# US 12,009,106 B2 · Emergency demand prediction device, emergency demand prediction method, and program

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

- Number: US 12,009,106 B2
- Document ID: US12009106B2
- Status: Granted
- Kind: B2
- Application number: 17/294,514
- Filed: 2019-11-08
- Published: 2022-01-20
- Granted: 2024-06-11
- Earliest claimed: 2018-11-19
- Assignees: Nippon Telegraph and Telephone Corporation
- Applicants: Nippon Telegraph and Telephone Corporation
- Inventors: Sunyong Kim; Ippei Shake; Kazuaki Obana; Atsuhiko Maeda; Michiharu Takemoto; Yukio Kikuya; Hiroshi Sato; Tetsuo Kawano; Kenichi Fukuda
- Web page: https://patentgrep.com/patent/US12009106B2

## CPC

- G16H 50/70: for mining of medical data, e.g. analysing previous cases of other patients
- G06Q 10/04: Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem" (market predictions or forecasting for commercial activities G06Q30/0202)
- G06Q 50/26: Government or public services (business processes related to the transportation industry G06Q50/40)
- G16H 40/20: for the management or administration of healthcare resources or facilities, e.g. managing hospital staff or surgery rooms
- G16H 50/20: for computer-aided diagnosis, e.g. based on medical expert systems
- G16H: HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR

## Abstract

Provided is a technique for efficiently predicting the number (quantity) of occurrences of emergency medical service requests in a target area. An emergency medical service demand prediction device according to an embodiment obtains actual history data including information about dates and times of occurrences of emergency medical service requests, information about positions of the occurrences of the emergency medical service requests, and information about illnesses and injuries that caused the emergency medical service requests; the device generates a first learning model that receives an input of first learning-purpose data generated on a basis of learning-purpose actual history data and outputs illness/injury groups; the device generates a second learning model that receives an input of second learning-purpose data generated on a basis of the learning-purpose actual history data and the illness/injury groups output from the first learning model and outputs a value indicating a quantity of occurrences of emergency medical service requests for each unit area; and the device predicts a quantity of occurrences of emergency medical service requests in each unit area, by inputting, to the second learning model having been trained, prediction-purpose data generated on a basis of prediction-purpose actual history data and the illness/injury groups output from the first learning model having been trained.

## Claims

### Claim 1

An emergency medical service demand prediction device that predicts a quantity of occurrences of emergency medical service requests in a target area, the emergency medical service demand prediction device comprising:
a memory configured to store computer program instructions; and
a processor configured to execute the computer program instructions so as to:
obtain actual history data including transport data from an external emergency medical service database, information transmitted from ambulances currently responding to the emergency medical service requests, date/time information indicating dates and times of occurrences of the emergency medical service requests, position information indicating places of the occurrences of the emergency medical service requests, illness/injury information indicating illnesses and injuries that caused the emergency medical service requests, and attribute information of patients including ages and genders;
pre-process the actual history data by dividing the actual history data into sections, extracting related items, and performing a normalization process to generate:
illness/injury group learning-purpose data used for learning the illnesses and injuries;
prediction model learning-purpose data used for learning a prediction model based on an actual number of the occurrences, the date/time information of each of the illnesses and injuries, and the position information; and
request occurrence number prediction-purpose data used for predicting a quantity of occurrences of future emergency medical service requests;
generate a first learning model, based on the illness/injury group learning-purpose data, including illness/injury groups in which the illnesses and injuries are grouped by a type of the illnesses and injuries by learning occurrence patterns of the illnesses and injuries, and similarities or distances among the illnesses and injuries in association with the dates and times, the position information, and the attribute information of the patients;
generate a second learning model, based on the first leaning model and the prediction model learning-purpose data, including a learned prediction model in which a relationship among a number of occurrences of the illnesses and injuries, the illness/injury groups, the dates and times, and the position information are learned, the learned prediction model indicating an actual number of the occurrences of the emergency medical service requests by the illness/injury groups, unit time, and unit areas in the position information;
predict the quantity of the occurrences of the future emergency medical service requests in each of the unit areas within the target area based on the second learning model and the request occurrence number prediction-purpose data; and
output the predicted quantity of the occurrences of the future emergency medical service requests to an external device.

### Claim 2 (depends on 1)

The emergency medical service demand prediction device according to claim 1,
wherein the processor is further configured to:
obtain environment data including information related to meteorology of the places of the occurrences of the emergency medical service requests;
generate the second learning model by further receiving an input of the environment data; and
predict the quantity of the occurrences of the future emergency medical service requests for each of the unit areas, by further inputting the environment data to the second learning model having been trained.

### Claim 3 (depends on 1)

The emergency medical service demand prediction device according to claim 1,
wherein the processor is further configured to:
obtain region data including information related to a regional statistic of the places of the occurrences of the emergency medical service requests;
generate the second learning model by further receiving an input of the region data; and
predict the quantity of the occurrences of the future emergency medical service requests for each of the unit areas, by further inputting the region data to the second learning model having been trained.

