Back to 2018 IMAG Futures Agenda
IMAG MEETING NOTES: Breakout Session on MSM Applications and Clinical Translation
- Jay Humphrey (email@example.com)
- Bruce Lee (firstname.lastname@example.org)
- Michael Henson (email@example.com)
IMAG Points of Contact
- Tina M. Morrison (FDA/CDRH) (firstname.lastname@example.org)
- Orlando Lopez (NIH/NIDCR) [E] (email@example.com)
Action Items: Planned Deliverables From Breakout Session Discussions
- Three whitepaper archival publications will be prepared to address the following topics:
- Success stories of clinical translation where MSM had impact on clinical care
- Success stories of institutional infrastructure that fascilitates MSM clinical translation by enhanced communication between modelers-clinicians-engineers-entrepenuers-other relevant experts
- Speak to future opportunities and needs that promote MSM clinical translation by enhanced communication between modelers-clinicians-engineers-entrepenuers-other relevant experts
- PI leads listed above will lead efforts on manuscript preparation and seek contributions from IMAG members.
Day 1 - Notes
- focus on success stories below
- Patient specific applications of modeling (e.g., from imaging information)
- FDA approval was not needed initially - accelerated clinical translation - built good relationship with clinicians
- Primary barriers
- lack of animal models fuels/advocatest need for modeling
- access to data that is suited to support modeling efforts - "clinical study that supports modeling"
- Resources oriented linitations for large scale applications
- Facilitating agents
- Collaborations/build good relationship with clinicians-engineers-modelists-basic scientists-imagers-patients-others
- Imagers are applied mathmatetians interested/knowledgable about modeling
- Look for clinical driven needs to develop models - this is a needs poll not a technology push
- Focus on sociology efforts - not only on technology efforts - develop strategic plan about what clinicians will do/be in future (10-20yrs)
- Consider generational changes in landscape of medical education/practice
- Understand clinician's thought process and culture which is very different from that of scientists/technologists
- need terminology standarizaton between relevant stakeholders
- need respect and deep understanding between stakeholders
- Think about what clinicians need and also what they are necessarily asking for
- use of MSM in systematic way to predict evolution of disease and outcomes of interventions
- Identify target audience for upcoming publications from this group
- attend/publish/present to groups outside modeling field - get buyin and build awareness across multidisciplinary fields
- Medical training is a moving target - use general algorithm/protocol driven protocols
- Surgery training requires special training compared to most other medical practicioners
- Visualization/Presentation/Interpretation of Modeling information/outcomes - presented to clinicians should be translated on their languge - biologically and clinically relevant significance
Reinforced Ideas - take aways
- Reinforce relationships between stakeholders
- Buyin from stakeholders - include clinicians (and statisticians) in conversation early
- Opportunities of application of models to illusidation of disease and model disease progression and treatment
- Communicate better with clinical community
- Visualization/Presentation/Interpretation of Modeling information/outcomes in way clinician is used to seeing - cartoons matter
- How do we link current strategies to clinical complexity?
- Needs assessment - respect for different view points
- Understand FDA regulatory pathways associated with modeling technologies - understand options
- Medical practice in 2030 - different medical specialties have different needs and different technical backgrounds - Respect clinician and their differences of needs and backgrounds
- Target dual publications for in-silico experiments on clinical and modeling focused journals - standarization of formats to go with computaion bioscience with journals
- Remove preconceptions and unrealistics expectations - remove barriers of use for models - advocate for the usefulness of the modeling to get buyin
Day 2 - Notes
Where can we learn from? Resources-Alliances-Model Organizations/Centers
- Consider Standford's Biodesign Program in context of how to approach enhanced multidiciplinary interactions between stakeholders for needs identification and product translation/commercialization
- Coulter Centers
- Moffit Cancer Center - Mathematical Oncology - application of CM to guide oncologic treatment - CM combined uses of imaging, genomic, proteomic, radiation, etc...
- Bakken Center UMN - entreprenuership program modeled after Standford's Biodesign Program
- NSF - look for common denominators between modeling and simulation - develop core of standards
- Look at academic degree programs that offer modeling and simulation in curriculum
- Sloan Memorial Cancer Center - use CM (softwares) to enhance/aid clinical desicion making
- Arteries - company in California - FDA cleared applications AI/deep learning
- DOD - computational working group on human modeling - lethality, injury and imparement
- Undiagnosed diseases network - Use of modeling, similation, analytics to assess EHRs for identification of diseases and adequate treatment options
- Institute for Computational Medicine at JHU
- Radiation oncology field as a whole uses CM to guide treatment
- Cardiovascular applications - (Heartflow) and other applications
- Deep brain stimulation/neuro modulation - CUNY group and SKI group (Utah) - used to optimize electrode configuration for certain targetted brain regions
- Human Factors engineering groups
- Look for patient specific diseases
- Orthopedic applications - there are a number of FDA cleared devices that use/have use CM software components
- CM used to support development of MRI safe/conditional medical products cleared by FDA - next gen pace maker lead deign
- Guide craniotomies for epilepsie procedures
- population exposure assessment - Modeling across infections diseases - obesity - HIV - epidimiologic prediction
- Design of new materials - implants
- Guide design of clinical trials
Relevant FDA considerations on translating CM to patients/FDA approved products - (Contact Tina.Morrison@fda.hhs.gov for questions/comments)
- FDA-wide Modeling and Simulation working group was started in 2017, supported by the Office of the Chief Scientist
- as of Jan 2018, there are 200 FDA scientists as a part of this working group - see presentation
- Slides for FDA perspective on in silico medicine, with an example about the virtual patient framework is here.
- Additional information about the Avicenna Alliance and their working groups can be found here.
