• Name

Reorganization and Plasticity to Accelerate Injury Recovery 

• What is the item

April 23, 2009, DARPA's DSO began soliciting for " DARPA seeks new methods for analysis and decoding of neural signals in order to understand how neural-based sensory stimulation could be applied to accelerate recovery from brain injury. Ultimately, it is desired to develop models of neural codes and temporal patterns that can provide an ability to interpret and predict changes in neural organization through plasticity at multiple scales of measurement." http://www.darpa.mil/dso/solicitations/baa09-27.htm

• What Horizon is it on

Sixth Horizon 

• Explanation of the item 

According to DARPA's DSO:

Major advances in neuroscience over the past decade have led to a number of theories and applications for the control of systems, biological or mechanical, through neural signals.  While many of these applications access anatomically relevant regions for appropriate signals, few of them address the brain as a distributed computational network made up of systems that act in parallel.  In addition, the knowledge of how the brain reorganizes these networks over time and through task acquisition is limited.   Further, the most enabling processes for extracting neural signals requires interfaces at the neuron level, which may present a number of biological problems over chronic use.

DARPA is soliciting research proposals that demonstrate innovative approaches to analyzing brain activity in order to develop more advanced bio-computational models of how the brain organizes and operates.  Such approaches should examine relevant structures at numerous scales of activity and across anatomically distributed regions during performance of a complex task.  In addition to analyzing electrical and non-electrical signals extracted from the brain, approaches should evaluate how the brain analyzes external stimuli to the biological sensory networks, including the eyes, ears, nose, and others.

This BAA is novel in that it seeks a multi-disciplined perspective on modeling brain activity, across fields such as neurobiology and network engineering, to take into account the processes of plasticity by which the brain reorganizes and optimizes performance. Through understanding the principles that allow networks in different anatomic regions to coordinate and communicate in order to perform a task, we seek to understand the means through which the brain enables improved performance over time. Further, by evaluating brain activity at several scales simultaneously (EEG, local field potential, single neuron, neurotransmitter, and corresponding scales for other transmission means), investigators may be able to determine which properties generated at the single neuron level can be correlated and predicted in a meaningful manner at grosser levels of measurement, such as EEG.

The ultimate goal of the program is the ability to create a biologically accurate computational model of relevant structures and networks in the brain, generalizable to the whole brain.  The model should accurately depict brain activity during task learning and in response to sensory stimuli.  Investigators must verify the model can predict permutations of brain activity in response to novel stimuli and new task acquisition in an experimental setting.  Researchers should show how the model can compensate for injury by resolving new brain patterns based on the dynamic organization of the brain.

Technical Area One:  Creation of an in silico, bio-computationally accurate model based on collection of multi-region, multi-scale neural activity in a non-human primate performing a complex dexterous task

Performers should construct an experimental system to collect neural activity in a non-human primate performing a complex dexterous task.  In addition to collecting data through multiple methodologies, the task should be constructed in such a manner as to allow for analysis of performance metrics, such as, time to target or targets, assessment of accuracy, control of simultaneous degrees of freedom, and learning time for task acquisition.  The performer must create an in silico bio computational model that accurately depicts experimental data.  The model should be capable of resolving algorithms that depict perturbed neural pathways and how plasticity might be used to determine an alternate path to task accomplishment based on observations of neurons early in the task performance path. It is expected that the bio computational model employ appropriately sophisticated mathematical analysis to capture the complexity of parallel processing, inter-neural activity, and plastic adaptation within the brain.  

Technical Area Two: Stimulation of non-human primates in order to evoke a biomimetic response

In addition to extraction and evaluation of neural signals, investigators will be expected to determine how the brain encodes the surrounding environment and transfers sensory information between networks of neurons. Investigators should be able to demonstrate the ability to stimulate relevant regions of the brain in such a manner that will evoke a response in the primate similar to that evoked through natural interaction with their surrounding environment.  Responses will be evaluated both through the subject’s ability to appropriately respond to the task stimuli vis-á-vis task performance as well as through observations of neural behavior compared to baseline performance in the absence of stimuli.  Ideally, investigators will be able to demonstrate ability of a non-human-primate to complete the task outlined in technical area one without the use of traditional sensory inputs. 

Technical Area Three: Perturbation of the system to mimic the injured state

Investigators should validate the ability of their model to address and enable recovery from injury by perturbing their system so as to mimic or simulate an injured brain. Approaches could include, but are not limited to, reduction in the amount and type of data used to accomplish the goals of Technical Area One and Two or through demonstration of the ability to complete the goals of Technical Area One or Two with an impaired primate.  Performance of the complex task after perturbation should be relatively stable as compared with performance prior to perturbation.  Additionally, performers should characterize system degradation as a function of “injury” severity and location as well as any resulting plastic responses in either the in silico model or in the nature and type of resulting outputs recorded from the primate brain.

Program Phases and Milestones

The REPAIR Program is a two-phase program corresponding to completion of investigations in Technical Areas One through Three above.  Note that preference will be given to efforts that propose to satisfy the metrics in less time or will provide study designs allowing for greater statistical significance under abbreviated research timelines. The multi-disciplined capabilities of proposer teams will be a consideration in the evaluation of proposals.  

The following information on the DARPA Phase I and II portions of the work plan are offered as guidelines. Proposers may use alternative milestones but they should be aware that deviation from those described below will reduce the likelihood of proposal acceptance. 

REPAIR Phase I Milestones

• In silico bio computational modeling of natural neural control of a complex dexterous task by a non-human primate
• Modeled perturbation of neural pathways to predict routes to task accomplishment in the absence of primary stimuli-response path used during initial task learning
• Measurable improvement on speed and accuracy in task completion through the neural interface by a minimum of a factor of two over a three month period
• Increase in control of simultaneous degrees of freedom by a factor of three over a three-month period
• Demonstration of ability to complete a dexterous task through the use of a traditional interface through generated sensory stimuli (see Technical Area Two)

REPAIR Phase II Milestones

·         Neural control of a dexterous task by a non-human primate in the absence of natural sensory stimuli using information generated using artificial sensory input (see Technical Area Two)

·         Validation of the model through successful task completion using seventy percent or less of neurally generated data and maintaining ninety-five percent of speed and accuracy comparable to prior trials

·         Demonstration of task performance through the use of neural signals without the use of neural spike recordings

http://www.darpa.mil/dso/solicitations/baa09-27.htm 

• Photos (if available)

• Issues

• Sources

http://www.darpa.mil/dso/solicitations/baa09-27.htm

This is a stub article.  Please make it better by adding to it!