EMS Central Communications Centre (CCC)

EMS Central Communications Centre (CCC)

Staffing Analysis – Final Presentation & Deliverables

Shuang E Scott Van Bolhuis

Derek Hewitt Jenny Morrow

Problem Recap

Reduced Effectiveness and Planning Ability

Over/understaffing Unable to plan for the future

Inefficiency and Simplicity

Lacked Robustness Lacked Scalability Suboptimal Staffing Forecast

Simple Rudimentary

Solution Recap

Increased Effectiveness and Planning Ability

Implementation and Intelligence Possible solutions depending on demand and Increased Planning Ability

Created Model and Controlled for Variability and Inefficiency

Created Staffing model to account for variability

Analyzed Results in ARENA and Improved Robustness

Analyzed Large amounts of Data

CAD and Telephony Found patterns and discovered service times and demand figures

Solution Recap

Queue length = number of customers waiting for service (=state of system minus number of customers being served)N(t) = number of customers in queueing system at time t (t>=0)Pn(t) = probability that exactly n customers are in queueing system at time t, given number at time 0s = number of servers in queueing systemλn = mean arrival rate (expected number of arrivals per unit time) of new customers when n customers are in the systemЧn = mean service rate (expected number of customers completing service per unit time)

Solution Recap (cont’d)

Variable Demand

Evaluator Service Times Based on Call Type

Schedule Model Optimal Staff Required and Optimal Shifts

Performance Measurement and Analysis (ARENA)

Planning Ability

Observations-Call Demand

Demand may increase over time, however, the percentage of weekly demand in each hour should remain about the same

The model uses the percentage of weekly demand per hour to find hourly demand given expected weekly demand

Distribution of Service time

Service Rates Log Normal

Minimum Servers is a Stepwise Function

M/M/s Queueing Model Displays no

Significant Variation

Distribution of Time in System

Observations

0:002:00

4:006:00

8:0010:00

12:0014:00

16:0018:00

20:0022:00

0

5

10

15

20

25

30

StrathERCCCCC

Hour of the Day

Arriv

al R

ate

per H

our

Observations

0:002:00

4:006:00

8:0010:00

12:0014:00

16:0018:00

20:0022:00

0

5

10

15

20

25

30

OtherAir IFTAir EmergTransferNon EmergEmerg

Hours of the Day

Arriv

al R

ate

per H

our

Testing for Correlations

12:00:00 AM

01:00:00 AM

02:00:00 AM

03:00:00 AM

04:00:00 AM

05:00:00 AM

06:00:00 AM

07:00:00 AM

08:00:00 AM

09:00:00 AM

10:00:00 AM

11:00:00 AM

12:00:00 PM

01:00:00 PM

02:00:00 PM

03:00:00 PM

04:00:00 PM

05:00:00 PM

06:00:00 PM

07:00:00 PM

08:00:00 PM

09:00:00 PM

10:00:00 PM

11:00:00 PM0

5

10

15

20

25

30

35

40

WednesdaySaturday

Hour of the Day

Arriv

al R

ate

per H

our

Testing for Correlations

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 240

5

10

15

20

25

30

35

40

WednesdayThursday

Hour of the Day

Arriv

al R

ate

per H

our

Testing for Correlations

Saturday Friday Thursday Wednesday Tuesday MondaySunday 0.934 -4.615 -5.734 -5.677 -4.075 -3.524Monday 3.107 -0.138 -1.844 -1.924 -0.538Tuesday -3.359 -0.948 -4.064 5.092Wednesday 4.582 2.226 0.207Thursday 4.651 2.056Friday 3.840

T-test for Significant Differences

If value <-2 or >2, the two corresponding days have significantly different arrival patterns.The red cells indicates the days that have similar arrival patterns.

Emergency Arrival Queue

Evaluators

Evaluators

Evaluators

IFT Arrival

Pre Alert Pro QA Paramedics Notified

Evaluators Coordination Scheduling

Air Arrival Evaluators

Helicopter Notified

Contact With Ground

Coordination

Hang Up Call back

Process Flow More detailed version in appendices and written report

Station Capacity

Building a Model

Discovering Intermediate Steps

Minimize Inputs Maximize Output

Defining Output

Usefulness PracticalityDeciding Inputs

Minimal Achieve Desired Result

Finding Basic Constraints

• Optimal staffing schedule with the minimum number of call evaluators that can provide desired service level