- Simulations that are used to support clinical decision making, whether for medical devices or pharmaceuticals will be regulated as "software as a medical device"
- learn more here about definitions and guidance
- Regarding use of simulation in the clinic, do we know if physicians are using the tools that are currenty available?
Provider led entity (PLE) - To get CMS approval for “appropriate use criteria” for PLE.
- JACR March 2018; vol 15#3PB-Special Issue
- Data Science: Big Data, Machine Learning and AI
- COMPUTATIONAL MEDICINE - "Computational Medicine: Translating Models to Clinical Care"
Raimond L. Winslow,1* Natalia Trayanova,2 Donald Geman,3 Michael I. Miller4
Because of the inherent complexity of coupled nonlinear biological systems, the development of computational models is necessary for achieving a quantitative understanding of their structure and function in health and disease. Statistical learning is applied to high-dimensional biomolecular data to create models that describe relationships between molecules and networks. Multiscale modeling links networks to cells, organs, and organ systems. Computational approaches are used to characterize anatomic shape and its variations in health and disease. In each case, the purposes of modeling are to capture all that we know about disease and to develop improved therapies tailored to the needs of individuals. We discuss advances in computational medicine, with specific examples in the fields of cancer, diabetes, cardiology, and neurology. Advances in translating these computational methods to the clinic are described, as well as challenges in applying models for improving patient health.
MSM Application WGs
MSM U01 suggestions for new MSM Application Challenges
- Influence of multi-scale modeling on different biomedical fields. Applications of the models upstream to identify likely failures of new treatments.
- Precision vs personalized medicine.
- How to best balance the simplicity that models should bring with the complexity of biology through a rational integration of biological data and hypothesis in multi-modeling frameworks.
- Models that facilitate the design and application of single or multiple treatments in adaptive therapy based clinical trials.
- A next challenge in cardiac metabolism/energetics is protein metabolism. Protein fluxes are large; the balance between proteolysis and protein synthesis fluctuates with activity. Athletic stresses require long recoveries. The pathway to protein synthesis begins with the signaling and control of mRNA production to supply the recipe for transcription. The influences of environment, organism activities, humoral and neural signals, and cellular state come together drive many cells simultaneously to transcription in a coordinated fashion: the nuclear (or mitochondrial) DNA response to these drivers are guided by the composite set of input signals from all of these levels. Defining these signals and their origins and connections throughout the body's systems is a hugely complex problem.
- Modeling complexity – that is, going beyond single steady state situations to model how cells adapt to changing internal and external conditions. Internal conditions may be due to inherent cycles or oscillations within metabolism or the cell cycle, for instance. Non-equilibrium, nonsteady state dynamics.
- Technology that allows us to identify putative drug targets or interventions that take into account that the system is complex and adaptive. Technology that allows us to also predict unintended cellular consequences that are manifest as side-effects of interventions –and how to mitigate these.
- o Metabolism is modeled using constraint-based approaches that only model steady states. Even so-called dynamic constraint-based approaches are simply a series of steady states strung together. There is no adaptivity in these approaches.
- o What happens when one perturbs a metabolic community?
- Physical understanding of natural selection. How much physical work/energy is required to replicate different cell types? The work/energy required to replicate is the physical quantity that determines the outcome of natural selection, and hence long-term adaptivity. This is especially important for understanding, predicting and controlling human microbiomes and cancers. But it is also important for brain and tissue repair, as well.
- Models of inter-cellular interactions, from microbiomes to the brain. How do cells turn material and energy gradients into information and knowledge? How do the principles of thermodynamics, control theory and dynamical systems lead to abstractions that become encoded as knowledge? How does perception lead to specific molecular interactions between specific cells/neurons and how do these specific interactions lead to self-awareness?
- We have categorized our cell-to-organ-to-organism-to-population work under item (11) from the 2009 IMAG report. It may be more appropriate to make a new challenge heading to span the full range with models grounded in experimental data that capture as much mechanistic understanding as possible at the various scales. The mechanism component, when grounded in observations, would aid linkage across scales by helping direct how smaller-scale models might fit into higher-level models either parametrically or spatially. Models would need to explicitly address the spatiotemporal nature of organ function, while also being tied to a functional/clinical readout that could differentiate healthy and diseased individuals (perhaps also allowing endo/phenotyping of the diseased population into subpopulations based on model-captured differences).
- Increased focus on process simulation/manufacturing
- We believe it will be important to expand SDO application to optical recording data where individual spikes are not recorded but rather a convolution of spiking processes and calcium dynamics produces the recorded signal events. This will support larger scale applications of the methods. What expertise are on your team (e.g. engineering, math, statistics, computer science, clinical, industry
- the integration of immune science into modeling efforts
- the integration of biomechanics with cellular systems modeling
- How to incorporate single cell sequencing and epigenetics regulation in systems modeling?
- Diversity of cell stimuli. We tend to focus on one or perhaps two aspects of the extracellular environment that drive cell behavior. How do we integrate experiments and modeling to tackle the greater complexity of cues that affect cell behavior/fate?
- Molecular and cellular heterogeneity in tissues. Tissues are composed of multiple cell types, and in many contexts the cells are in motion and changing their behavior as they differentiate. This presents an enormous yet exciting challenge for modeling.
- Improved annotation, data structures, and data capture for complex, multi-clinician medical interventions such as surgery, resuscitation therapy, anesthesiology.
- evidence based clinical data integrated into modeling
- How can systems modeling address drug resistance and recurrence in clinic studies?
- How to consider mutations in systems modeling to simulate clinical treatment?
- As a community, we need to address the issue of computational speed vis-à-vis patient variability. How can we implement patient-specific models effectively when the model takes days or weeks to run?