Emergency Demand Forecast (Collaborative Research with Yokohama City University)

In fiscal 2017, we have been conducting research with the goal of modeling the number of emergency participations and calculating predicted values. We announce the result at the mayor regular press conference and are summarized in press release document.

The following is the data used for modeling.

For each item in the data

• Year of recognition: The year when City of Yokohama, Fire Bureau was aware of the request for an ambulance (four digits AD)
• A month when City of Yokohama, Fire Bureau was aware of the request for an ambulance (1-12)
• Date of recognition: The day when City of Yokohama, Fire Bureau perceived the request for an ambulance (1-31)
• When City of Yokohama, Fire Bureau is aware of the request for an ambulance (0-23)
• Number of injured persons: Includes not only those who transported but also those who did not finally transport for some reason.

※If the data is not entered, it will be "Not entered".

For each item in the data

• Year of recognition: The year when City of Yokohama, Fire Bureau was aware of the request for an ambulance (four digits AD)
• A month when City of Yokohama, Fire Bureau was aware of the request for an ambulance (1-12)
• Housing category: Address of the victim (city, suburbs, foreign countries, others)
  • City: Injury living in Yokohama City
  • Outside the city: Injured living outside Yokohama (Japan)
  • Foreign: Injuries living outside Japan
  • Others: Injuries who do not know Address, such as unknown address or unknown address
• Number of injured persons: Includes not only those who transported but also those who did not finally transport for some reason.

※If the data is not entered, it will be "Not entered".

For each item in the data

• Year of recognition: The year when City of Yokohama, Fire Bureau was aware of the request for an ambulance (four digits AD)
• A month when City of Yokohama, Fire Bureau was aware of the request for an ambulance (1-12)
• 18 administrative districts (Tsurumi Ward, Kanagawa Ward, Nishi Ward, Naka Ward, Minami Ward, Konan Ward, Hodogaya Ward, Asahi Ward). Isogo Ward, Kanazawa Ward, Kohoku Ward, Midori Ward, Aoba Ward, Tsuzuki Ward  , Totsuka Ward, Sakae Ward, Izumi Ward, Seya Ward) and outside the city
• Number of injured persons: Includes not only those who transported but also those who did not finally transport for some reason.

※If the data is not entered, it will be "Not entered".

For each item in the data

• Year of recognition: The year when City of Yokohama, Fire Bureau was aware of the request for an ambulance (four digits AD)
• A month when City of Yokohama, Fire Bureau was aware of the request for an ambulance (1-12)
• Accident type: Reason for emergency request (sudden illness, general injury, traffic accident)
  • Sudden illness: illness, treated as emergency services
  • General injury: Accidents that are not classified as fire, natural disaster, water shortage, work related accidents, athletic competition, injury, self-harm, etc.
  • Traffic accidents: Accidents caused by collisions or contact with traffic or pedestrians
• Age classification: Age classification of the victim (elderly, adult, boy or younger)
  • Elderly people: 65 years old or older
  • Adults: 18 to under 64 years old
  • Juveniles and younger: 17 years old or younger
• Number of injured persons: Includes not only those who transported but also those who did not finally transport for some reason.

※If the data is not entered, it will be "Not entered".

This is the annual data obtained by extracting the budget for proper use of emergency services from the public relations expenses of the City of Yokohama, Fire Bureau Emergency Planning Division.

Yokohama City Emergency Telephone Consultation is a service that started in 2016, and since there is still little data, it was fixed at 3.07%, which is the utilization rate (compared to the city population) in 2016.

Using "Chapter 2_Population_5_ Age (each age, 5 years old class), administrative district, population by gender" of Yokohama City Statistical Report, we obtained data on city population.

Similarly, using the Yokohama City Statistical Report “Chapter 2_Population_19_Administrative Situation of Commuting / School Population_(1) Daytime / Night Population, Inflow Population and Population Density”, there will be no major fluctuations in the influx population until 2030. Based on the assumption that there is no basic data was created.

We used future population projections calculated by Yokohama City based on the census (at that time, the 2010 standard).

We used weather data for each date and time during 2016 from the past weather data (outside site) published by the Japan Meteorological Agency (outside site).

