Investigators
Zachary Danziger; Elie Alhajjar; Deniz Erdogmus; Giovanna Guidoboni; John Yin
Contact info (email)
zdanzige@fiu.edu
1. Define context(s)
identify/explore new therapies
reveal new biological insights
Current Conformance Level / Target Conformance Level
TBD
Primary goal of the model/tool/database
The goal of the model is to demonstrate a framework for linking local mechanistic models together via machine learning to predict organ level function and response to neuromodulation in the urinary bladder. The spatial extent encompasses all spinal level tissues in the lower urinary tract and includes updates on the order of seconds, integrated across minutes.
Biological domain of the model
Urinary tract - organ level
Structure(s) of interest in the model
Bladder, urethra, reflex voiding
Spatial scales included in the model
Organ level
Time scales included in the model
seconds to hours
Other uses for the model (optional)
Lower urinary tract (LUT) disorders are associated with incontinence, bladders that are overactive or obstructed, and other symptoms; they are highly prevalent among adults worldwide. The treatment of such disorders by peripheral neuromodulation holds significant promise, but the lack of predictive models has forced researchers to explore the vast space of nerve targets, electrode designs, and stimulation strategies in animals by trial and error. Motivated by eventual human clinical needs, our long-term goal is to advance models to predict the effects of neuromodulation on lower urinary tract function.
To take a first step toward our goal, we propose to develop a new modeling framework that integrates biophysics through machine learning, emulating the entire LUT organ system. We will unite multiple mechanistic models, where each model accounts for a component function of the organ system; appropriate coupling of function and dynamics between component models will be learned by a deep recurrent neural network.
Critical to our long-term success will be for our team to: (i) adopt a consistent modeling and simulation terminology, (ii) develop guidelines and procedures for credible practice, (iii) demonstrate workflows for credibility assessment, and (iv) promote credible practice of modeling and simulation. Toward these ends, we embrace the guidelines that have been developed by the Committee on Credible Practice of Modeling & Simulation in Healthcare, and we look forward to actively participating in their evolution, providing feedback on best practices and lessons learned as we pursue organ system modeling and simulation.
2. Data for building and validating the model
| Data for building the model | Published? | Private? | How is credibility checked? | Current Conformance Level / Target Conformance Level |
|---|---|---|---|---|
| in vitro (primary cells cell, lines, etc.) | ||||
| ex vivo (excised tissues) | yes | citations | ||
| in vivo pre-clinical (lower-level organism or small animal) | yes | SPARC DRC on completion | ||
| in vivo pre-clinical (large animal) | ||||
| Human subjects/clinical | yes | |||
| Other: ________________________ | Since our organ system project will be based on an integration of existing (published) mechanistic models, the data for building and validation are the existing data that were used to develop the published models as well as data that we will collect. In most cases, these data are in vivo and pre-clinical, based on small animal studies, and published. Our first major task has focused on identifying and vetting component mechanistic models from the literature, which has enabled the identification of preliminary criteria for model exclusion and inclusion. In short, we will be performing a “ten rules” assessment as a step toward deciding which component models are most credible for inclusion as we build our system model. Currently, criteria for model exclusion are models that entail: (i) complex implicit solution schemes, (ii) spatial and/or finite element solution schemes, and (iii) multi-parameter biomechanical material descriptions that lack independent measures or estimates of parameters. Inclusion criteria favor submodels based on mass (fluid) conservation principles in bladder and urethral volumes and flows. Discussions are ongoing among team members who have expertise in mechanistic modeling; we are exploring to what extent inclusion criteria should be favored for semi-empirical models based on direct measurements of parameters such as bladder pressure, or empirically-derived parameters, such as time-derivatives of bladder pressure, which may drive afferent signaling. |
| Data for validating the model | Published? | Private? | How is credibility checked? | Current Conformance Level / Target Conformance Level |
|---|---|---|---|---|
| in vitro (primary cells cell, lines, etc.) | ||||
| ex vivo (excised tissues) | ||||
| in vivo pre-clinical (lower-level organism or small animal) | Comparisons with prior published work to check data validity. | |||
| in vivo pre-clinical (large animal) | ||||
| Human subjects/clinical | ||||
| Other: ________________________ | Although most of the data for building our organ system model will be based on published data used to build its published component models, we anticipate that new data will also be collected in P.I. Danzinger’s lab; these new data will be used to build and validate the organ system model. Like the published data used to develop the component models (above), the new data will be in vivo and pre-clinical, based on small animal studies, and initially private. The credibility of the new data will be checked internally and by the broader community through work with collaborators, informal and conference presentations and during submission for publication, by anonymous peer review. |
3. Validate within context(s)
| Who does it? | When does it happen? | How is it done? | Current Conformance Level / Target Conformance Level | |
|---|---|---|---|---|
| Verification | Participating labs. | During and post development. | ||
| Validation | ||||
| Uncertainty quantification | ||||
| Sensitivity analysis | ||||
| Other:__________ | ||||
