Toward mechanistic medical digital twins: some use cases in immunology.

A fundamental challenge for personalized medicine is to capture enough of the complexity of an individual patient to determine an optimal way to keep them healthy or restore their health. This will require personalized computational models of sufficient resolution and with enough mechanistic information to provide actionable information to the clinician. Such personalized models are increasingly referred to as medical digital twins. Digital twin technology for health applications is still in its infancy, and extensive research and development is required. This article focuses on several projects in different stages of development that can lead to specific-and practical-medical digital twins or digital twin modeling platforms. It emerged from a two-day forum on problems related to medical digital twins, particularly those involving an immune system component. Open access video recordings of the forum discussions are available.

Forum on immune digital twins: a meeting report.

Medical digital twins are computational models of human biology relevant to a given medical condition, which are tailored to an individual patient, thereby predicting the course of disease and individualized treatments, an important goal of personalized medicine. The immune system, which has a central role in many diseases, is highly heterogeneous between individuals, and thus poses a major challenge for this technology. In February 2023, an international group of experts convened for two days to discuss these challenges related to immune digital twins. The group consisted of clinicians, immunologists, biologists, and mathematical modelers, representative of the interdisciplinary nature of medical digital twin development. A video recording of the entire event is available. This paper presents a synopsis of the discussions, brief descriptions of ongoing digital twin projects at different stages of progress. It also proposes a 5-year action plan for further developing this technology. The main recommendations are to identify and pursue a small number of promising use cases, to develop stimulation-specific assays of immune function in a clinical setting, and to develop a database of existing computational immune models, as well as advanced modeling technology and infrastructure.

A systematic efficacy analysis of tuberculosis treatment with BPaL-containing regimens using a multiscale modeling approach.

Tuberculosis (TB) is a life-threatening infectious disease. The standard treatment is up to 90% effective; however, it requires the administration of four antibiotics (isoniazid, rifampicin, pyrazinamide, and ethambutol [HRZE]) over long time periods. This harsh treatment process causes adherence issues for patients because of the long treatment times and a myriad of adverse effects. Therefore, the World Health Organization has focused goals of shortening standard treatment regimens for TB in their End TB Strategy efforts, which aim to reduce TB-related deaths by 95% by 2035. For this purpose, many novel and promising combination antibiotics are being explored that have recently been discovered, such as the bedaquiline, pretomanid, and linezolid (BPaL) regimen. As a result, testing the number of possible combinations with all possible novel regimens is beyond the limit of experimental resources. In this study, we present a unique framework that uses a primate granuloma modeling approach to screen many combination regimens that are currently under clinical and experimental exploration and assesses their efficacies to inform future studies. We tested well-studied regimens such as HRZE and BPaL to evaluate the validity and accuracy of our framework. We also simulated additional promising combination regimens that have not been sufficiently studied clinically or experimentally, and we provide a pipeline for regimen ranking based on their efficacies in granulomas. Furthermore, we showed a correlation between simulation rankings and new marmoset data rankings, providing evidence for the credibility of our framework. This framework can be adapted to any TB regimen and can rank any number of single or combination regimens.

Development and Analysis of Multiscale Models for Tuberculosis: From Molecules to Populations.

Although infectious disease dynamics are often analyzed at the macro-scale, increasing numbers of drug-resistant infections highlight the importance of within-host modeling that simultaneously solves across multiple scales to effectively respond to epidemics. We review multiscale modeling approaches for complex, interconnected biological systems and discuss critical steps involved in building, analyzing, and applying such models within the discipline of model credibility. We also present our two tools: CaliPro, for calibrating multiscale models (MSMs) to datasets, and tunable resolution, for fine- and coarse-graining sub-models while retaining insights. We include as an example our work simulating infection with Mycobacterium tuberculosis to demonstrate modeling choices and how predictions are made to generate new insights and test interventions. We discuss some of the current challenges of incorporating novel datasets, rigorously training computational biologists, and increasing the reach of MSMs. We also offer several promising future research directions of incorporating within-host dynamics into applications ranging from combinatorial treatment to epidemic response.

Semi-automated colony-forming unit counting for biosafety level 3 laboratories.

