On inductive biases for the robust and interpretable prediction of drug concentrations using deep compartment models.

Conventional pharmacokinetic (PK) models contain several useful inductive biases guiding model convergence to more realistic predictions of drug concentrations. Implementing similar biases in standard neural networks can be challenging, but might be fundamental for model robustness and predictive performance. In this study, we build on the deep compartment model (DCM) architecture by introducing constraints that guide the model to explore more physiologically realistic solutions. Using a simulation study, we show that constraints improve robustness in sparse data settings. Additionally, predicted concentration-time curves took on more realistic shapes compared to unconstrained models. Next, we propose the use of multi-branch networks, where each covariate can be connected to specific PK parameters, to reduce the propensity of models to learn spurious effects. Another benefit of this architecture is that covariate effects are isolated, enabling model interpretability through the visualization of learned functions. We show that all models were sensitive to learning false effects when trained in the presence of unimportant covariates, indicating the importance of selecting an appropriate set of covariates to link to the PK parameters. Finally, we compared the predictive performance of the constrained models to previous relevant population PK models on a real-world data set of 69 haemophilia A patients. Here, constrained models obtained higher accuracy compared to the standard DCM, with the multi-branch network outperforming previous PK models. We conclude that physiological-based constraints can improve model robustness. We describe an interpretable architecture which aids model trust, which will be key for the adoption of machine learning-based models in clinical practice.

A Generative and Causal Pharmacokinetic Model for Factor VIII in Hemophilia A: A Machine Learning Framework for Continuous Model Refinement.

In rare diseases, such as hemophilia A, the development of accurate population pharmacokinetic (PK) models is often hindered by the limited availability of data. Most PK models are specific to a single recombinant factor VIII (rFVIII) concentrate or measurement assay, and are generally unsuited for answering counterfactual ("what-if") queries. Ideally, data from multiple hemophilia treatment centers are combined but this is generally difficult as patient data are kept private. In this work, we utilize causal inference techniques to produce a hybrid machine learning (ML) PK model that corrects for differences between rFVIII concentrates and measurement assays. Next, we augment this model with a generative model that can simulate realistic virtual patients as well as impute missing data. This model can be shared instead of actual patient data, resolving privacy issues. The hybrid ML-PK model was trained on chromogenic assay data of lonoctocog alfa and predictive performance was then evaluated on an external data set of patients who received octocog alfa with FVIII levels measured using the one-stage assay. The model presented higher accuracy compared with three previous PK models developed on data similar to the external data set (root mean squared error = 14.6 IU/dL vs. mean of 17.7 IU/dL). Finally, we show that the generative model can be used to accurately impute missing data (< 18% error). In conclusion, the proposed approach introduces interesting new possibilities for model development. In the context of rare disease, the introduction of generative models facilitates sharing of synthetic data, enabling the iterative improvement of population PK models.

ChatGPT in pharmacometrics? Potential opportunities and limitations.

The potential of using ChatGPT in pharmacometrics was explored in this study, with a focus on developing a population pharmacokinetic (PK) model for standard half-life factor VIII. Our results demonstrated that ChatGPT can be utilized to accurately obtain typical PK parameters from literature, generate a population PK model in R and develop an interactive Shiny application to visualize the results. ChatGPT's language generation capabilities enabled the development of R codes with minimal programming knowledge and helped to identify as well fix errors in the code. While ChatGPT presents several advantages, such as its ability to streamline the development process, its use in pharmacometrics also has limitations and challenges, including the accuracy and reliability of AI-generated data, the lack of transparency and reproducibility regarding codes generated by ChatGPT. Overall, our study demonstrates the potential of using ChatGPT in pharmacometrics, but researchers must carefully evaluate its use for their specific needs.

Adoption of Machine Learning in Pharmacometrics: An Overview of Recent Implementations and Their Considerations.

Pharmacometrics is a multidisciplinary field utilizing mathematical models of physiology, pharmacology, and disease to describe and quantify the interactions between medication and patient. As these models become more and more advanced, the need for advanced data analysis tools grows. Recently, there has been much interest in the adoption of machine learning (ML) algorithms. These algorithms offer strong function approximation capabilities and might reduce the time spent on model development. However, ML tools are not yet an integral part of the pharmacometrics workflow. The goal of this work is to discuss how ML algorithms have been applied in four stages of the pharmacometrics pipeline: data preparation, hypothesis generation, predictive modelling, and model validation. We will also discuss considerations before the use of ML algorithms with respect to each topic. We conclude by summarizing applications that hold potential for adoption by pharmacometricians.

Deep compartment models: A deep learning approach for the reliable prediction of time-series data in pharmacokinetic modeling.

