Digital dosing and the future of therapeutic drug monitoring: insights from the 23rd IATDMCT congress

When it comes to therapeutics, the core clinical challenge is often distilled down to a critical question: how much drug, and for how long? For too long, the aspiration toward truly personalised medicine has been constrained by the limitations of static, one-size-fits-all dosing and outdated pharmacokinetic (PK) models – a rigid system ill-equipped for advanced medications like biological treatments and powerful new compounds. 

But what if the digital tools transforming our industry could give you back more time?

The integration of modern digital technologies can fundamentally change this dynamic, shifting therapeutic drug monitoring (TDM) from a reactive measurement tool to a proactive, predictive clinical strategy. This digital transformation provides clinicians with efficient, accurate decision-making capabilities while enabling laboratories and hospitals to achieve better outcomes, reduced costs, and improved patient experience [1, 2].

The feasibility of this vision was one of the central themes at the 23rd International Association of Therapeutic Drug Monitoring and Clinical Toxicology (IATDMCT) conference in Singapore last year. Held in September, the event convened an elite cohort of the industry’s foremost experts and strategic pioneers to showcase precisely how these tools are operationalised. 

The Lab Insights team noted key insights, drawn from four insightful presentations, which demonstrate a multi-faceted approach to digitalising TDM and clinical toxicology, validating its role in the move toward personalised care: 

• • Dr Amy Legg explored the strategic role of artificial intelligence (AI) and Big Data in reshaping TDM research, emphasising the need for dynamic predictive models. Her work, rooted in clinical pharmacy and antimicrobial stewardship, highlights the potential of embedding research within digital electronic health record (EHR) systems to address complex clinical questions, from rare drug interactions to population-specific PK.
• • Dr Tomoyuki Yamada showcased how leveraging vast sources of real-world data (RWD) provides critical safety signals and advances clinical practices in toxicology and drug monitoring. As Head Pharmacist leading pharmacovigilance efforts, Dr Yamada’s case studies – such as assessing daptomycin-statin interactions and evaluating follow-up TDM to reduce vancomycin toxicity – demonstrate the tangible impact of RWD analytics on patient outcomes.
• • Dr Dirk Jan Moes detailed the strategic application of Model-Informed Precision Dosing (MIPD) for complex biologics in oncology, highlighting the quantifiable financial and clinical benefits of integrated digital systems. As an associate professor of clinical pharmacometrics, Dr Moes focused on how MIPD strategies can personalise dosing for specific biologics – including Daratumab and Nivolumab – to reduce both financial and clinical toxicity.
• • Dr Tomoyuki Mizuno demonstrated the cutting-edge use of machine learning (ML) and integrated decision support dashboards to facilitate hyper-precision dosing, particularly within challenging pediatric patient care settings. Dr Mizuno, who directs a Pharmacometrics Centre of Excellence, focused on enhancing the MIPD framework using ML to address the unique complexities of dosing for children, providing clinicians with innovative bedside tools.

Together, these experts illuminate the essential digital pillars that will define the next generation of diagnostic value and patient care.

Model-informed precision dosing (MIPD) as the foundational platform

No two bodies are identical, and there is a clinical need for individualisation. “Our body is not one homogeneous fish tank alike. Drugs are distributed differently,” stressed Professor Andrea Kwa of Singapore General Hospital in her keynote speech. The goal of precision dosing is in the identification of the most optimal dosing regimen that will provide required efficacy with minimal toxicity in each individual patient [1]. MIPD has emerged as the essential quantitative framework for achieving this goal. It utilises sophisticated pharmacokinetic/pharmacodynamic (PK/PD) models to integrate and employ patient-specific information and clinical observations, individualising doses at multiple steps throughout drug therapy.

This approach functions as a predict, confirm, learn, and apply cycle:

• • Prediction (front dose adaption): Baseline patient data – such as bodyweight, age, genetic makeup, and organ function – are collected as covariates to inform the model and design the initial starting dose. Dr Moes referred to this crucial first step as “front dose adaptation” [2].
  • Confirmation and refinement: To control for high between-patient variability, observed data (e.g. drug concentration measurement and PD biomarkers) are used as Bayesian feedback to update the model, allowing it to reflect the patient’s actual drug handling and phenotype. [2] This iterative process allows for the necessary dose refinement post biomarker collection, which is especially critical when managing more complex drugs like biologics in hemato-oncology, where Dr Moes stressed the importance of efficacy and safety for challenging patient populations. [2]

A key enabler to the effective use of MIPD therefore requires robust clinical infrastructure, reliable laboratory assays (e.g., Liquid Chromatography-Mass Spectrometry for multiplexing multiple antibodies), and, crucially, the integration of EHRs and laboratory information systems (LIS) directly into the dosing tool [1]. One example of the latter is a programme named RoadMAB, an operation dosing software developed by Dr Mizuno and his team at Cincinnati Children’s Hospital Medical Center (CCHMC) [3, 4]. This tool is specifically designed to be integrated into EHRs to automatically correct necessary patient data for analysis, effectively closing the digital loop between data, diagnosis, and decision.  

