US 12,330,047 B2
System for generating simulated animal data and models
Vivek Khare, Cupertino, CA (US); Mark Gorski, Royal Oak, MI (US); Stanley Mimoto, Bethel Island, CA (US); and Anuroop Yadav, Livingston, NJ (US)
Assigned to Sports Data Labs, Inc., Royal Oak, MI (US)
Appl. No. 17/251,092
Filed by SPORTS DATA LABS, INC., Royal Oak, MI (US)
PCT Filed Sep. 8, 2020, PCT No. PCT/US2020/049678
§ 371(c)(1), (2) Date Dec. 10, 2020,
PCT Pub. No. WO2021/046519, PCT Pub. Date Mar. 11, 2021.
Claims priority of provisional application 63/027,491, filed on May 20, 2020.
Claims priority of provisional application 62/897,064, filed on Sep. 6, 2019.
Prior Publication US 2022/0323855 A1, Oct. 13, 2022
Int. Cl. A63F 13/212 (2014.01); A63F 13/217 (2014.01); A63F 13/65 (2014.01); A63F 13/90 (2014.01)
CPC A63F 13/212 (2014.09) [A63F 13/217 (2014.09); A63F 13/65 (2014.09); A63F 13/90 (2014.09); A63F 2300/8094 (2013.01)] 62 Claims
OG exemplary drawing
 
1. A method comprising:
receiving, by a computing device, one or more sets of real animal data at least partially obtained from one or more sensors that receive, store, and/or send information related to one or more targeted individuals, the one or more sensors including a biological sensor configured to measure electrical signals in or derived from a targeted subject's body, the one or more targeted individuals including a player in an athletic event;
transforming the electrical signals into one or more heart rate values and/or electrocardiogram (ECG) data;
generating, by the computing device, simulated animal data from at least a portion of real animal data or one or more derivatives thereof, wherein one or more parameters or variables of the one or more targeted individuals are modified, the simulated animal data including artificially-created data that shares at least one biological function with a human or another Animal; and
assessing fatigue for the player in the athletic event from the simulated animal data from a running total for an amount of time the player has an elevated heart rate, diastolic blood pressure, systolic blood pressure, perspiration rate and/or distance run wherein the simulated animal data is generated by a method using a recurrent neural network (RNN) comprising:
(a) receiving real animal data from multiple events, wherein training animal data readings are timestamped and recorded at predetermined time intervals;
(b) training an untrained recurrent neural network model using the real animal data to form a trained recurrent neural network model, wherein the training includes:
(i) using a sequence of observations as input;
(ii) performing the training over a plurality epochs using an error metric and an optimizer for updating network weights;
(c) generating simulated animal data by applying the trained neural recurrent network model, wherein the trained recurrent neural network model predicts physiological data including heart rate based on the real animal data;
(d) performing in-sample forecasting, wherein the simulated animal data is generated from a subset of the real animal data to predict future values based on known outcomes;
(e) performing out-of-sample forecasting, wherein the simulated animal data is generated from observations that were not part of the real animal data to predict future values based on previously unseen data; and
(f) modifying one or more parameters of simulated animal data to alter physiological, biomechanical, or environmental conditions for predictive analysis.