Biosafety level 3 decontamination precautions motivate measuring microbial colonies using consumable photography instead of expensive automated plate counters or smartphones, and assaying drug treatments-with multiple concentrations per treatment, replicates, and controls-produces hundreds of images. Here, we present a protocol for semi-automated image analysis by hand-tuning three parameters. The parameters control for non-uniform colony growth and artifacts such as lid condensation, reflections, and plating streaks. We describe steps to prepare images, tune parameters, and plot dose-response relationships. For complete details on the use and execution of this protocol, please refer to Larkins-Ford et al.1.

Optimizing tuberculosis treatment efficacy: Comparing the standard regimen with Moxifloxacin-containing regimens.

Tuberculosis (TB) continues to be one of the deadliest infectious diseases in the world, causing ~1.5 million deaths every year. The World Health Organization initiated an End TB Strategy that aims to reduce TB-related deaths in 2035 by 95%. Recent research goals have focused on discovering more effective and more patient-friendly antibiotic drug regimens to increase patient compliance and decrease emergence of resistant TB. Moxifloxacin is one promising antibiotic that may improve the current standard regimen by shortening treatment time. Clinical trials and in vivo mouse studies suggest that regimens containing moxifloxacin have better bactericidal activity. However, testing every possible combination regimen with moxifloxacin either in vivo or clinically is not feasible due to experimental and clinical limitations. To identify better regimens more systematically, we simulated pharmacokinetics/pharmacodynamics of various regimens (with and without moxifloxacin) to evaluate efficacies, and then compared our predictions to both clinical trials and nonhuman primate studies performed herein. We used GranSim, our well-established hybrid agent-based model that simulates granuloma formation and antibiotic treatment, for this task. In addition, we established a multiple-objective optimization pipeline using GranSim to discover optimized regimens based on treatment objectives of interest, i.e., minimizing total drug dosage and lowering time needed to sterilize granulomas. Our approach can efficiently test many regimens and successfully identify optimal regimens to inform pre-clinical studies or clinical trials and ultimately accelerate the TB regimen discovery process.

Disease phenotypic and geospatial features vary across genetic lineages for Tuberculosis within Arkansas, 2010-2020.

Tuberculosis (TB) elimination in the United States remains elusive, and community-specific, localized intervention strategies may be necessary to meet elimination goals. A better understanding of the genotypic diversity of Mtb, the population subgroups affected by different TB strains, and differences in disease presentation associated with these strains can aid in identifying risk groups and designing tailored interventions. We analyze TB incidence and genotype data from all Arkansas counties over an 11-year time span from 2010 through 2020. We use statistical methods and geographic information systems (GIS) to identify demographic and disease phenotypic characteristics that are associated with different Mtb genetic lineages in the study area. We found the following variables to be significantly associated with genetic lineage (p<0.05): patient county, patient birth country, patient ethnicity, race, IGRA result, disease site, chest X-ray result, whether or not a case was identified as part of a cluster, patient age, occupation risk, and date arrived in the US. Different Mtb lineages affect different subpopulations in Arkansas. Lineage 4 (EuroAmerican) and Lineage 2 (East Asian) are most prevalent, although the spatial distributions differ substantially, and lineage 2 (East Asian) is more frequently associated with case clusters. The Marshallese remain a particularly high-risk group for TB in Arkansas.

Calibration methods to fit parameters within complex biological models.

Mathematical and computational models of biological systems are increasingly complex, typically comprised of hybrid multi-scale methods such as ordinary differential equations, partial differential equations, agent-based and rule-based models, etc. These mechanistic models concurrently simulate detail at resolutions of whole host, multi-organ, organ, tissue, cellular, molecular, and genomic dynamics. Lacking analytical and numerical methods, solving complex biological models requires iterative parameter sampling-based approaches to establish appropriate ranges of model parameters that capture corresponding experimental datasets. However, these models typically comprise large numbers of parameters and therefore large degrees of freedom. Thus, fitting these models to multiple experimental datasets over time and space presents significant challenges. In this work we undertake the task of reviewing, testing, and advancing calibration practices across models and dataset types to compare methodologies for model calibration. Evaluating the process of calibrating models includes weighing strengths and applicability of each approach as well as standardizing calibration methods. Our work compares the performance of our model agnostic Calibration Protocol (CaliPro) with approximate Bayesian computing (ABC) to highlight strengths, weaknesses, synergies, and differences among these methods. We also present next-generation updates to CaliPro. We explore several model implementations and suggest a decision tree for selecting calibration approaches to match dataset types and modeling constraints.

Concomitant immunity to M. tuberculosis infection.

