01 Data
Automatic retrieval
The assay result and the patient context (identity, dosage, dosing times, weight, age, creatinine) are retrieved from the LIS or EHR. No manual re-entry, fewer identification errors.
Therapeutic drug monitoring · MIPD
The right dose, for the right patient, at the right time.
Plug'n'Dose distributes, integrates and supports Tucuxi, decision support software developed by CHUV and HEIG-VD. Using population pharmacokinetic models and Bayesian inference, Tucuxi helps personalise dose adjustment.
- Developed since 2012 by CHUV and HEIG-VD
- Open source
- Already in use in several university hospitals
- CE marking (class IIa) in progress
- Predicted median
- 25th to 75th percentiles
- 5th to 95th percentiles
- Trough target
- Dose
- Measurement
- Without adjustment
- Measured concentration
- 17.0mg/L Sampled at 35.5 h, before the 4th dose
- Estimated clearance
- 3.2L/h 90% interval: 2.7 to 3.8 L/h
- Steady-state trough
- 18.1mg/L Above target Current dosage, target 8 to 15 mg/L
- Suggested dosage
- 750mg every 12 h Predicted trough: 13.6 mg/L, from the 5th dose
Illustration with a fictitious drug and a simplified model. Try the real Tucuxi core
Plug'n'Dose
Making therapeutic drug monitoring simpler and more accessible
Plug'n'Dose is a French company founded in 2025 to make Tucuxi accessible to healthcare institutions and laboratories. We do not rewrite the software, whose core remains open. Our job is to make it installable, connected to your systems and usable day to day by your teams.
-
Distribution
Provision of Tucuxi, installation in your environment and long-term maintenance.
-
HIS and LIS integration
Automatic retrieval of patient data (identity, dosing, covariates, drug assays) from the EHR or LIS, with no manual re-entry and stronger patient identification safeguards.
-
Support and training
Creation of drug models, or help in creating them, and training for clinicians, whether they already practise TDM or are new to it.
-
Regulatory stewardship
CE marking process for Tucuxi as class IIa medical device software, under Regulation (EU) 2017/745.
Three audiences, one need
Personalising dose adjustment for narrow therapeutic index drugs, without adding manual re-entry or complexity to the teams' work.
-
Hospital pharmacy and clinical wards
Healthcare institutions
Make dosing safer and more personalised with a tool already in use in university hospitals, connected to your HIS. Patients eligible for TDM are centralised and identified by drug and by subpopulation.
-
Laboratory medicine
Laboratories
Add value to drug assays for prescribers with a reasoned interpretation and a dosing recommendation, based on data retrieved automatically from the LIS.
-
Analysers and automated systems
Diagnostics manufacturers
Pair dosing decision support with the results produced by your analysers, as a differentiating software component for your installed base.
A clinical, technical and operational alliance
The team brings together the initiator of Tucuxi, its lead developer and a pharmacokinetics engineer dedicated to its deployment.
-
Julien Massari
Founder and President
Pharmacokinetics engineer. Strategy, integration and client relations.
-
Prof. Thierry Buclin
Scientific and clinical lead
Physician and clinical pharmacologist, professor emeritus at CHUV. Initiator of the Tucuxi project, specialist in Bayesian TDM.
-
Prof. Yann Thoma
Technical lead
Professor of computer science at HEIG-VD. Lead developer of Tucuxi, interoperability and HIS integration.
Regulatory
A regulatory pathway under way
CE marking of Tucuxi is in progress, as class IIa medical device software (Regulation (EU) 2017/745, Rule 11). The process follows the sector's standards.
- ISO 13485
- Quality management system
- IEC 62304
- Software life cycle
- ISO 14971
- Risk management
- IEC 62366-1
- Usability
- IEC 81001-5-1
- Cybersecurity
- MDR, Art. 61
- Clinical evaluation
Tucuxi
Each patient has their own pharmacokinetics. And their own dosage.
Tucuxi is open-source decision support software for therapeutic drug monitoring, developed since 2012 by the Clinical Pharmacology Service of CHUV and HEIG-VD.
It combines population pharmacokinetic models from the literature with a Bayesian inference computation engine.
Tucuxi places each measured concentration against the percentiles expected for the patient, assesses exposure against the therapeutic target, suggests a dosage adjustment and predicts its effect. It produces an interpretation report ready to send to the prescriber.
Tucuxi is already in use in several leading university hospitals, including CHUV, AP-HP, AP-HM and HCL. Use of the open-source version, at the initiative of the institutions.
Technical specifications
- Licence
- Open source, AGPL v3
- Origin
- Clinical pharmacology at CHUV and HEIG-VD, since 2012
- Computation
- Bayesian inference: population, a priori and a posteriori predictions, percentiles
- Targets
- Trough, peak, AUC, cumulative AUC, AUC/MIC, time above MIC
- Routes
- Intravenous bolus, infusion, extravascular route
- Connectivity
- Reading from and writing to a remote database (LIS, EHR), import of requests in XML
- Output
- Interpretation report for the prescriber
How it works
From a measurement to a personalised dosage
Four steps, from the sample to the recommendation sent to the prescriber. Fictitious example, consistent with the demonstration above.