Nonlinear mixed effect (NLME) models are the gold standard for the analysis of patient response following drug exposure. However, these types of models are complex and time-consuming to develop. There is great interest in the adoption of machine-learning methods, but most implementations cannot be reliably extrapolated to treatment strategies outside of the training data. In order to solve this problem, we propose the deep compartment model (DCM), a combination of neural networks and ordinary differential equations. Using simulated datasets of different sizes, we show that our model remains accurate when training on small data sets. Furthermore, using a real-world data set of patients with hemophilia A receiving factor VIII concentrate while undergoing surgery, we show that our model more accurately predicts a priori drug concentrations compared to a previous NLME model. In addition, we show that our model correctly describes the changing drug concentration over time. By adopting pharmacokinetic principles, the DCM allows for simulation of different treatment strategies and enables therapeutic drug monitoring.

Application of SHAP values for inferring the optimal functional form of covariates in pharmacokinetic modeling.

In population pharmacokinetic (PK) models, interindividual variability is explained by implementation of covariates in the model. The widely used forward stepwise selection method is sensitive to bias, which may lead to an incorrect inclusion of covariates. Alternatives, such as the full fixed effects model, reduce this bias but are dependent on the chosen implementation of each covariate. As the correct functional forms are unknown, this may still lead to an inaccurate selection of covariates. Machine learning (ML) techniques can potentially be used to learn the optimal functional forms for implementing covariates directly from data. A recent study suggested that using ML resulted in an improved selection of influential covariates. However, how do we select the appropriate functional form for including these covariates? In this work, we use SHapley Additive exPlanations (SHAP) to infer the relationship between covariates and PK parameters from ML models. As a case-study, we use data from 119 patients with hemophilia A receiving clotting factor VIII concentrate peri-operatively. We fit both a random forest and a XGBoost model to predict empirical Bayes estimated clearance and central volume from a base nonlinear mixed effects model. Next, we show that SHAP reveals covariate relationships which match previous findings. In addition, we can reveal subtle effects arising from combinations of covariates difficult to obtain using other methods of covariate analysis. We conclude that the proposed method can be used to extend ML-based covariate selection, and holds potential as a complete full model alternative to classical covariate analyses.

The actual use of directional steering and shorter pulse width in selected patients undergoing deep brain stimulation.

INTRODUCTION: Directional deep brain stimulation (DBS) and pulse with <60µs increase side-effects threshold, enlarging the therapeutic window. However, new systems allowing these advanced features are more expensive and often available only for a limited number of patients in some centers. It is unknown how many and which DBS patients actually need the advanced features because of an insufficient improvement with standard parameters. METHODS: We included in the analysis all patients with Parkinson's disease, dystonia and tremor who were selected to receive implantation of advanced DBS systems based on specific preoperative or intraoperative clinical features. RESULTS: After a median follow-up of 15 months, 54.9% of the 51 patients implanted with directional leads were using the advanced features in one or both leads (n = 42 leads, 42%), meaning these leads were programmed either with directional stimulation (n = 9, 9%), a shorter pw (n = 20, 20%) or both (n = 13, 13%). This included 92% of patients implanted in the Vim, 44% of those implanted in the STN, and 40% of those implanted in the GPi. CONCLUSIONS: DBS systems with advanced features may be particularly indicated for selected patients based on some clinical characteristics and the chosen target. This data may help clinicians allocate resources in a more informed way.

Intraoperative Use of Cone-Beam Computed Tomography in the Treatment of Atlantoaxial Rotatory Subluxation.

BACKGROUND: Atlantoaxial rotatory subluxation (AARS) is a rare pathological condition of the upper cervical spine. It can be caused by multiple mechanisms, including minor neck manipulations. Children are more prone owing to the weaker periarticular soft tissue and a steeper slant of the C1 facet plane against the vertical axis of the dens. If AARS does not resolve spontaneously, a normal position of the atlantoaxial joint must be achieved by reduction and stabilization. CASE DESCRIPTION: A 15-year-old girl had presented with a painful torticollis that had already been present for 4 weeks after trimaxillary jaw correction for skeletal class II malocclusion. A computed tomography (CT) scan of the cervical spine showed AARS Field and Hawkins classification type I. We first attempted 1 week of conservative treatment with a soft collar and the prescription of a muscular relaxant. However, because the AARS persisted, we performed transoral closed reduction with the patient under general anesthesia, as previously described. During the procedure, we used intraoperative cone-beam CT to evaluate the degree of reduction. After obtaining complete reduction, immobilization with a halo-vest was applied. CONCLUSIONS: Complete reduction of the AARS was achieved with closed intraoral reduction. We used intraoperative cone-beam CT to confirm complete reduction. We found cone-beam CT to be a very useful tool.

Novel Low-Twist Bast Fibre Yarns from Flax Tow for High-Performance Composite Applications.