Some persistent infections provide a level of immunity that protects against reinfection with the same pathogen, a process referred to as concomitant immunity. To explore the phenomenon of concomitant immunity during Mycobacterium tuberculosis infection, we utilized HostSim, a previously published virtual host model of the immune response following Mtb infection. By simulating reinfection scenarios and comparing with data from non-human primate studies, we propose a hypothesis that the durability of a concomitant immune response against Mtb is intrinsically tied to levels of tissue resident memory T cells (Trms) during primary infection, with a secondary but important role for circulating Mtb-specific T cells. Further, we compare HostSim reinfection experiments to observational TB studies from the pre-antibiotic era to predict that the upper bound of the lifespan of resident memory T cells in human lung tissue is likely 2-3 years. To the authors' knowledge, this is the first estimate of resident memory T-cell lifespan in humans. Our findings are a first step towards demonstrating the important role of Trms in preventing disease and suggest that the induction of lung Trms is likely critical for vaccine success.

T cell transcription factor expression evolves over time in granulomas from Mycobacterium tuberculosis-infected cynomolgus macaques.

Mycobacterium tuberculosis (Mtb), the causative agent of tuberculosis (TB), is a global health concern, yearly resulting in 10 million new cases of active TB. Immunologic investigation of lung granulomas is essential for understanding host control of bacterial replication. Here, we identify and compare the pathological, cellular, and functional differences in granulomas at 4, 12, and 20 weeks post-infection in Chinese cynomolgus macaques. Original granulomas differ in transcription-factor expression within adaptive lymphocytes, with those at 12 weeks showing higher frequencies of CD8+T-bet+ T cells, while CD4+T-bet+ T cells increase at 20 weeks post-infection. The appearance of T-bet+ adaptive T cells at 12 and 20 weeks is coincident with a reduction in bacterial burden, suggesting their critical role in Mtb control. This study highlights the evolution of T cell responses within lung granulomas, suggesting that vaccines promoting the development and migration of T-bet+ T cells would enhance mycobacterial control.

A virtual host model of Mycobacterium tuberculosis infection identifies early immune events as predictive of infection outcomes.

Tuberculosis (TB), caused by infection with Mycobacterium tuberculosis (Mtb), is one of the world's deadliest infectious diseases and remains a significant global health burden. TB disease and pathology can present clinically across a spectrum of outcomes, ranging from total sterilization of infection to active disease. Much remains unknown about the biology that drives an individual towards various clinical outcomes as it is challenging to experimentally address specific mechanisms driving clinical outcomes. Furthermore, it is unknown whether numbers of immune cells in the blood accurately reflect ongoing events during infection within human lungs. Herein, we utilize a systems biology approach by developing a whole-host model of the immune response to Mtb across multiple physiologic and time scales. This model, called HostSim, tracks events at the cellular, granuloma, organ, and host scale and represents the first whole-host, multi-scale model of the immune response following Mtb infection. We show that this model can capture various aspects of human and non-human primate TB disease and predict that biomarkers in the blood may only faithfully represent events in the lung at early time points after infection. We posit that HostSim, as a first step toward personalized digital twins in TB research, offers a powerful computational tool that can be used in concert with experimental approaches to understand and predict events about various aspects of TB disease and therapeutics.

Structural identifiability analysis of age-structured PDE epidemic models.

Computational and mathematical models rely heavily on estimated parameter values for model development. Identifiability analysis determines how well the parameters of a model can be estimated from experimental data. Identifiability analysis is crucial for interpreting and determining confidence in model parameter values and to provide biologically relevant predictions. Structural identifiability analysis, in which one assumes data to be noiseless and arbitrarily fine-grained, has been extensively studied in the context of ordinary differential equation (ODE) models, but has not yet been widely explored for age-structured partial differential equation (PDE) models. These models present additional difficulties due to increased number of variables and partial derivatives as well as the presence of boundary conditions. In this work, we establish a pipeline for structural identifiability analysis of age-structured PDE models using a differential algebra framework and derive identifiability results for specific age-structured models. We use epidemic models to demonstrate this framework because of their wide-spread use in many different diseases and for the corresponding parallel work previously done for ODEs. In our application of the identifiability analysis pipeline, we focus on a Susceptible-Exposed-Infected model for which we compare identifiability results for a PDE and corresponding ODE system and explore effects of age-dependent parameters on identifiability. We also show how practical identifiability analysis can be applied in this example.

