# US 2017/0286569 A1 · System and method of modeling irregularly sampled temporal data using kalman filters

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2017/0286569 A1
- Document ID: US20170286569A1
- Status: Published application
- Kind: A1
- Application number: 15/083,466
- Filed: 2016-03-29
- Published: 2017-10-05
- Earliest claimed: 2016-03-29
- Assignees: Xerox Corporation
- Applicants: Xerox Corporation
- Inventors: Abhishek Sengupta; Prathosh Aragulla Prasad; Satya Narayan Shukla; Vaibhav Rajan; Katerina Sinclair; Stephen Fullerton
- Web page: https://patentgrep.com/patent/US20170286569A1

## CPC

- G06F 30/20: Design optimisation, verification or simulation (optimisation, verification or simulation of circuit designs G06F30/30)
- G06F 17/18: for evaluating statistical data {, e.g. average values, frequency distributions, probability functions, regression analysis (forecasting specially adapted for a specific administrative, business or logistic context G06Q10/04)}
- G06N 5/04: Inference or reasoning models
- G06N 20/00: Machine learning
- G06F 2111/08: Probabilistic or stochastic CAD
- G06F 2218/12: Classification; Matching
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

Systems and methods of modeling irregularly sampled time series signals with unknown temporal dynamics are disclosed wherein a temporal difference variable (TDV) is introduced to model irregular time differences between subsequent measurements. A hierarchical model is designed comprising two linear dynamical systems that model the effects of evolving TDV on temporal observations. All the parameters of the model, including the temporal dynamics are statistically estimated using historical data.

## Claims

### Claim 1

A method for data forecasting using machine learning said method comprising:
collecting a dataset with an irregular temporal sampling rate;
defining a state equation according to said collected dataset with said irregular temporal sampling rate;
defining a temporal difference variable to account for said irregular temporal sampling rate;
incorporating said temporal difference variable into said state equation in order to form a linear dynamic system model; and
forecasting future expected data in said dataset according to said linear dynamic system model.

### Claim 2 (depends on 1)

The method of claim 1 further comprising estimating at least one parameter associated with said linear dynamic system model according to historical data.

### Claim 3 (depends on 2)

The method of claim 2 said at least one parameter comprises at least one parameter indicative of temporal dynamics.

### Claim 4 (depends on 1)

The method of claim 1 wherein said linear dynamic system model comprises a KF3 model.

### Claim 5 (depends on 4)

The method of claim 4 wherein said temporal difference variable associated with said KF3 model is derived from said state equation given by:
{tilde over (Λ)}t,t−1=Ã{tilde over (Λ)}t−1,t−2+{tilde over (ε)}t
Δt,t−1={tilde over (C)}{tilde over (Λ)}t,t−1+{tilde over (δ)}t.

### Claim 6 (depends on 5)

The method of claim 5 wherein said temporal difference variable is a temporal evolving parameter of said KF3 model.

### Claim 7 (depends on 6)

The method of claim 6 wherein said dataset comprises data associated with one of:
patient vital signs;
electronic medical records; and
chemical processing.

### Claim 8

A method for identifying a patient at risk of sepsis said method comprising:
collecting a dataset of at least one vital sign of said patient;
defining a state equation according to said collected dataset;
defining a temporal difference variable;
incorporating said temporal difference variable into said state equation in order to form a linear dynamic system model; and
forecasting the onset of sepsis in said patient according to said linear dynamic system model.

### Claim 9 (depends on 8)

The method of claim 8 wherein said dataset of at least one vital sign of said patient comprises a dataset with an irregular temporal sampling rate.

### Claim 10 (depends on 9)

The method of claim 9 further comprising estimating at least one parameter associated with said linear dynamic system model according to historical vital sign data.

### Claim 11 (depends on 10)

The method of claim 10 wherein said at least one parameter comprises at least one parameter indicative of temporal dynamics.

### Claim 12 (depends on 10)

The method of claim 10 wherein said linear dynamic system model comprises a KF3 model.

### Claim 13 (depends on 12)

The method of claim 12 wherein said temporal difference variable is a temporal evolving parameter of said KF3 model.

### Claim 14 (depends on 13)

The method of claim 13 wherein said dataset of at least one vital sign of said patient comprises at least one of:
body temperature;
heart rate;
respiratory rate; and
white blood cell count.

### Claim 15

A clinical decision support system comprising:
at least one dataset collection module configured to collect data;
a processor-readable medium storing computer code representing instructions to cause a process for identifying content relevance, said computer code comprising code to:
define a state equation according to said collected dataset with said irregular temporal sampling rate;
define a temporal difference variable to account for said irregular temporal sampling rate;
incorporate said temporal difference variable into said state equation in order to form a linear dynamic system model; and
forecast future expected data in said dataset according to said linear dynamic system model.

### Claim 16 (depends on 15)

The clinical decision support system of claim 15 further comprising computer code configured for:
estimating at least one parameter associated with said linear dynamic system model according to historical data.

### Claim 17 (depends on 16)

The clinical decision support system of claim 16 wherein said at least one parameter comprises at least one parameter indicative of temporal dynamics.

### Claim 18 (depends on 15)

The clinical decision support system of claim 15 wherein said linear dynamic system model comprises a KF3 model.

### Claim 19 (depends on 18)

The clinical decision support system of claim 18 wherein said temporal difference variable associated with said KF3 model is derived from said state equation given by:
{tilde over (Λ)}t,t−1=Ã{tilde over (Λ)}t−1,t−2+{tilde over (ε)}t
Δt,t−1={tilde over (C)}{tilde over (Λ)}t,t−1+{tilde over (δ)}t.

### Claim 20 (depends on 19)

The clinical decision support system of claim 19 wherein said temporal difference variable is a temporal evolving parameter of said KF3 model.
