# US 11,443,204 B2 · Computer system and method of presenting information on basis of prediction result for input data

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

- Number: US 11,443,204 B2
- Document ID: US11443204B2
- Status: Granted
- Kind: B2
- Application number: 16/702,702
- Filed: 2019-12-04
- Published: 2020-07-30
- Granted: 2022-09-13
- Earliest claimed: 2019-01-30
- Assignees: Hitachi, Ltd.
- Applicants: Hitachi, Ltd.
- Inventors: Naofumi Hama; Masashi Egi; Yasuhide Mori
- Web page: https://patentgrep.com/patent/US11443204B2

## CPC

- G06N 5/04: Inference or reasoning models
- G16H 50/20: for computer-aided diagnosis, e.g. based on medical expert systems
- G06F 3/14: Digital output to display device {; Cooperation and interconnection of the display device with other functional units}
- G06N 20/00: Machine learning
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G16H: HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

A computer system stores interpretation factor conversion information for managing an interpretation factor interpreting a basis of a prediction result for input data, the interpretation factor is determined by a value of each of a plurality of feature quantities contained in the input data including values of the plurality of feature quantities, and a first evaluation value of each of the plurality of feature quantities contained in the input data. When evaluation target data is input, the computer system calculates a prediction result, calculates a contribution value of each of the plurality feature quantities contained in the evaluation target data, specifies a corresponding interpretation factor, based on a value and a contribution value of each of the plurality of feature quantities contained in the evaluation target data, by referring to the interpretation factor conversion information, and generates and outputs display information for presenting the specified interpretation factor.

## Claims

### Claim 1

A computer system comprising:
at least one computer that includes a processor and a memory connected to the processor, wherein
the memory stores:
model information for predicting an event of a target based on input data including a value of a plurality of feature quantities indicating a state of the target, and
interpretation factor conversion information for managing an interpretation factor interpreting a basis of a prediction result for the input data, the interpretation factor being determined by a value of each of the plurality of feature quantities contained in the input data and a first evaluation value of the value of each of the plurality of feature quantities contained in the input data, the first evaluation value showing magnitude of contribution of the value of the feature quantity to a prediction result for the input data, and
the processor is configured to:
calculate, when evaluation target data which is the input data to be evaluated is input, a prediction result for the evaluation target data based on the model information,
calculate a first evaluation value of each of the plurality of feature quantities contained in the evaluation target data,
specify a corresponding interpretation factor, based on the value and first evaluation value of each of the plurality of feature quantities contained in the evaluation target data, by referring to the interpretation factor conversion information, and
generate display information for presenting the specified interpretation factor and output the display information.

### Claim 2 (depends on 1)

The computer system according to claim 1, wherein
the interpretation factor conversion information includes conversion data in which a set is associated with the interpretation factor, the set including, as elements, a combination of an area of the value of the feature quantity contained in the input data and an area of the first evaluation value of the feature quantity contained in the input data, and
the processor is configured to:
specify an area including a value of each of the plurality of feature quantities contained in the evaluation target data,
specify an area including a first evaluation value of each of the plurality of feature quantities contained in the evaluation target data,
retrieve, by referring to the interpretation factor conversion information, the conversion data corresponding to a subset, the subset including, as elements, a combination of the area including the value of each of the plurality of feature quantities contained in the evaluation target data and the area including the first evaluation value of each of the plurality of feature quantities contained in the evaluation target data, and
specify the interpretation factor from the retrieved conversion data.

### Claim 3 (depends on 2)

The computer system according to claim 2, wherein
the processor is configured to:
acquire area division information defining areas of the value and the first evaluation value of each of the plurality of feature quantities contained in the input data,
acquire data for generation which is input data for generating the interpretation factor conversion information,
calculate a prediction result of the data for generation and the first evaluation value of each of the plurality of feature quantities contained in the data for generation,
generate subsets of the plurality of feature quantities contained in the input data,
select a target subset from the subsets,
calculate a second evaluation value showing magnitude of contribution of a combination of values of a plurality of feature quantities to a prediction result for the data for generation, the plurality of feature quantities being contained in the data for generation and being elements of the target subset,
specify, when the second evaluation value is larger than a threshold, an area including the value of the feature quantity, based on the area division information, for the value of each of the plurality of feature quantities that are contained in the data for generation and are elements of the target subset, and specify an area including the first evaluation value of the feature quantity, based on the area division information, for the first evaluation value of each of the plurality of feature quantities that are contained in the data for generation and are contained in the target subset,
generate a set as candidate conversion data, the set including the specified area of the values of the plurality of feature quantities and the specified area of the first evaluation values as elements, and
generate the conversion data by associating the interpretation factor with the candidate conversion data and store the conversion data in the interpretation factor conversion information.

### Claim 4 (depends on 3)

The computer system according to claim 3, wherein
the processor is configured to:
acquire integration condition information defining a condition under which a plurality of pieces of the candidate conversion data is integrated,
specify, based on the integration condition information, the plurality of pieces of candidate conversion data which are able to be integrated,
integrate the plurality of pieces of specified candidate conversion data, and
generate the conversion data by associating the interpretation factor with the plurality of pieces of integrated candidate conversion data.

