Data Generation Manual

1.Introduction

1.1 Purpose of this manual

This manual introduces the data used in the demo of the policy making support technology "Policy Twin."

2. Use cases

2.1 Use case： Multiple Duplicated Medications

2.1.1 Overview

The input and output of simulations are defined for each use case.

In the Policy Twin demo, you need to prepare the input files described below.

Training data (csv)
Testing data (csv)
Configuration file (json)
Flowcharts (BPMN)
Target indicator display settings (json)

A sample dataset is also provided for executing simulations on the topic of multiple duplicated medications.

sample dataset(zip)

2.1.2 Training data（csv）

In some cases, the simulation uses machine learning to calculate target indicators, in such cases the training data related to the policy must be registered. This data primarily consists of people and thing attribute data used to calculate target indicators.

Since no learning is performed in the simulation of multiple duplicated medications, only the pid column in the first row are described as follows. The details on how to create and use data for learning will be explained when additional use cases are published.

pid

2.1.3 Test data (csv)

This test data is used in the Policy Twin demo to calculate target indicators during a simulation and primarily consists of attribute data for people and things. For example, in the case of multiple duplicated medications, it includes prescription data for medications from multiple people. It is necessary to include the ID (pid) that identifies the user and the attribute data used in the flowchart condition formula described below.

The following example describes the attributes used in the flowchart's conditional expressions, based on prescription data from multiple people. These attributes include individual-level information, such as the number of types of medications, the number of healthcare institutions visited, the number of visits, and whether the individual has illnesses such as cancer.

pid,NUMBER-OF-TYPES_MEDICATION_ENTIRE-PERIOD_1,NUMBER-OF-TYPES_MEDICATION_ENTIRE-PERIOD_2,NUMBER-OF-TYPES_MEDICATION_ENTIRE-PERIOD_3,NUMBER-OF-TYPES_MEDICATION_FIXED-PERIOD_1,NUMBER-OF-TYPES_MEDICATION_FIXED-PERIOD_2,NUMBER-OF-TYPES_MEDICATION_FIXED-PERIOD_3,NUMBER-OF-MEDICAL-INSTITUTIONS_REPEATED-MEDICATION_1,NUMBER-OF-MEDICAL-INSTITUTIONS_REPEATED-MEDICATION_2,NUMBER-OF-MEDICAL-INSTITUTIONS_REPEATED-MEDICATION_3,NUMBER-OF-VISITS_SAME-ILLNESS_1,NUMBER-OF-VISITS_SAME-ILLNESS_2,NUMBER-OF-VISITS_SAME-ILLNESS_3,NUMBER-OF-MEDICAL-INSTITUTIONS_SAME-PRESCRIPTION_1,NUMBER-OF-MEDICAL-INSTITUTIONS_SAME-PRESCRIPTION_2,NUMBER-OF-MEDICAL-INSTITUTIONS_SAME-PRESCRIPTION_3,TOTAL-DAYS_SAME-PRESCRIPTION_1,TOTAL-DAYS_SAME-PRESCRIPTION_2,TOTAL-DAYS_SAME-PRESCRIPTION_3,STATUS_CANCER,STATUS_RENAL-DIALYSIS,STATUS_DEMENTIA,STATUS_SHUNT,STATUS_INTRACTABLE-DISEASE,STATUS_MENTAL-ILLNESS,STATUS_PAST-RECOMMENDATION-NOTICE,FLAG_CONSENT,FLAG_LOCAL-MEDICAL-INSTITUTION
1,10,0,6,3,8,0,4,4,2,20,9,23,0,0,0,21,25,25,0,0,1,0,0,0,0,0,1
2,4,13,19,5,0,4,0,1,2,22,26,9,3,0,1,50,47,54,0,0,0,0,0,0,0,0,1
3,16,0,17,8,7,4,2,4,2,21,28,16,3,0,2,52,17,47,0,1,0,0,1,1,0,0,1
...

2.1.4 Configuration file (json)

This is a configuration file used for evaluating the results of a simulation. The evaluation engine name must be specified according to the type of issue addressed in the simulation.

In the example below, the name of the evaluation engine ("Repeated Medication") is specified to address the issue of multiple duplicated medications.

{
  "engine": "Repeated Medication",
  "config": {
    "treatmentNotation": {
      "visit": ["訪問指導A", "訪問指導" ,"HOME-VISIT-GUIDANCE"],
      "visit-resource": 25,
      "notice": ["勧奨通知A", "通知", "NOTIFICATION"],
      "notice-resource": 300
    },
    "costCoefficient": {
      "visit": 15822,
      "notice": -236
    },
    "improvementCoefficient": {
      "visit": 0.8
    }
  }
}

2.1.5 Flowchart (bpmn)

This is how to create the policy to be used in the simulation as a flowchart in BPMN (Business Process Model and Notation) format. In a simulation, multiple people (corresponding to the test data mentioned above) virtually flow through the flowchart following conditional branches.

The example below shows a policy for addressing multiple duplicated medications (a flowchart for handling multiple duplicated medications cases in a virtual municipality).

