the minimum necessary, disclaims any ownership of customer data in any form, and requires return or destruction of all PHI at termination, including PHI held by subcontractors.

What is not published is a retention window while a customer is active, a de identification practice, or whether saved patient cases, which the product lets clinicians share with colleagues for handoff, and customer data used to learn local population models ever reach the general model library. Safety engineering is partly published: the software warns when an area under the curve falls outside the recommended range or a patient approaches an ototoxic exposure. How a wrong dose traced to the software is reported and corrected is not described.

The business associate agreement is published in full at an open URL, precisepk.com/business-associate-agreement, last updated 18 March 2026, between Healthware, Inc. (doing business as PrecisePK) and any customer, so the same terms apply to every customer rather than to a tier.

It binds subcontractors: section 7 requires every agent or subcontractor that handles PHI to accept the same restrictions in writing, and commits PrecisePK to post notice of new subcontracts on its website within 30 days. It also sets a five business day window for reporting security incidents, a 30 day ceiling for breach notification with reimbursement of the customer's notification costs, return or destruction of all PHI at termination, and an audit report on request. Business associate status is stated in the privacy notice and in a HIPAA Privacy Statement, and the subcontractor list is published separately.

PrecisePK's compliance page, linked from every legal page on its site, names SOC 2 Type 2, ISO/IEC 27001:2022, HIPAA and GDPR, and states that copies of the certifications are available by emailing the company. Its business associate agreement adds an audit report on request, naming an independent HIPAA compliance report (AT-C 315) or another agreed third party audit.

That puts two recognized certifications, one with its version, in the vendor's own material, with the reports themselves behind a request rather than on an open portal. No audit period, certificate date or scope statement is published, so which systems and which period each report covers is not stated. No penetration testing cadence and no vulnerability disclosure program are published.

No clearance is claimed in any market. PrecisePK's terms of service scope the product as an aid for suitably qualified medical staff making dosing recommendations, not medical advice, with the institution or professional solely responsible for treatment decisions. That is the long standing position for therapeutic drug monitoring software used under a clinical pharmacist's professional judgment, and the pharmacist mediated workflow supports it.

The tension is that the product computes a dose, and computing a dose is the function regulators in the United States and Europe have treated as a device function for some dosing software, while PrecisePK publishes no reasoning for its own position. Thirty years of uneventful use is not a stated position.

Performance is published for a patient group where the product's own method is most likely to fail, and not compared across groups. A Bayesian dose starts from a prior taken from a published population and predicts worst where drug exposure is least predictable, and critically ill patients, with fluid shifts and changing kidney function, are among the least predictable.

They are the population in Turner and colleagues' evaluation (Pharmacotherapy, 2018), where PrecisePK's area under the curve estimates had a median bias of 5.1 percent across 19 patients; PrecisePK cites the study. The product serves far more than that group, across wards, pediatrics and intensive care.

PrecisePK addresses the prior problem in two ways, by selecting population models by patient type such as critically ill, obese or pediatric, and by learning population models from a hospital's own patients. No results compare accuracy across age, sex, race and ethnicity, renal function band or body habitus, no bias assessment is published, and there is no statement of how the models are monitored after deployment. Results compared across those groups would move this to A.

A published error figure stands behind the output, in a study PrecisePK itself cites. Turner and colleagues at Pacific University School of Pharmacy (Pharmacotherapy, 2018) tested five Bayesian dosing programs against a reference area under the curve in a published data set of 19 critically ill patients, and PrecisePK's estimates had a median bias of 5.1 percent, the lowest of the five; the metric is a median percentage error across the 19 patients, so there is no count numerator.

PrecisePK's own articles cite the study as validation. A pharmacist also reviews every calculation before any dose is ordered. What is missing is anything that makes the vendor answerable, and the terms of service run the other way: the software is provided as is without warranty, and a customer indemnifies Healthware against product liability claims arising from its use. The study is small, and no error figure is published for other drugs or other patient groups.

Integration is real and narrow. A third party review directory reports full Epic integration through that vendor's app marketplace with automatic import of medication orders, prior doses, drug levels and serum creatinine, which are precisely the four fields the calculation needs, plus single sign on support and an integration with the Sanford Guide that puts infectious disease guidance beside the dosing decision.

PrecisePK's own material says that when integrated with an EHR the software retrieves patient information from it, and treats integration as a purchasing consideration alongside clinical surveillance tools and reporting compliance. Only one electronic health record is named, no FHIR conformance statement or public API documentation is published, and the specifics come from a secondary source rather than from the vendor's own documentation.

PrecisePK is a web platform, having moved from desktop software, and its subprocessor page states where the data sits: Amazon Web Services provides the data center, in the United States, and the page lists each subprocessor's country of processing, with one ticketing tool processing support tickets in Australia. That states the hosting option and the residency.

Tenant isolation is not described, no on premise or institution hosted option is stated for customers who ran the desktop era product, and the processing path between the hospital record and the platform is not documented. Nothing says whether the published evidence and the current platform are the same software.

No price is published, and the route to start is. The site offers a sixty day free trial of the software prominently and repeatedly, a concrete commercial term that lets a pharmacy department test accuracy against its own patients before committing. Unlimited one to one onboarding training and ongoing technical and clinical support are described as included rather than quoted as separate professional services. What remains undisclosed is the price itself, the unit of charge, the contract term, whether the trial converts automatically, and whether EHR integration carries a separate implementation charge.

Coverage is deep within a narrow function. The product supports therapeutic drug monitoring across the drug classes where measured concentrations guide dosing, with vancomycin dominating current practice and the aminoglycosides and other narrow therapeutic index agents alongside it, and the company states use in all types of patients. Pediatric capability is credible rather than asserted, given beta testing at a children's hospital in the product's development history.

Institutional coverage is concentrated in American academic medicine, with University of California sites named and a thirty year relationship at one of them. The footprint outside academic centers is not described, there is no evidence of use outside the United States, and the function is confined to drugs with routine concentration monitoring rather than to prescribing generally.