You are on page 1of 105
(12) INTERNATIONAL APPLICATION PUBLISH (09) Worid Intellect Property ‘Organization = Inston areas =F Se (43) International Publication Date = 13 February 2014 (13.02.2014) (10) International Publication Number WO 2014/025765 A2 WIPO!PCT (1) International Patent Classification: AGIN 1/18 (2006.01) (21) International Application Number: pew) o130s3 (22) International Filing Date 6 August 2013 (06.08.2013) (25) Filing Language: English (26) Publication Language: Degli (80) Priority Date: 51/679,939 6 Angust 2012 (06.08.2012) us (71) Applicant: UNIVERSITY OF MIAMI [USUS}; 1951 NW ‘is Avenue - Suite 110, Location Code: C234, Miami FL 33136(US), Inventors: MAHMOUDI, Babak; 1951 NW Tih Aveave Suite 120, Location Cod C234, Miami, FL (Us), SANC- HZ, Justin, C2, 16809 SW Sih Court, Miranat, FL 33027 (US). Agents: MATOS, Peter, A. ct al Malloy & Malloy, PL, ‘Attomeys For Applicant, 2800 Sv, Third Avenue, Mian, FL 3512905), m~ om (81) Designated States (unlevsotherwise indicoied, for every Kind of national protection available): AE, AG, AL, AM, AO, AT, AU, AZ, BA, BB, BG, BH, BN, BR, BW, BY, BZ, CA, CH, CL, CN, CO, CR, CU, CZ, DE, DK, DM DO, Dz, E 5, FI, GB, GD, GE, GH, GM, GT, HN, HR, HU, ID, 1, IN, IS, JP, KE, KG, KN, KP, KR KZ, LA, LC, LK, Li, LS, LT, LU, LY, MA, MD, ME MG, MK, MN, MW, MX, MY, MZ, NA, NG. NT, NO, NZ, (OM, PA, BE, PG, PH, PL, PT, QA, RO, RS, RU, RW, SA. SC, SD, SK, SL, SM, ST. SV. SY, TH, TH, TM. TN, TR, TF, TZ, UA, UG, US, UZ, VC, VN, ZA, ZN. a Designated States (unless etherwise Indico, for every ind of regional protection avaiable): ARIPO (BW, Gil GM, KE, TR. 1S, MW. MZ, NA, RW. SD, SL, SZ, TZ, UG, ZM, ZW), Purasisn (AM, AZ, BY, KG, RZ, RUTH ‘TM), Furopean (AL, AT, BE, BG, CH, CY, CZ, DE, DK, FE, PS, FI FR, GB, GR. HR, HU, IE 18, IT, LT, LU. LY, MC, MK, MTT, NL, NO, PL, PT, RO, RS, SE, SISK, SM, TR), OAPI (BF. DI. CF. CG, Ci, CM, GA, GN. GO. GW KM, ML, MR, NE, SN, TD, TG) Published: without international search report and to be republished spon receipt ofthat report Ruse 48.218) ow WO 2014/025765 A2 (Ga) Title SYSTEMS AND METHODS FOR ADAPTIVE NEURAL DECODING nuns (67) Abstract: This invention is directed to a system ad ‘method for adaptive newal desing of «user A plral- Ay of itersonnected processing units comprising seuNory ods, hidden nodes, a output nodes fons thee l= fer network. The seasory nods receive signals fromm & teural veclor and cvsputs gna to the hidlen modes, The hidden nodes are connected to the sensory nodes, and the vip esto the hidden nodes tough each of 8 symaptie’ conection Raving aa individual synaptic weight. The hidden ant ouput nodes are indvidaly ‘contigred 1o calculate a probability based an is synaptic ‘weighs, and canput a signal based on the probably ‘The out signal of the pur node having the highest rally is ransmited to the environment. feodback ‘inal is received y’a focthack module hased on the out Dt signal, and associat synapse weight (8) ae altered base on the foedhack signal 10 15 20 25 30 39 WO 2014025765 PCT/US2013/083772 scription SYSTEMS AND METHODS FOR ADAPTIVE NEURAL DECODING Statement Regarding Federally-Sponsored Research and Development This invention was made with U.S. government support under grant mumber N65001-10-C-2008 and W31P4Q-12-C-0200 awarded by the Defense Advanced Research Projects Agency (DARPA), The U.S. government may have certain rights in the invention. BACKGROUND OF THE INVENTION Field of the Invention ‘The present invention relates to adaptive neural decoding and terfaces between the brain and machines. More specifically, the invention concerns automatically adjusting neural mapping for adaptive learning of neural decoders and neuroprosthetics. Additionally, the present invention is alse directed to responsive neurorehabilitation through the use of adaptive neural decoders. Description of the Related Art Neural decoding involves the reconstruction of stimuli from information that is represented in the brain by networks neurons, in order to convert the impulses of the brain into ons that can be performed in the subject's environment. With neural decoding, thought can drive mechanical motion of objects in the environment, such as neuroprosthetic dev decoder is a device made up of a plurality of synthetic neurons or a ces. A neural nodes that process impulses and convey information to other nodes further downstream by “firing” an action potential. Ideally, the pattern of synthetic neurons or nodes models the decision-making process in the brain, in order to convert neural impulses to commands for action to be taken in an environment. Neural decoding can be particularly useful in medical cases in waich a person’s control of a limb or extremity has been reduced or is gone entirely. For instance, neural decoding can assist in rehabilitation of arm, hand, leg, or foot movement, such 10 15 20 25 30 39 WO 2014025765 PCT/US2013/083772 as following injury or impairment due to a medical condition such as Parkinsons, full or pai al paralysis, and muscle wasting diseases or conditions. It can also be used for neuroprosthetic control, such as to replace an amputated arm, leg, hand, or foot. However, many challenges still remain. For example, central nervous system (CNS) injuries, such as spinal cord injuries, impair the ability of the brain to communicate with and direct the body. Physical therapy is often used, but it does not fully engage the CNS, only affecting the brain indirectly and not quantitatively. Therefore, the connection between the brain and body is not fully restored, which is needed for successful rehabilitation. In addition, the way the brain produces intent adapts and changes over time, a phenomenon called neural plasticity. It is a complicated and constantly ongoing process, and is influenced by numerous factors. Being able to account for and adapt to this ever-changing neural landscape is critical for successful neural decoding, but as yet remains only rudimentarily achieved. For instance, machines can “learn” to adapt to changes or perturbations, but the various approaches to machine learning all fall short of ideal. There are currently three main machine learning paradigms: supervised, unsupervised, and reinforcement learning methods. In supervised learning approaches, the controllé neural impulses to the machine needs a detailed copy of the desired response to compute a low-level feedback for adaptation. For example, in the case of reaching tasks, the desired response or converter would be the trajectory of arm movements or the location of the target in the environment. Taere have been attempts to “infer” a desired response so that supervised learning approaches can be used, however these approaches do not exactly match the user’s intent and may not be feasible in unstructured environments such as those encountered during daily living, Moreover, these static neural decoding algorithms assume stationary input/output relationships, limiting their usefulness since they cannot easily adapt to perturbations in the input space.