2.1.6 Target indicator display settings (json)

This is a configuration file used when displaying the results of a simulation on screen.

In the example below, display names in Japanese and English are specified for the three target indicator values calculated by evaluation engine for duplicated medications.

{
    "widgetFactorComparisonRadar": {
        "chart": {
            "totalCost": {
                "ja": "医療費節減効果（円）",
                "en": "Medical cost savings"
            },
            "citizenRisk": {
                "ja": "通知勧奨数（人）",
                "en": "Number of Notifications"
            },
            "hospitalDays": {
                "ja": "保健指導数（人）",
                "en": "Number of Health Guidance"
            },
            "effect": {
                "ja": "健康改善効果（人）",
                "en": "Health Improvement Effects"
            },
            "userResponses": {
                "ja": "リソース：通知勧奨",
                "en": "Resources : Notifications"
            },
            "resources": {
                "ja": "リソース：保健指導",
                "en": "Resources : Health Guidance"
            }
        }
    },
    "widgetPolicyEffect": {
        "totalCost": {
            "ja": "医療費節減効果（円）",
            "en": "Medical cost savings"
        },
        "citizenRisk": {
            "ja": "通知勧奨数（人）",
            "en": "Number of Notifications"
        },
        "hospitalDays": {
            "ja": "保健指導数（人）",
            "en": "Number of Health Guidance"
        },
        "effect": {
            "ja": "健康改善効果（人）",
            "en": "Health Improvement Effects"
        },
        "userResponses": {
            "ja": "リソース：通知勧奨数",
            "en": "Resources : Notifications"
        },
        "resources": {
            "ja": "リソース：保健指導数",
            "en": "Resources : Health Guidance"
        }
    },
    "dialogFilterCandidates": {
        "totalCost": {
            "ja": "医療費節減効果（円）",
            "en": "Medical cost savings"
        },
        "citizenRisk": {
            "ja": "通知勧奨数（人）",
            "en": "Number of Notifications"
        },
        "hospitalDays": {
            "ja": "保健指導数（人）",
            "en": "Number of Health Guidance"
        },
        "effect": {
            "ja": "健康改善効果（人）",
            "en": "Health Improvement Effects"
        },
        "userResponses": {
            "ja": "リソース：通知勧奨数",
            "en": "Resources : Notifications"
        },
        "resources": {
            "ja": "リソース：保健指導数",
            "en": "Resources : Health Guidance"
        }
    },
    "dialogTargetCandidates": {
        "totalCost": {
            "ja": "医療費節減効果（円）",
            "en": "Medical cost savings"
        },
        "citizenRisk": {
            "ja": "通知勧奨数（人）",
            "en": "Number of Notifications"
        },
        "hospitalDays": {
            "ja": "保健指導数（人）",
            "en": "Number of Health Guidance"
        },
        "effect": {
            "ja": "健康改善効果（人）",
            "en": "Health Improvement Effects"
        },
        "userResponses": {
            "ja": "リソース：通知勧奨数",
            "en": "Resources : Notifications"
        },
        "resources": {
            "ja": "リソース：保健指導数",
            "en": "Resources : Health Guidance"
        }
    },
    "chartAxises": {
        "totalCost": {
            "ja": "医療費節減効果（円）",
            "en": "Medical cost savings"
        },
        "citizenRisk": {
            "ja": "通知勧奨数（人）",
            "en": "Number of Notifications"
        },
        "hospitalDays": {
            "ja": "保健指導数（人）",
            "en": "Number of Health Guidance"
        },
        "effect": {
            "ja": "健康改善効果（人）",
            "en": "Health Improvement Effects"
        },
        "userResponses": {
            "ja": "リソース：通知勧奨数",
            "en": "Resources : Notifications"
        },
        "resources": {
            "ja": "リソース：保健指導数",
            "en": "Resources : Health Guidance"
        }
    },
    "widgetKPI": {
        "totalCost": {
            "ja": "医療費節減効果（円）",
            "en": "Medical cost savings"
        },
        "citizenRisk": {
            "ja": "通知勧奨数（人）",
            "en": "Number of Notifications"
        },
        "hospitalDays": {
            "ja": "保健指導数（人）",
            "en": "Number of Health Guidance"
        },
        "effect": {
            "ja": "健康改善効果（人）",
            "en": "Health Improvement Effects"
        },
        "userResponses": {
            "ja": "リソース：通知勧奨数",
            "en": "Resources : Notifications"
        },
        "resources": {
            "ja": "リソース：保健指導数",
            "en": "Resources : Health Guidance"
        }
    }
}

2.1.7 Description of each target indicator

The effectiveness of the policy is evaluated using the following five target indicators for multiple duplicated medications.

Medical cost savings: Amount of medical cost savings (yen)
Number of Notifications: The number of people requiring notification by postcard, pharmacy, etc. (number of people)
Number of Health Guidance: The number of notified subjects who agree to participate in home visit guidance (number of people)
Health Improvement Effects: The number of participants in the health guidance who are expected to show improvement.
Resources: The correlation between the predetermined maximum resource capacity for health guidance and the number of health guidance mentioned above.