02 Model
Population model
Each drug is described by a drug model file: a published population pharmacokinetic model, its covariates, its therapeutic targets and the available doses.
03 Computation
Bayesian inference
The computation engine confronts the model with the patient's measurement. The broad population distribution narrows into an individual a posteriori estimate, with its uncertainty.
04 Recommendation
Personalised dosage
Tucuxi assesses the current dosage against the target, suggests an adjustment and predicts its effect. The interpretation report is ready to send; the decision remains with the prescriber.
An evaluated and published technology
More than ten years of academic work and scientific publications, from the design of the software to its evaluation on patient data.
Piperacillin: target attainment by dosing strategy
Estimated probability of reaching the target trough (8 to 32 mg/L), 80 treatment courses analysed with Tucuxi
A posteriori Bayesian adjustment (MIPD)Other strategies
Publications
- 2025Individualization of piperacillin dosage based on therapeutic drug monitoring with or without model-informed precision dosing: a scenario analysisHaefliger D. et al. J Antimicrob Chemother
- 2024Bayesian vancomycin model selection for therapeutic drug monitoring in neonatesAlrahahleh D. et al. Clin Pharmacokinet
- 2024Pharmacokinetic consideration of venetoclax in acute myeloid leukemia patients: a potential candidate for TDM?Philippe M. et al. Ther Drug Monit
- 2023AUC-based monitoring and model-informed precision dosing of vancomycin in critically ill patients: why and how?Goutelle S. et al. Anaesth Crit Care Pain Med
- 2022Implementation and cross-validation of a pharmacokinetic model for precision dosing of busulfan in hematopoietic stem cell transplanted childrenGoutelle S. et al. Pharmaceutics
- 2022Therapeutic drug monitoring of cefepime in a non-critically ill population: retrospective assessment and potential role for model-based dosingSuttels V. et al. JAC Antimicrob Resist
- 2022Implementation and comparison of two pharmacometric tools for model-based therapeutic drug monitoring and precision dosing of daptomycinHeitzmann J. et al. Pharmaceutics
Show the 5 earlier publications
- 2021Bayesian forecasting for intravenous tobramycin dosing in adults with cystic fibrosis using one versus two serum concentrations in a dosing intervalDrennan P.G. et al. Ther Drug Monit
- 2021Predictive performance of Bayesian vancomycin monitoring in the critically illNarayan S.W. et al. Crit Care Med
- 2020The steps to therapeutic drug monitoring: a structured approach illustrated with imatinibBuclin T. et al. Front Pharmacol
- 2017TUCUXI: an intelligent system for personalized medicine, from individualization of treatments to research databases and backDubovitskaya A. et al. ACM-BCB
- 2012Benchmarking therapeutic drug monitoring software: a review of available computer toolsFuchs A. et al. Clin Pharmacokinet
Demonstrator
The Tucuxi core, in your browser
A fictitious patient, five public drug models and the actual Tucuxi computation engine. Change a value: the prediction and the suggested dosages are recalculated.
The demonstrator requires JavaScript and WebAssembly. For a guided demonstration, contact us.
Local computation by tucuxi-core (HEIG-VD and CHUV, AGPL v3 licence, sources), drug models from the public tucuxi-drugs repository. Nothing is sent to a server. This demonstrator is not CE marked and must never be used to determine a patient's dosage.
Context
Therapeutic drug monitoring
TDM consists of measuring the concentration of a drug in the patient in order to adjust its dosage. It applies to narrow therapeutic index drugs, whose effective and toxic doses are close, and to drugs whose exposure varies widely from one patient to another.
Underdosing exposes the patient to treatment failure, overdosing to adverse effects. TDM also helps to check adherence, control exposure to expensive drugs and make better use of antibiotics in the face of antimicrobial resistance.
Empirical TDM
The measurement, then a rule
The concentration is compared with a reference range, then the dose is adjusted proportionally. Interpretation assumes steady state and precise sampling times.
MIPD
The measurement, within a model
A population model and Bayesian inference use every measurement, including outside steady state, take covariates into account and predict the effect of an adjustment.
A recommended practice, with growing adoption
Guidelines and recent literature agree: Bayesian adjustment improves the attainment of exposure targets. Its uptake is held back by access to tools, their integration and training.