Corrigendum: Spatial Organization and Recruitment of Non-Specific T Cells May Limit T Cell-Macrophage Interactions Within Mycobacterium tuberculosis Granulomas.

[This corrects the article DOI: 10.3389/fimmu.2020.613638.].

Neutrophil Dynamics Affect Mycobacterium tuberculosis Granuloma Outcomes and Dissemination.

Neutrophil infiltration into tuberculous granulomas is often associated with higher bacteria loads and severe disease but the basis for this relationship is not well understood. To better elucidate the connection between neutrophils and pathology in primate systems, we paired data from experimental studies with our next generation computational model GranSim to identify neutrophil-related factors, including neutrophil recruitment, lifespan, and intracellular bacteria numbers, that drive granuloma-level outcomes. We predict mechanisms underlying spatial organization of neutrophils within granulomas and identify how neutrophils contribute to granuloma dissemination. We also performed virtual deletion and depletion of neutrophils within granulomas and found that neutrophils play a nuanced role in determining granuloma outcome, promoting uncontrolled bacterial growth in some and working to contain bacterial growth in others. Here, we present three key results: We show that neutrophils can facilitate local dissemination of granulomas and thereby enable the spread of infection. We suggest that neutrophils influence CFU burden during both innate and adaptive immune responses, implying that they may be targets for therapeutic interventions during later stages of infection. Further, through the use of uncertainty and sensitivity analyses, we predict which neutrophil processes drive granuloma severity and structure.

To Sobol or not to Sobol? The effects of sampling schemes in systems biology applications.

Computational and mathematical models in biology rely heavily on the parameters that characterize them. However, robust estimates for their values are typically elusive and thus a large parameter space becomes necessary for model study, particularly to make translationally impactful predictions. Sampling schemes exploring parameter spaces for models are used for a variety of purposes in systems biology, including model calibration and sensitivity analysis. Typically, random sampling is used; however, when models have a high number of unknown parameters or the models are highly complex, computational cost becomes an important factor. This issue can be reduced through the use of efficient sampling schemes such as Latin hypercube sampling (LHS) and Sobol sequences. In this work, we compare and contrast three sampling schemes - random sampling, LHS, and Sobol sequences - for the purposes of performing both parameter sensitivity analysis and model calibration. In addition, we apply these analyses to different types of computational and mathematical models of varying complexity: a simple ODE model, a complex ODE model, and an agent-based model. In general, the sampling scheme had little effect when used for calibration efforts, but when applied to sensitivity analyses, Sobol sequences exhibited faster convergence. While the observed benefit to convergence is relatively small, Sobol sequences are computationally less expensive to compute than LHS samples and also have the benefit of being deterministic, which allows for better reproducibility of results.

CaliPro: A Calibration Protocol That Utilizes Parameter Density Estimation to Explore Parameter Space and Calibrate Complex Biological Models.

INTRODUCTION: Mathematical and computational modeling have a long history of uncovering mechanisms and making predictions for biological systems. However, to create a model that can provide relevant quantitative predictions, models must first be calibrated by recapitulating existing biological datasets from that system. Current calibration approaches may not be appropriate for complex biological models because: 1) many attempt to recapitulate only a single aspect of the experimental data (such as a median trend) or 2) Bayesian techniques require specification of parameter priors and likelihoods to experimental data that cannot always be confidently assigned. A new calibration protocol is needed to calibrate complex models when current approaches fall short. METHODS: Herein, we develop CaliPro, an iterative, model-agnostic calibration protocol that utilizes parameter density estimation to refine parameter space and calibrate to temporal biological datasets. An important aspect of CaliPro is the user-defined pass set definition, which specifies how the model might successfully recapitulate experimental data. We define the appropriate settings to use CaliPro. RESULTS: We illustrate the usefulness of CaliPro through four examples including predator-prey, infectious disease transmission, and immune response models. We show that CaliPro works well for both deterministic, continuous model structures as well as stochastic, discrete models and illustrate that CaliPro can work across diverse calibration goals. CONCLUSIONS: We present CaliPro, a new method for calibrating complex biological models to a range of experimental outcomes. In addition to expediting calibration, CaliPro may be useful in already calibrated parameter spaces to target and isolate specific model behavior for further analysis.

A multi-scale pipeline linking drug transcriptomics with pharmacokinetics predicts in vivo interactions of tuberculosis drugs.