### Claim 4 (depends on 1)

The emergency medical service demand prediction device according to claim 1,
wherein the processor is further configured to:
obtain region data including information related to a regional statistic of the places of the occurrences of the emergency medical service requests;
generate a third learning model which receives an input of the region data and output a regional feature value for each of the unit areas;
generate the second learning model by further receiving an input of the regional feature value output from the third learning model; and
predict the quantity of the occurrences of the future emergency medical service requests for each of the unit areas by further inputting the regional feature value output from the third learning model having been trained, to the second learning model having been trained.

### Claim 5 (depends on 1)

The emergency medical service demand prediction device according to claim 1, wherein
the first learning model is structured by using a first layer of a neural network, and
the second learning model is structured by using a second layer of the neural network that receives an output of the first layer as an input.

### Claim 6 (depends on 1)

The emergency medical service demand prediction device according to claim 1,
wherein the processor is configured to generate and output the predicted quantity of the occurrences of the future emergency medical service requests to the external device for visually presenting a prediction result.

### Claim 7 (depends on 1)

The emergency medical service demand prediction device according to claim 1,
wherein the processor is further configured to:
determine whether the predicted quantity of the occurrences of the future emergency medical service requests deviates an actual quantity of the occurrences of the future emergency medical service requests in each of the unit areas within the target area in the future for a predetermined period of time; and
perform a re-learn process in which the obtaining of the actual history data, the pre-processing of the actual history data, the generating the first learning model, the generating the second learning model, the predicting of the quantity of the occurrences of the future emergency medical service requests, and the outputting the predicted quantity of the occurrences of the future emergency medical service requests are repeated when the processor determines that the predicted quantity of the occurrences of the future emergency medical service requests deviates the actual quantity of the occurrences of the future emergency medical service requests in each of the unit areas within the target area in the future for the predetermined period of time.

### Claim 8

An emergency medical service demand prediction method for causing a processor to execute computer program instructions and being implemented by an emergency medical service demand prediction device that predicts a quantity of occurrences of emergency medical service requests in a target area, the emergency medical service demand prediction method comprising executing on the processor the steps of:
obtaining actual history data including transport data from an external emergency medical service database, information transmitted from ambulances currently responding to the emergency medical service requests, date/time information indicating dates and times of occurrences of the emergency medical service requests, position information indicating places of the occurrences of the emergency medical service requests, illness/injury information indicating illnesses and injuries that caused the emergency medical service requests, and attribute information of patients including ages and genders;
pre-processing the actual history data by dividing the actual history data into sections, extracting related items, and performing a normalization process to generate:
illness/injury group learning-purpose data used for learning the illnesses and injuries;
prediction model learning-purpose data used for learning a prediction model based on an actual number of the occurrences, the date/time information of each of the illnesses and injuries, and the position information; and
request occurrence number prediction-purpose data used for predicting a quantity of occurrences of future emergency medical service requests;
generating a first learning model, based on the illness/injury group learning-purpose data, including illness/injury groups in which the illnesses and injuries are grouped by a type of the illnesses and injuries by learning occurrence patterns of the illnesses and injuries, and similarities or distances among the illnesses and injuries in association with the dates and times, the position information, and the attribute information of the patients;
generating a second learning model, based on the first leaning model and the prediction model learning-purpose data, including a learned prediction model in which a relationship among a number of occurrences of the illnesses and injuries, the illness/injury groups, the dates and times, and the position information are learned, the learned prediction model indicating an actual number of the occurrences of the emergency medical service requests by the illness/injury groups, unit time, and unit areas in the position information;
predicting the quantity of the occurrences of the future emergency medical service requests in each of the unit areas within the target area based on the second learning model and the request occurrence number prediction-purpose data; and
outputting the predicted quantity of the occurrences of the future emergency medical service requests to an external device.

### Claim 9 (depends on 8)

The emergency medical service demand prediction method according to claim 8, further comprising:
obtaining environment data including information related to meteorology of the places of the occurrences of the emergency medical service requests;
generating the second learning model by further receiving an input of the environment data; and
predicting the quantity of the occurrences of the future emergency medical service requests for each of the unit areas, by further inputting the environment data to the second learning model having been trained.

### Claim 10 (depends on 8)

The emergency medical service demand prediction method according to claim 8, further comprising:
obtaining region data including information related to a regional statistic of the places of the occurrences of the emergency medical service requests;
generating the second learning model by further receiving an input of the region data; and
predicting the quantity of the occurrences of the future emergency medical service requests for each of the unit areas, by further inputting the region data to the second learning model having been trained.

### Claim 11 (depends on 8)

The emergency medical service demand prediction method according to claim 8, further comprising:
obtaining region data including information related to a regional statistic of the places of the occurrences of the emergency medical service requests;
generating a third learning model which receives an input of the region data and output a regional feature value for each of the unit areas;
generating the second learning model by further receiving an input of the regional feature value output from the third learning model; and
predicting the quantity of the occurrences of the future emergency medical service requests for each of the unit areas by further inputting the regional feature value output from the third learning model having been trained, to the second learning model having been trained.