-
AUC
For vancomycin, the 2020 North American consensus guideline recommends monitoring guided by the area under the curve, preferably estimated with Bayesian software.[1]
-
15 studies
A 2026 systematic review identifies uses of Bayesian MIPD in clinical practice, mostly for anti-infectives. Barriers cited: training, IT support and licence costs.[2]
-
Randomised trial
In kidney transplantation, a tacrolimus MIPD application embedded in the patient record enabled more patients to reach the target than usual prescribing.[3]
- Rybak MJ et al. Therapeutic monitoring of vancomycin for serious MRSA infections: a revised consensus guideline. Am J Health Syst Pharm. 2020;77(11):835-864. doi:10.1093/ajhp/zxaa036
- Alghamdi WA. Clinical use of Bayesian model-informed precision dosing in routine practice: a focused systematic review. J Clin Med. 2026;15(10):3838. doi:10.3390/jcm15103838
- Kuypers DRJ et al. Electronic patient file-embedded model-informed precision dosing compared with physician dosing of tacrolimus in kidney transplantation. Clin Pharmacol Ther. 2026;119:381-392. doi:10.1002/cpt.70090
Application areas and models
Tucuxi lists 19 drugs and 43 models. Depending on the area, other uses of TDM can be explored with a suitable drug model.
- Anti-infectivesCefepime, daptomycin, darunavir, dolutegravir, doravirine, gentamicin, lopinavir, meropenem, piperacillin, rifampicin, teicoplanin, tobramycin, vancomycin.
- Oncology and haematologyBusulfan, imatinib, ponatinib, venetoclax.
- TransplantationTacrolimus.
- CardiologyApixaban.
- Inflammatory diseasesOther TDM uses to be exploredBiologics, monoclonal antibodies, anti-TNF agents: selection or creation of a drug model depending on the project.
- Neurology and psychiatryOther TDM uses to be exploredAntiepileptics, lithium, antidepressants, antipsychotics: selection or creation of a drug model depending on the project.
See the references of the 43 published models
- Apixaban
- Cirincione
- Busulfan
- Ben Hassine (adults), Ben Hassine (paediatrics), Paci
- Cefepime
- Buclin
- Daptomycin
- Dvorchik
- Darunavir
- Daskapan
- Dolutegravir
- Barcelo
- Doravirine
- Yee
- Gentamicin
- Fuchs
- Imatinib
- Gotta
- Lopinavir
- Fuchs
- Meropenem
- Li
- Piperacillin
- Chen, Li
- Ponatinib
- Hanley
- Rifampicin
- Svensson
- Tacrolimus
- Andrews, Cai, Monchaud, Nanga, Sikma, Storset, Han
- Teicoplanin
- Ogami
- Tobramycin
- Hennig
- Vancomycin
- Colin, Dao, De Cock, Frymoyer, Goti, Grimsley, Liu, Llopis-Salvia, Staatz, Thomson, Yamamoto, Kimura, Lee, Lo, Mehrotra, Mulubwa
- Venetoclax
- Brackman
Catalogue checked on 30 September 2026 against the official list of Tucuxi models. Several models may apply to different populations. The demonstrator above loads five drug models; this catalogue lists all the models published on the official website.
Populations
Dose adjustment is particularly useful when pharmacokinetics are variable or hard to predict: intensive care, paediatrics and neonatology, renal or hepatic impairment, obesity, renal replacement therapy and extracorporeal circulation.
Questions
Frequently asked questions
Another question about Tucuxi or a deployment project? Write to us.
Is Tucuxi open-source software?
Yes. Tucuxi is released as open source under the AGPL v3 licence, and its drug models are available in a public repository. Plug'n'Dose provides installation, integration with information systems, training and regulatory stewardship.
What is the regulatory status of Tucuxi?
CE marking is in progress, for class IIa medical device software under Regulation (EU) 2017/745. In the meantime, the open-source version is used under the responsibility of the institutions that install it.
Which drugs are available?
The catalogue published by Tucuxi lists 43 models for 19 drugs (see application areas and models). Five of these drug models are loaded in our demonstrator. Any drug with a published population pharmacokinetic model can be given a new drug model, created with the online editor. We support model selection and the creation of the drug model.
Is a concentration measurement needed to use Tucuxi?
No. Without a measurement, Tucuxi provides an a priori prediction based on the population model and the patient's characteristics. Each measured concentration then refines the a posteriori prediction, including outside steady state.
Does Tucuxi replace the pharmacologist's expertise?
No. Tucuxi is a decision support tool: it computes, positions the measurement and suggests a reasoned adjustment. The decision remains with the prescriber, informed by the opinion of the pharmacologist, pharmacist or laboratory biologist.
How does a deployment work?
We start from your priority drugs and your existing data flows. Together, we define the scope, the connection to the LIS or EHR, the drug models to configure and the training of your teams.
Contact
Let's talk about your project
Healthcare institutions, clinical laboratories, diagnostics manufacturers: write to us to explore how Tucuxi could be set up in your context.
Julien Massari, founder and President
LinkedIn profile
Plug'n'Dose SASU · RCS Nanterre 987 993 201 · Courbevoie