# US 2022/0044809 A1 · Systems and methods for using deep learning to generate acuity scores for critically ill or injured patients

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

- Number: US 2022/0044809 A1
- Document ID: US20220044809A1
- Status: Published application
- Kind: A1
- Application number: 17/309,975
- Filed: 2020-02-21
- Published: 2022-02-10
- Earliest claimed: 2019-02-22
- Assignees: University of Florida Research Foundation, Inc.
- Applicants: University of Florida Research Foundation, Incorporated
- Inventors: Azra Bihorac; Tyler J. Loftus; Tezcan Ozrazgat Baslanti; Parisa Rashidi; Benjamin P. Shickel
- Web page: https://patentgrep.com/patent/US20220044809A1

## CPC

- G16H 50/20: for computer-aided diagnosis, e.g. based on medical expert systems
- G06N 3/044: Recurrent networks, e.g. Hopfield networks
- G06N 3/0442: characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
- G06N 3/09: Supervised learning
- G16H 10/60: for patient-specific data, e.g. for electronic patient records
- G16H 50/30: for calculating health indices; for individual health risk assessment
- G16H 50/50: for simulation or modelling of medical disorders
- G06N 3/048: Activation functions
- G06N 3/084: Backpropagation, e.g. using gradient descent
- G06N 5/01: Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
- G06N 7/01: Probabilistic graphical models, e.g. probabilistic networks
- G06N 20/20: Ensemble learning
- G16H: HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

Methods, apparatus, systems, and computer program products for providing patient predictions are provided in various embodiments. Responsive to receiving an indication of initiation of a patient interaction, a model for the patient is initiated by an assessment computing entity. The model has been trained using machine learning and the model is configured to generate a prediction for the patient. The prediction comprises at least one of an acuity score or a mortality prediction. Responsive to identifying a prediction trigger, the assessment computing entity updates the model for the patient based at least in part on medical data corresponding to the patient. The assessment computing entity generates the prediction using the updated deep learning model. The assessment computing entity provides at least a portion of the prediction such that the at least a portion of the prediction may be used to update an electronic health record corresponding to the patient and/or provided to a clinician for review.

## Claims

### Claim 1

A method for providing a patient prediction, the method comprising:
responsive to receiving an indication of initiation of a medical event for a patient, initiating a deep learning model for the patient, wherein the deep learning model (a) comprises a modified recurrent neural network (RNN) with gated recurrent units (GRUs), (b) is executed by an assessment computing entity, (c) has been trained using machine learning, and (d) is configured to generate a prediction for the patient comprising at least one of an acuity score or a mortality prediction;
responsive to identifying a prediction trigger, automatically updating, by the assessment computing entity, the deep learning model for the patient based at least in part on medical data (a) corresponding to the patient and (b) captured during the medical event as part of one or more time series of medical data;
automatically generating, by the assessment computing entity, the prediction using the updated deep learning model; and
automatically providing, by the assessment computing entity, at least a portion of the prediction such that the at least a portion of the prediction may be used to update an electronic health record corresponding to the patient and/or provided to a clinician for review.

### Claim 2 (depends on 1)

The method of claim 1, wherein the acuity score is a sequential organ failure assessment (SOFA) score.

### Claim 3 (depends on 1)

The method of claim 1, wherein the deep learning model comprises a self-attention mechanism.

### Claim 4 (depends on 3)

The method of claim 3, wherein the self-attention mechanism determines a weight for each time step of the one or more time series that indicates the time steps influence on the prediction.

### Claim 5 (depends on 1)

The method of claim 1, wherein the prediction comprises at least one self-attention parameter and/or distribution.

### Claim 6 (depends on 5)

The method of claim 5, wherein the at least one self-attention parameter and/or distribution is flagged as to whether or not the self-attention parameter and/or distribution represents a therapeutic target.