The power of real-world data (RWD) and big data analytics

While MIPD provides the crucial quantitative framework for individualised dosing, the true power of this digital approach is realised when its predictive models are continuously refined by real-world evidence. This is where RWD and big data analytics become indispensable tools for advancing both TDM and clinical toxicology [5]. 

The strength of RWD lies in its wide-ranging collection. RWD is sourced from spontaneous reporting databases, claims databases and EHRs, providing complementary insights crucial for detecting adverse events and evaluating current clinical practice patterns [6]. This vast, complex data set is essential for hypothesis generation: large, observational RWD sets – such as the BLUeY registry for beta-lactam TDM as cited by Dr Legg [5] – allow ML to identify early, hypothesis-generating signals that are often missed in smaller, traditional studies due to rare outcomes or complex patient variability [8]. 

More crucially, this data translates directly into demonstrable clinical impact. Analysis of claims data, for example, has shown that consistent follow-up TDM can dramatically reduce the incidence of drug-induced toxicity, with one study noting that this approach lowered vancomycin-induced nephrotoxicity incidence from 60.9% to 9.9%, as shared by Dr Yamada [6]. 

AI and machine learning (ML) for next-generation decision support

The ability to generate hypotheses and validate practice patterns using RWD creates a vital demand for sophisticated analytical tools capable of processing such massive, complex datasets. It is through the advanced computational power of AI and ML that diagnostic laboratories and clinical teams can convert this deluge of data into predictive, actionable, and personalised dosing insights.

• • Dynamic modelling: AI is critical for handling the large sample sizes and confounding variables necessary to develop more complex and dynamic population PK (PopPK) models that can adjust to a patient’s changing clinical parameters over time, as highlighted by Dr Legg [8].
  • Hybrid models: Dr Mizuno also stressed the most promising approaches move beyond siloed technologies. This involves combining the clinical plausibility of conventional Bayesian estimation with ML (e.g., using an XGBoost model) to correct prediction errors. This results in robust hybrid models that consistently outperform conventional PopPK approaches [3].
  • Automation and agentic AI: Looking ahead, future systems are using reinforcement learning to create AI agents that can automatically collect patient data, assess treatment options, and suggest dosing regimens and intervals. Dr Mizuno has also stressed that this advancement streamlines the entire dosing process and elevates human decision-making [3].

Strategic value and return on investment (ROI) for diagnostic leadership

Altogether, the integration of MIPD frameworks, RWD, and predictive AI is more than an academic concept; it represents a fundamental shift in the value proposition of diagnostics. For laboratory and hospital leadership, this convergence translates directly into quantifiable benefits that impact both the patient bottom line and the organisational bottom line.

A) Enhanced patient safety and clinical outcomes

The primary benefit of digital transformation is its power to elevate diagnostic services from passive reporting to active, personalised guidance, leading to better patient care:

• • Precision and accuracy: AI algorithms provide a level of predictive capability that was previously unattainable. For instance, AI tools have demonstrated the ability to predict key clinical values, such as the day five International Normalised Ratio (INR) for warfarin, with accuracy that can match or slightly outperform human experts [8]. This predictive power minimises dangerous delays in optimal treatment.
  • Proactive intervention: By linking predictive models with clinical workflow, digital decision support dashboards enable clinicians to receive immediate, PK-guided dose recommendations. According to Dr Mizuno, this facilitates proactive intervention, supporting better treatment outcomes by ensuring patients achieve and maintain therapeutic targets faster, thereby reducing morbidity and hospital stays [3]. 

B) Operational efficiency and financial toxicity reduction

Digital TDM acts as a powerful lever for cost control, particularly regarding expensive specialty drugs, turning drug monitoring into a source of tangible financial return:

• • Cost containment: Optimised dosing directly addresses the high cost of expensive therapies by minimising drug waste and exposure. By ensuring every dose is as precise as possible, hospitals can significantly mitigate the growing challenge of financial toxicity associated with advanced treatments [2]. 
  • Quantifiable ROI: The strategic use of PK/PD models and in silico dose adjustments has delivered significant, measurable cost savings:
    • • Drug waste reduction: In dose adjustments for certain antibody-drug conjugates (ADCs), the strategy of aiming to use only whole drug vials resulted in an estimated 25% reduction in drug expenses [2]. 
      • Extended intervals: In Dr Moes’s same presentation, model-informed dose optimisation can also improve patient convenience and reduce costs simultaneously. Developing body weight-dependent dosing intervals for subcutaneous immune checkpoint inhibitors (e.g., nivolumab) has been shown to save more than 30% of yearly drug expenses while maintaining stringent clinical efficacy criteria [2].