Tuberculosis (TB) is the deadliest infectious disease worldwide. The design of new treatments for TB is hindered by the large number of candidate drugs, drug combinations, dosing choices, and complex pharmaco-kinetics/dynamics (PK/PD). Here we study the interplay of these factors in designing combination therapies by linking a machine-learning model, INDIGO-MTB, which predicts in vitro drug interactions using drug transcriptomics, with a multi-scale model of drug PK/PD and pathogen-immune interactions called GranSim. We calculate an in vivo drug interaction score (iDIS) from dynamics of drug diffusion, spatial distribution, and activity within lesions against various pathogen sub-populations. The iDIS of drug regimens evaluated against non-replicating bacteria significantly correlates with efficacy metrics from clinical trials. Our approach identifies mechanisms that can amplify synergistic or mitigate antagonistic drug interactions in vivo by modulating the relative distribution of drugs. Our mechanistic framework enables efficient evaluation of in vivo drug interactions and optimization of combination therapies.

Systems biology predicts that fibrosis in tuberculous granulomas may arise through macrophage-to-myofibroblast transformation.

Mycobacterium tuberculosis (Mtb) infection causes tuberculosis (TB), a disease characterized by development of granulomas. Granulomas consist of activated immune cells that cluster together to limit bacterial growth and restrict dissemination. Control of the TB epidemic has been limited by lengthy drug regimens, antibiotic resistance, and lack of a robustly efficacious vaccine. Fibrosis commonly occurs during treatment and is associated with both positive and negative disease outcomes in TB but little is known about the processes that initiate fibrosis in granulomas. Human and nonhuman primate granulomas undergoing fibrosis can have spindle-shaped macrophages with fibroblast-like morphologies suggesting a relationship between macrophages, fibroblasts, and granuloma fibrosis. This relationship has been difficult to investigate because of the limited availability of human pathology samples, the time scale involved in human TB, and overlap between fibroblast and myeloid cell markers in tissues. To better understand the origins of fibrosis in TB, we used a computational model of TB granuloma biology to identify factors that drive fibrosis over the course of local disease progression. We validated the model with granulomas from nonhuman primates to delineate myeloid cells and lung-resident fibroblasts. Our results suggest that peripheral granuloma fibrosis, which is commonly observed, can arise through macrophage-to-myofibroblast transformation (MMT). Further, we hypothesize that MMT is induced in M1 macrophages through a sequential combination of inflammatory and anti-inflammatory signaling in granuloma macrophages. We predict that MMT may be a mechanism underlying granuloma-associated fibrosis and warrants further investigation into myeloid cells as drivers of fibrotic disease.

Predicting the second wave of COVID-19 in Washtenaw County, MI.

The COVID-19 pandemic has highlighted the patchwork nature of disease epidemics, with infection spread dynamics varying wildly across countries and across states within the US. To explore this issue, we study and predict the spread of COVID-19 in Washtenaw County, MI, which is home to University of Michigan and Eastern Michigan University, and in close proximity to Detroit, MI, a major epicenter of the epidemic in Michigan. We apply a discrete and stochastic network-based modeling framework allowing us to track every individual in the county. In this framework, we construct contact networks based on synthetic population datasets specific for Washtenaw County that are derived from US Census datasets. We assign individuals to households, workplaces, schools, and group quarters (such as prisons or long term care facilities). In addition, we assign casual contacts to each individual at random. Using this framework, we explicitly simulate Michigan-specific government-mandated workplace and school closures as well as social distancing measures. We perform sensitivity analyses to identify key model parameters and mechanisms contributing to the observed disease burden in the three months following the first observed cases of COVID-19 in Michigan. We then consider several scenarios for relaxing restrictions and reopening workplaces to predict what actions would be most prudent. In particular, we consider the effects of 1) different timings for reopening, and 2) different levels of workplace vs. casual contact re-engagement. We find that delaying reopening does not reduce the magnitude of the second peak of cases, but only delays it. Reducing levels of casual contact, on the other hand, both delays and lowers the second peak. Through simulations and sensitivity analyses, we explore mechanisms driving the magnitude and timing of a second wave of infections upon re-opening. We find that the most significant factors are workplace and casual contacts and protective measures taken by infected individuals who have sought care. This model can be adapted to other US counties using synthetic population databases and data specific to those regions.