The use of natural fibres for components subjected to higher mechanical requirements tends to be limited by the high price of high-quality semi-finished products. Therefore, the present study deals with the development of more cost-effective staple fibre yarns made from flax tow. In the subsequent processing stage, the yarns were processed into quasi-unidirectional (UD) fabrics. The results of the fibre characterisation along the process chain have shown that no significant mechanical fibre damage occurs after slivers' production. Fibres prepared from yarns and fabrics show comparable characteristics. The yarns were processed to composites by pultrusion to verify the reinforcement effect. The mechanical properties were comparable to those of composites made from a high-quality UD flax roving. The fabrics were industrially processed into composite laminates using a vacuum infusion and an autoclave injection process (vacuum injection method in an autoclave). While impact strength compared to a reference laminate based on the UD flax roving was achieved, tensile and flexural properties were not reached. An analysis showed that the staple fibre yarns in the fabric show an undulation, leading to a reorientation of the fibres and lower characteristic values, which show 86-92% of the laminate made from the flax roving. Hybrid laminates with outer glass and inner flax layers were manufactured for the intended development of a leaf spring for the bogie of a narrow-gauge railroad as a demonstrator. The hybrid composites display excellent mechanical properties and showed clear advantages over a pure glass fibre-reinforced composite in lightweight construction potential, particularly flexural stiffness.

Multiple nocardial abscesses of the brainstem and spinal cord diagnosed after an open biopsy through a cervical partial central corpectomy: case report.

Nocardiosis of the central nervous system is a challenging and difficult diagnosis for the clinician. The combination of infections of the brain and spinal cord is even more rare. The authors report on a patient with multiple lesions in the brainstem and cervical spinal cord. This 81-year-old immunocompetent woman presented with symptoms of progressive walking difficulty and ataxia. The results of an extensive workup with laboratory investigation, MRI, lumbar puncture, positron emission tomography (PET), and bone marrow biopsy remained inconclusive. Only after an open biopsy of a cervical lesion by an anterior approach through a partial central corpectomy of the cervical spine, was the diagnosis of nocardiosis made, allowing for specific antibiotic treatment.

Comparison of different equations to assess glomerular filtration in critically ill patients.

PURPOSE: To evaluate equations for estimation of glomerular filtration rate (GFR) and measured urinary creatinine clearance, compared to measured GFR in critically ill patients. METHODS: GFR was measured using inulin clearance. Multiple blood samples were collected per patient for determination of serum creatinine, cystatin C and inulin. GFR was estimated by the use of the following estimation equations (eGFR): four commonly used creatinine-based equations [Cockcroft-Gault, Modification of Diet in Renal Disease (both the short and long formula) and Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI)], five cystatin C based estimation equations (Hoek, Larsson, Filler, Le Bricon, CKD-EPIcys) and one equation combining cystatin C and serum creatinine (CKD-EPIcr-cys). In addition we measured urinary creatinine clearance. Bias, precision and accuracy of all estimates were compared to those of the inulin clearance. RESULTS: Data were collected from 83 patients, of whom 68 were considered evaluable. The median age was 58 years [interquartile range (IQR) 39-68]. The median inulin clearance was 80 mL/min/1.73 m(2) (IQR 31-114). Equations based on creatinine had much bias and poor precision and accuracy. Measured urinary creatinine clearances overestimated GFR. Equations based on cystatin C were free of bias, but also had limited precision and accuracy. CONCLUSIONS: In this cohort of patients, estimates of GFR had low accuracy and precision. Cystatin C based formulas, especially CKD-EPIcr-cys, showed limited bias; however, the accuracy and precision of these estimates were still insufficient. Measured urinary creatinine clearance overestimates GFR, but may provide a cheap alternative, when this is taken into account.

Second family with the Boston-type craniosynostosis syndrome: novel mutation and expansion of the clinical spectrum.

Craniosynostosis, caused by early fusion of one or more cranial sutures, can affect the coronal or lambdoid sutures, or include premature fusion of the sagittal (scaphocephaly) or metopic suture (trigonocephaly). Often occurring as isolated finding, their co-existence in a craniosynostosis syndrome is infrequent. We describe a four-generation family with variable expression of a craniosynostosis phenotype with scaphocephaly and a particularly severe trigonocephaly. Molecular analysis revealed a missense mutation in the MSX2-associated with the Boston-type craniosynostosis syndrome-affecting the same amino-acid residue as in the original Boston family. Besides unique features such as the cranial sutures involved, minor limb abnormalities and incomplete penetrance, our patients share with the original family autosomal dominant inheritance and the presence of multiple endocranial erosions on CT imaging. Though these findings appear to be important diagnostic clues for MSX2-related craniosynostosis, it is noteworthy that the first affected generation in this family presented merely with isolated sagittal or unicoronal craniosynostosis and cutaneous syndactyly. Molecular analysis of MSX2 should therefore be considered in patients with isolated scaphocephaly/unicoronal synostosis, especially in the presence of a family history for craniosynostosis or syndactyly.