### Claim 12 (depends on 8)

The emergency medical service demand prediction method according to claim 8, wherein
the first learning model is structured by using a first layer of a neural network, and
the second learning model is structured by using a second layer of the neural network that receives an output of the first layer as an input.

### Claim 13 (depends on 8)

The emergency medical service demand prediction method according to claim 8, further comprising:
determining whether the predicted quantity of the occurrences of the future emergency medical service requests deviates an actual quantity of the occurrences of the future emergency medical service requests in each of the unit areas within the target area in the future for a predetermined period of time; and
performing a re-learn process in which the obtaining of the actual history data, the pre-processing of the actual history data, the generating the first learning model, the generating the second learning model, the predicting of the quantity of the occurrences of the future emergency medical service requests, and the outputting the predicted quantity of the occurrences of the future emergency medical service requests are repeated when the processor determines that the predicted quantity of the occurrences of the future emergency medical service requests deviates the actual quantity of the occurrences of the future emergency medical service requests in each of the unit areas within the target area in the future for the predetermined period of time.

### Claim 14

A non-transitory computer-readable medium storing computer program instructions for causing a computer to execute a process by a processor so as to perform the steps of:
obtaining actual history data including transport data from an external emergency medical service database, information transmitted from ambulances currently responding to emergency medical service requests, date/time information indicating dates and times of occurrences of the emergency medical service requests, position information indicating places of the occurrences of the emergency medical service requests, illness/injury information indicating illnesses and injuries that caused the emergency medical service requests, and attribute information of patients including ages and genders;
pre-processing the actual history data by dividing the actual history data into sections, extracting related items, and performing a normalization process to generate:
illness/injury group learning-purpose data used for learning the illnesses and injuries;
prediction model learning-purpose data used for learning a prediction model based on an actual number of the occurrences, the date/time information of each of the illnesses and injuries, and the position information; and
request occurrence number prediction-purpose data used for predicting a quantity of occurrences of future emergency medical service requests;
generating a first learning model, based on the illness/injury group learning-purpose data, including illness/injury groups in which the illnesses and injuries are grouped by a type of the illnesses and injuries by learning occurrence patterns of the illnesses and injuries, and similarities or distances among the illnesses and injuries in association with the dates and times, the position information, and the attribute information of the patients;
generating a second learning model, based on the first leaning model and the prediction model learning-purpose data, including a learned prediction model in which a relationship among a number of occurrences of the illnesses and injuries, the illness/injury groups, the dates and times, and the position information are learned, the learned prediction model indicating an actual number of the occurrences of the emergency medical service requests by the illness/injury groups, unit time, and unit areas in the position information within a target area;
predicting the quantity of the occurrences of the future emergency medical service requests in each of the unit areas within the target area based on the second learning model and the request occurrence number prediction-purpose data; and
outputting the predicted quantity of the occurrences of the future emergency medical service requests to an external device.

### Claim 15 (depends on 14)

The non-transitory computer-readable medium according to claim 14, further comprising:
obtaining environment data including information related to meteorology of the places of the occurrences of the emergency medical service requests;
generating the second learning model by further receiving an input of the environment data; and
predicting the quantity of the occurrences of the future emergency medical service requests for each of the unit areas, by further inputting the environment data to the second learning model having been trained.

### Claim 16 (depends on 14)

The non-transitory computer-readable medium according to claim 14, further comprising:
obtaining region data including information related to a regional statistic of the places of the occurrences of the emergency medical service requests;
generating the second learning model by further receiving an input of the region data; and
predicting the quantity of the occurrences of the future emergency medical service requests for each of the unit areas, by further inputting the region data to the second learning model having been trained.

### Claim 17 (depends on 14)

The non-transitory computer-readable medium according to claim 14, further comprising:
obtaining region data including information related to a regional statistic of the places of the occurrences of the emergency medical service requests;
generating a third learning model which receives an input of the region data and output a regional feature value for each of the unit areas;
generating the second learning model by further receiving an input of the regional feature value output from the third learning model; and
predicting the quantity of the occurrences of the future emergency medical service requests for each of the unit areas by further inputting the regional feature value output from the third learning model having been trained, to the second learning model having been trained.

### Claim 18 (depends on 14)

The non-transitory computer-readable medium according to claim 14, wherein
the first learning model is structured by using a first layer of a neural network, and
the second learning model is structured by using a second layer of the neural network that receives an output of the first layer as an input.

### Claim 19 (depends on 14)

The non-transitory computer-readable medium according to claim 14, further comprising:
determining whether the predicted quantity of the occurrences of the future emergency medical service requests deviates an actual quantity of the occurrences of the future emergency medical service requests in each of the unit areas within the target area in the future for a predetermined period of time; and
performing a re-learn process in which the obtaining of the actual history data, the pre-processing of the actual history data, the generating the first learning model, the generating the second learning model, the predicting of the quantity of the occurrences of the future emergency medical service requests, and the outputting the predicted quantity of the occurrences of the future emergency medical service requests are repeated when the processor determines that the predicted quantity of the occurrences of the future emergency medical service requests deviates the actual quantity of the occurrences of the future emergency medical service requests in each of the unit areas within the target area in the future for the predetermined period of time.