C) Advancing the diagnostic continuum

Digital transformation fundamentally shifts the position of the laboratory within the care continuum – it is no longer just a service provider, but a strategic partner in therapeutics:

• • EHR integration: Developing tools like the RoadMAB software, which integrate MIPD and ML directly into the EHR, automates the correction of necessary patient data and simplifies the analysis for clinicians [3, 4]. This innovation shifts the diagnostic value proposition from a static lab result to a final, actionable therapeutic recommendation.
  • New assays and multiplexing: Strategic investment in next-generation assays, such as LCMS as noted by Dr Moes, enables the simultaneous measurement of multiple drug concentrations or antibodies in a single patient sample [2]. This enhances the lab’s service portfolio, increases efficiency, and supports the complex needs of biologics monitoring. 

These realised efficiencies underscore a critical paradigm shift: digital transformation in TDM operates as a value generator, yielding a discernible and substantial return on investment (ROI) rather than merely incurring overhead.

Roadmap to transformation: a call to action for leadership

Realising this significant ROI and sustaining the advanced diagnostic continuum requires a deliberate, forward-thinking strategy that addresses both organisational readiness and technical governance. For laboratory and hospital leadership, the path forward is not just about adopting technology, but about driving institutional change. For this to be achievable, a clear set of actions are required:

• • Interdisciplinary collaboration: Attaining a holistic vision of integrated care requires breaking down traditional silos. Effective collaboration is essential across key stakeholders, including clinical pharmacy, laboratory services, IT/Software development, and hospital management, to ensure seamless tool integration and workflow adoption [1].
  • Governance and quality control: Digital technologies, especially AI, require careful management. Leadership must prioritise robust data quality, rigorous external validation, and strong governance to ensure that models – particularly those leveraging RWD – are safe and trustworthy for broad patient care [8]. Establishing clear protocols protects against liability and ensures patient safety.
  • Investment focus: Future investment should focus on integrating systems rather than buying isolated tools. This includes integrating full pharmacodynamics (PK/PD models), advanced concepts like digital twins, and refined AI capabilities directly into clinical decision-support dashboards. This approach will further modernise and transform TDM, ensuring the ultimate goal: the right drug, right dose, at the right time for every patient [2, 3].

Ultimately, the digital future of TDM – as confirmed by these industry leaders – is about precision with a purpose. It’s a chance for diagnostics to become an active, integrated partner in care. This transformation offers leaders a clear competitive advantage with measurable returns, but the real win is the ability to fundamentally elevate the human experience of medicine by making personalised safety the core standard for everyone.

References

[1] Mizuno, T., Vinks, A.A., Fukuda, T., Rosenheck, R., Wetterland, L., de Leon, J., Hartman, S., Arnold, L.M. and Patino, L.R. (2022) ‘Clinical implementation of pharmacogenetics and model-informed precision dosing to improve patient care’, British Journal of Clinical Pharmacology, 88(4), pp. 1418–1426.

[2] Moes, D.J. (2025) ‘Model Informed (precision) dosing of biologics in (hemato)oncology: challenges and opportunities?’, 23rd International Association of Therapeutic Drug Monitoring and Clinical Toxicology (IATDMCT) Conference. Grand Copthorne Waterfront Hotel, Singapore, 21–24 September.

[3] Mizuno, T. (2025) ‘Leveraging machine learning and decision support dashboard to facilitate precision dosing in pediatric patient care’, 23rd International Association of Therapeutic Drug Monitoring and Clinical Toxicology (IATDMCT) Conference. Grand Copthorne Waterfront Hotel, Singapore, 21–24 September.

[4] Colman, R.J., Xiong, Y., Mizuno, T., Xie, C., Vinks, A.A., Hyams, J.S., Denson, L.A. and Minar, P. (2023) ‘Model-informed Precision Dosing for Biologics Is Now Available at the Bedside: A Roadmap for Strategies and Considerations for Real-world Implementation’, Inflammatory Bowel Diseases, 29(1), pp. 119–127. doi: 10.1093/ibd/izac063.

[5] Maier, C., Hartung, N., Kloft, C., Huisinga, W. and Henrich, W. (2021) ‘A continued learning approach for model-informed precision dosing: updating models in clinical practice’, CPT: Pharmacometrics & Systems Pharmacology, 10(4), pp. 311–321.

[6] Yamada, T. (2025) ‘Real-World Big Data Insights for Advancing Therapeutic Drug Monitoring and Clinical Toxicology’, 23rd International Association of Therapeutic Drug Monitoring and Clinical Toxicology (IATDMCT) Conference. Grand Copthorne Waterfront Hotel, Singapore, 21–24 September.

[7] Pai Mangalore, R., Peel, T.N., Udy, A.A. and Peleg, A.Y. (2023) ‘The clinical application of beta-lactam antibiotic therapeutic drug monitoring in the critical care setting’, Journal of Antimicrobial Chemotherapy, 78(10), pp. 2395–2405. Available at: https://doi.org/10.1093/jac/dkad223 (Accessed: January 26, 2026).

[8] Legg, A. (2025) ‘TDM research in the digital age?’, 23rd International Association of Therapeutic Drug Monitoring and Clinical Toxicology (IATDMCT) Conference. Grand Copthorne Waterfront Hotel, Singapore, 21–24 September.