| Additional Comments | Several components will contribute to the credibility of our organ system model: verification, validation, uncertainty quantification, and sensitivity analysis. These will be performed by the labs that are advancing the organ system model, and this work will be concurrent with and following development of the model. The current two-year project is a first step toward what we envision as a quite comprehensive eventual organ system model, so the broad activities associated with validation will be an ongoing process. |
4. Limitations
| Disclaimer statement (explain key limitations) | Who needs to know about this disclaimer? | How is this disclaimer shared with that audience? | Current Conformance Level / Target Conformance Level |
|---|---|---|---|
| Connections between components may not be fully mechanistic. | All users. | Listed prominently in all publications and documentation. | |
| Our proposed organ system model will be based in part, but not entirely, on component mechanistic models. Other features of the model will be the recurrent neural network that learns how to orchestrate the interactions between the component models. Thus, a key limitation of the organ system model will be that connections between components may not be fully mechanistic. All users of the organ system model will need to know about this limitation; a disclaimer will be listed prominently in all presentations, publications and model documentations. | |||
5. Version control
| Current Conformance Level / Target Conformance Level |
|---|
| TBD |
| Naming Conventions? | Repository? | Code Review? | |
|---|---|---|---|
| individual modeler | SimTK is a free project-hosting platform for the biomedical computation community that enables sharing of software, data, and models, tracks the impact of shared resources, provides infrastructure to support and grow a community around a project, and connects users and their projects to a broad cross-section of researchers working at the intersection of biology, medicine, and computation. | We anticipate that all versions of the organ system model, associated data and component models, and documentation will be managed using the subversion repository provided through SimTK (simtk.org). | |
| within the lab | |||
| collaborators |
6. Documentation
| Current Conformance Level / Target Conformance Level | |
|---|---|
| Code commented? | |
| Scope and intended use described? | All publications and technical reports will be published in accordance with NIH guidelines on Pub-Med indexing and open access. The full UM and associated code (e.g., network training and system simulation) generated by this project, along with documentation and relevant pointers to the LUT knowledge inventory, will be uploaded to o2S2PARC with the logistical support of the SIM-CORE team. |
| User’s guide? | |
| Developer’s guide? |
7. Dissemination
| Current Conformance Level / Target Conformance Level |
|---|
| The model, data used to create the simulations, documentation, and publications will be freely available for download from the o2S2PARC and SimTK project sites. The user community will be encouraged to provide corrections, edits and refinements and share them. Peer-reviewed articles, conference presentations and technical memos will also be regularly produced and disseminated among o2S2PARC stakeholders and the broader biomedical computation community. |
| Target Audience(s): | “Inner circle” | Scientific community | Public |
|---|---|---|---|
| Simulations | TBD | O2SPARC | Publications |
| Models | |||
| Software | |||
| Results | |||
| Implications of results |
8. Independent reviews
| Current Conformance Level / Target Conformance Level |
|---|
| TBD |
| Reviewer(s) name & affiliation: | Prior to making the model publically available, it will be submitted for independent peer review in conjunction with publication. |
|---|---|
| When was review performed? | |
| How was review performed and outcomes of the review? |
9. Test competing implementations
| Current Conformance Level / Target Conformance Level |
|---|
| TBD |
| Yes or No (briefly summarize) | |
|---|---|
| Were competing implementations tested? | The testing of competing implementations will be an ongoing process in the development of the organ level model. It is conceivable that the performance of component mechanistic models as independently run models will differ from their performance in the context of the full organ level model, necessitating multiple tests of competing implementations. |
| Did this lead to model refinement or improvement? |
10. Conform to standards
| Current Conformance Level / Target Conformance Level |
|---|
| TBD |
| Yes or No (briefly summarize) | |
|---|---|
| Are there operating procedures, guidelines, or standards for this type of multiscale modeling? | For the handling of animals, data collection, processing, and reporting methods, we will conform to practices generally accepted by the biomechanics community. For the organ level modeling that we envision, operating procedures, guidelines, or standards have yet to be defined. We look forward to contributing our experience and ideas toward defining such standards as we develop our integrative modeling approach. |
| How do your modeling efforts conform? |
11. (optional) Additional information to support items 1-10
This project is in the proposal stage, and as such model components have not yet been produced for dissemination or use. We will update this page as the project is developed.
Domain of Use: Translational research - The BLADDER model is being developed as a research tool to study lower urinary tract function; clinical applications are beyond the scope of the model.
Use Capacity: By integrating data and component-models, the intent of the BLADDER model is to: (1) elucidate mechanisms of organ-level function, and (2) enable testing of organ control strategies by neuromodulation.
Strength of influence: The aim of the BLADDER model is to identify and account for gaps in our mechanistic understanding of bladder function and generate hypotheses for further investigations. The model is not being developed to predict which individuals will experience bladder dysfunction.