### Claim 5 (depends on 4)

The computer system according to claim 4, wherein
the integration condition information includes a type and a threshold of an index indicating similarity between the candidate conversion data, and
the processor is configured to:
calculate the index between the candidate conversion data, and
specify, based on a comparison result of the index and the threshold, the plurality of pieces of candidate conversion data which are able to be integrated.

### Claim 6

A method of presenting information on a prediction result of prediction for input data executed by a computer system, wherein:
the computer system includes
at least one computer that includes a processor and a memory connected to the processor,
the memory storing model information for predicting an event of a target based on input data including a plurality of feature quantities indicating a state of the target, and interpretation factor conversion information for managing an interpretation factor interpreting a basis of a prediction result for the input data, the interpretation factor being determined by a value of each of the plurality of feature quantities contained in the input data and a first evaluation value of the value of each of the plurality of feature quantities contained in the input data, the first evaluation value showing magnitude of contribution of the value of the feature quantity to a prediction result for the input data,
the method of presenting information comprising:
a first step in which the processor is configured to calculate, when an evaluation target data which is the input data to be evaluated is input, a prediction result for the evaluation target data based on the model information;
a second step in which the processor is configured to calculate a first evaluation value of each of the plurality of feature quantities contained in the evaluation target data;
a third step in which the processor is configured to specify a corresponding interpretation factor, based on the value and first evaluation value of each of the plurality of feature quantities contained in the evaluation target data, by referring to the interpretation factor conversion information; and
a fourth step in which the processor is configured to generate display information for presenting the specified interpretation factor and output the display information.

### Claim 7 (depends on 6)

The method of presenting information according to claim 6, wherein
the interpretation factor conversion information includes conversion data in which a set is associated with the interpretation factor, the set including, as elements, a combination of an area of the value of the feature quantity contained in the input data and an area of the first evaluation value of the feature quantity contained in the input data,
wherein the third step includes:
a step in which the processor is configured to specify an area including a value of each of the plurality of feature quantities contained in the evaluation target data;
a step in which the processor is configured to specify an area including a first evaluation value of each of the plurality of feature quantities contained in the evaluation target data;
a step in which the processor is configured to retrieve, by referring to the interpretation factor conversion information, the conversion data corresponding to a subset, the subset including, as elements, a combination of the area including the value of each of the plurality of feature quantities contained in the evaluation target data and the area including the first evaluation value of each of the plurality of feature quantities contained in the evaluation target data; and
a step in which the processor is configured to specify the interpretation factor from the retrieved conversion data.

### Claim 8 (depends on 7)

The method of presenting information according to claim 7, further comprising:
a fifth step in which the processor is configured to acquire area division information defining areas of the value and the first evaluation value of each of the plurality of feature quantities contained in the input data;
a sixth step in which the processor is configured to acquire data for generation which is input data for generating the interpretation factor conversion information;
a seventh step in which the processor is configured to calculate a prediction result of the data for generation and a first evaluation value of each of a plurality of feature quantities contained in the data for generation;
an eighth step in which the processor is configured to generate subsets of the plurality of feature quantities contained in the input data;
a ninth step in which the processor is configured to select a target subset from the subsets;
a tenth step in which the processor is configured to calculate a second evaluation value showing magnitude of contribution of a combination of values of a plurality of feature quantities to a prediction result for the data for generation, the plurality of feature quantities being contained in the data for generation and being elements of the target subset;
an eleventh step in which the processor is configured to specify, when the second evaluation value is larger than a threshold, an area including the value of the feature quantity, based on the area division information, for the value of each of the plurality of feature quantities that are contained in the data for generation and are elements of the target subset, and specify an area including the first evaluation value of the feature quantity, based on the area division information, for the first evaluation value of each of the plurality of feature quantities that are contained in the data for generation and are contained in the target subset;
a twelfth step in which the processor is configured to generate a set as candidate conversion data, the set including the specified area of the values of the plurality of feature quantities and the specified area of the first evaluation values as elements; and
a thirteenth step in which the processor is configured to generate the conversion data by associating the interpretation factor with the candidate conversion data and store the conversion data in the interpretation factor conversion information.

### Claim 9 (depends on 8)

The method of presenting information according to claim 8, wherein
the twelfth step includes:
a step in which the processor is configured to acquire integration condition information defining a condition under which a plurality of pieces of the candidate conversion data is integrated;
a step in which the processor is configured to specify, based on the integration condition information, the plurality of pieces of candidate conversion data which are able to be integrated; and
a step in which the processor is configured to integrate the plurality of pieces of specified candidate conversion data,
the thirteenth step includes
a step in which the processor is configured to generate the conversion data by associating the interpretation factor with the plurality of pieces of integrated candidate conversion data.

### Claim 10 (depends on 9)

The method of presenting information according to claim 9, wherein
the integration condition information includes a type and a threshold of an index indicating similarity between the candidate conversion data, and
the method of presenting information further comprising:
a step in which the processor is configured to calculate the index between the candidate conversion data; and
a step in which the processor is configured to specify, based on a comparison result of the index and the threshold, the plurality of pieces of candidate conversion data which are able to be integrated.