US 11,817,193 B1
Systems and methods for autoregressive recurrent neural networks for identifying actionable vital alerts
Ashwyn Sharma, Seattle, WA (US)
Assigned to CADENCE SOLUTIONS, INC., New York, NY (US)
Filed by CADENCE SOLUTIONS, INC., New York, NY (US)
Filed on Feb. 17, 2023, as Appl. No. 18/171,078.
Int. Cl. G06N 3/04 (2023.01); G16H 10/60 (2018.01); G06N 3/0442 (2023.01)
CPC G16H 10/60 (2018.01) [G06N 3/0442 (2023.01)] 17 Claims
OG exemplary drawing
 
1. A system comprising:
a server comprising one or more processors; and
a non-transitory memory, in communication with the server, storing instructions that when executed by the one or more processors, cause the one or more processors to implement a method comprising:
receiving one or more first patient embeddings as input for a first autoregressive recurrent neural network including a first internal state;
generating, by the first autoregressive recurrent neural network, a first output, wherein the first output includes a probability of triggering a first actionable alert associated with patient vitals data;
receiving one or more second patient embeddings, the first internal state, and first output, as input for a second autoregressive recurrent neural network including a second internal state, wherein the second autoregressive recurrent neural network is configured to receive the second internal state and a third future internal state as input;
generating, by the second autoregressive recurrent neural network, a second output, wherein the second output includes a probability of triggering a second actionable alert associated with the patient vitals data; and
identifying one or more dependencies associated with the first internal state and the second internal state.