### Claim 7 (depends on 1)

The method of claim 1, wherein the at least a portion of the prediction is provided such that the at least a portion of the prediction is received by a user computing entity configured to provide the at least a portion of the prediction to the clinician for review.

### Claim 8 (depends on 1)

The method of claim 1, wherein the at least a portion of the prediction is provided such that the at least a portion of the prediction is received by a records computing entity configured to update an electronic health record corresponding to the patient based at least in part on the at least a portion of the prediction.

### Claim 9 (depends on 1)

The method of claim 1, wherein the prediction trigger is the determination that a period of time corresponding to a periodicity with which predictions are to be provided has passed.

### Claim 10 (depends on 1)

The method of claim 1, wherein the deep learning model is configured to provide a prediction periodically.

### Claim 11 (depends on 10)

The method of claim 10, wherein the deep learning model is configured to provide a prediction every hour.

### Claim 12 (depends on 10)

The method of claim 10, wherein the deep learning model is configured to provide a prediction every minute.

### Claim 13 (depends on 1)

The method of claim 1, wherein the medical event is at least one of an intensive care unit (ICU) stay, a medical procedure, or a surgical procedure.

### Claim 14 (depends on 1)

The method of claim 1, further comprising receiving medical data provided via user interaction with one or more user computing entities and/or data collection via one or more sensors.

### Claim 15 (depends on 1)

The method of claim 1, further comprising receiving medical data from a patient electronic health record (EHR).

### Claim 16 (depends on 1)

The method of claim 1, wherein the at least one of the acuity score or the mortality prediction is a (near) real-time acuity score or mortality prediction for the patient.

### Claim 17 (depends on 1)

The method of claim 1, wherein the deep learning model comprises a static module for receiving and processing background information corresponding to the patient and a time series module for receiving and processing the medical data corresponding to the patient.

### Claim 18

An apparatus comprising at least one processor, a communications interface configured for communicating via at least one network, and at least one memory storing computer program code, the at least one memory and the computer program code configured to, with the processor, cause the apparatus to at least:
responsive to receiving an indication of initiation of a medical event, initiate a deep learning model for the patient, wherein the deep learning model (a) comprises a modified recurrent neural network (RNN) with gated recurrent units (GRUs), (b) is executed by an assessment computing entity, (c) has been trained using machine learning, and (d) is configured to generate a prediction for the patient comprising at least one of an acuity score or a mortality prediction;
responsive to identifying a prediction trigger, automatically update the deep learning model for the patient based at least in part on medical data (a) corresponding to the patient and (b) captured during the medical event as part of one or more time series of medical data;
automatically generate the prediction using the updated deep learning model; and
automatically provide at least a portion of the prediction such that the at least a portion of the prediction may be used to update an electronic health record corresponding to the patient and/or provided to a clinician for review.

### Claim 19 (depends on 18)

The apparatus of claim 18, wherein the deep learning model comprises a static module for receiving and processing background information corresponding to the patient and a time series module for receiving and processing the medical data corresponding to the patient.

### Claim 20

A computer program product comprising at least one non-transitory computer-readable storage medium having computer-readable program code portions stored therein, the computer-readable program code portions comprising executable portions configured, when executed by a processor of an apparatus, to cause the apparatus to:
responsive to receiving an indication of initiation of a medical event, initiate a deep learning model for the patient, wherein the deep learning model (a) comprises a modified recurrent neural network (RNN) with gated recurrent units (GRUs), (b) is executed by an assessment computing entity, (c) has been trained using machine learning, and (d) is configured to generate a prediction for the patient comprising at least one of an acuity score or a mortality prediction;
responsive to identifying a prediction trigger, automatically update the deep learning model for the patient based at least in part on medical data (a) corresponding to the patient and (b) captured during the medical event as part of one or more time series of medical data;
automatically generate the prediction using the updated deep learning model; and
automatically provide at least a portion of the prediction such that the at least a portion of the prediction may be used to update an electronic health record corresponding to the patient and/or provided to a clinician for review.