We obtained the total number of foreign guests from 2010 to 2015 from the annual report published by Culture and Tourism Bureau, Yokohama City.

Estimates the total number of foreign guests up to 2030 based on this data and the inbound target value (outside site) announced by the Japanese government at the "Tourism Vision Initiative Conference to Support Japan tomorrow", and uses it as the basic data for predicting emergency demand in Yokohama City.

In addition to Saturday and Sunday, national holidays (outside site) are designated as "holidays", and other days are designated as weekdays.

The number of cases per day (number / day) and the number of cases per hour (number / hour) were calculated for each of the following categories.

• Housing classification
• Housing category (per hour)
• Administrative Districts
• Cross classification by age group and accident type

The optimal model was searched using a heavy regression analysis using the response variable and a factor candidate as an explanatory variable for the number of cases/days obtained above. In other words, when y is a vector of a number of cases/day (response variable) and X as a planning matrix of a factor candidate (explanatory variable), the following model was applied to the data.

y=Xβ+ε, however, ε to Nn(0, σ2P)

Here, ε is an error vector that follows n-dimensional regular distribution Nn with an average vector of 0 and a covariance matrix σ2P (P is a correlation matrix), and n is the number of days to be analyzed (number of specimens). The error section assumes a self-regression (AR) structure, and model selection is based on AIC. For the number/hour model, we used a model that added variables related to perception (dummy variables), demographics, and alternating effects during perception to the model selected in number/day.

The estimation of the coefficient parameter β of each model obtained by the above procedure is as follows.

Using the optimum model obtained from the search, we provided explanatory variables with estimates for the future year to calculate the number/day and number/time forecast for 2017-2030.

For each item in the data

•Year of recognition: The year when City of Yokohama, Fire Bureau perceives the request for an ambulance (four digits AD)
•Average number of emergency participations per day for residents in the city (predicted values)
•Average number of emergency participations per day for residents outside the city (predicted values)
•Average number of emergency services per day for foreign residents (predicted values)

For each item in the data

• Year of recognition: The year when City of Yokohama, Fire Bureau perceives the request for an ambulance (four digits AD)
• When City of Yokohama, Fire Bureau is aware of the request for an ambulance (0-23)
• Average number of emergency participations per day for residents in the city (predicted values)
• Average number of emergency participations per day for residents outside the city (predicted values)
• Average number of emergency services per day for foreign residents (predicted values)

For each item in the data

• Year of recognition: The year when City of Yokohama, Fire Bureau perceives the request for an ambulance (four digits AD)
• Average number of entries per day in Tsurumi Ward
• Average number of entries per day in Kanagawa Ward
• Average number of entries per day in Nishi Ward
• Average number of entries per day in Naka Ward
• Average number of entries per day in Minami Ward
• Average number of entries per day in Konan Ward
• Average number of entries per day in Hodogaya Ward
• Average number of entries per day in Asahi Ward
• Average number of entries per day in Isogo Ward
• Average number of entries per day in Kanazawa Ward
• Average number of entries per day in Kohoku Ward
• Average number of entries per day in Midori Ward
• Average number of entries per day in Aoba Ward
• Average number of entries per day in Tsuzuki Ward  
• Average number of entries per day in Totsuka Ward
• Average number of entries per day in Sakae Ward
• Average number of entries per day in Izumi Ward
• Average number of entries per day in Seya Ward
• Average number of participants per day outside the city (predicted values)

For each item in the data

• Year of recognition: The year when City of Yokohama, Fire Bureau perceives the request for an ambulance (four digits AD)
• Average number of injured patients per day: Sudden illness for boys and younger (predicted value)
• Average number of injured patients per day: General injuries under boys (predicted values)
• Average number of injured patients per day: Traffic accidents for boys and younger (predicted values)
• Average number of injured patients per day: Adult sudden illness (predicted value)
• Average number of injured patients per day: General injury in adults (predicted values)
• Average number of injured patients per day: Traffic accidents in adults (predicted values)
• Average number of injured patients per day: Emergency illness of elderly people (predicted value)
• Average number of injured patients per day: General injury of elderly people (predicted values)
• Average number of injured patients per day: Traffic accidents involving elderly people (predicted values)