What we Need

• No less than minimum required servers in each hour

Constraint

• Queueing Toolpak formulas• Inputs• Threshold time, service level, arrival rate, and service rate

How to Find Minimum Required Servers

Determining Threshold and Service Level

Sensitivity Analysis• Found min number of servers required

under different threshold and service levels

Minimum Required Work Stations• Conducted analysis for a peak hour in the

week, so selected results determined the min required work stations

Determining Threshold and Service Level

Model Assumptions

Weekend days follow the same pattern, with less demand

Min servers is a stepwise function so small differences do not matter

Weekdays follow the same pattern

Model Assumes 2 Different Days, Weekdays and Weekend Days

Minimum Required Servers

Weekday Ground Emergency and IFT Calls

Minimum Required Servers

Weekend Ground Emergency and IFT Calls

Model Assumptions

• Different call types have different arrival and service rates and must be accounted for separately

Different Arrival and Service Rates

• Call evaluators are able to answer all different call types

Cross Training

• The minimum required servers that constrains the model is the aggregate of the minimum required servers for each call type

Aggregate Minimum Required Servers

Binary Model - MechanicsSimplified model without union constraints

Binary Model Constraints

Shift start times must be

reasonable

Shift lengths must follow union

guidelines

Breaks must be accounted for

Days must be connected since

shifts wrap into the next day

Creating Useful OutputAdded Union Constraints

Added Start Times and Breaks into the Model to create a more useful schedule

TotalStaffing

hours per week

How to Operate the Model

Step 1 •The first tab “Staff Optimizer 3000” contains 3 input cells, these are expected weekly demand for each call type

•Those input cells properly constrain the model

Step 2 •Go to the “Scheduling Model” tab and press solve

Step 3 •Go to the “Week’s Schedule” tab to find the output•Filter out the 0s in the column labeled “Number of Shifts”

How to Operate the Model

Inputs

Sample Schedule OutputWednesday

Employee Number shift length Shift Start 15 min break 30 min break 15 min break525460 12 5:30:00 AM 8:15:00 AM 11:15:00 AM 2:30:00 PM966492 12 6:30:00 PM 9:15:00 PM 12:15:00 AM 3:30:00 AM308384 12 9:00:00 PM 11:45:00 PM 2:45:00 AM 6:00:00 AM584997 10.5 10:30:00 AM 1:00:00 PM 3:30:00 PM 6:30:00 PM733703 10.5 9:00:00 PM 11:30:00 PM 2:00:00 AM 5:00:00 AM206250 8.25 7:30:00 AM 9:15:00 AM 11:30:00 AM 1:45:00 PM275492 8.25 8:30:00 AM 10:15:00 AM 12:30:00 PM 2:45:00 PM529698 8.25 10:30:00 AM 12:15:00 PM 2:30:00 PM 4:45:00 PM516983 8.25 9:00:00 PM 10:45:00 PM 1:00:00 AM 3:15:00 AM757914 6.25 7:30:00 AM 8:45:00 AM 10:30:00 AM 12:15:00 PM654707 6.25 12:00:00 PM 1:15:00 PM 3:00:00 PM 4:45:00 PM737753 4.25 7:30:00 AM 9:30:00 AM167357 4.25 2:30:00 PM 4:30:00 PM589771 4.25 4:30:00 PM 6:30:00 PM607777 4.25 6:30:00 PM 8:30:00 PM

Model – Analysis/Simulation

Arena

Results of the Simulation

Call Type Average Wait Time (s) Average Service Time (min)Emergency 0.04 1.16Non Emergency 0.09 1.34IFT 0.11 1.17Air Emergency 11.65 4.80Air IFT 9.53 1.51Other 0.03 1.37

Position Utilization Rate Average Number ScheduledCall Evaluator 8.72% 4.09Flight Coordinator 3.88% 1.27

Impact on CCC and more

• Our analysis in action• Reducing costs• Confidence in staffing schedule• Applicability to all centers

Effectiveness

Updated Deliverables

•Summarized analysis, sensitivity Tests and recommendation

•User guide for model

Written report•Provides optimal staffing schedule

subject to Demand changes•Provides results and insightsSchedule

Model and ARENA outputs

•Detailed analysis of processes and actors

•Derived from our modeling

Process flow

What you need

• Runs and optimizes schedule model

Premium Solver

• Supports embedded queueing formulas in model

Queueing Toolpak

We will help you install and use these add-ins and applications through our personal demonstration and user’s

guide.

QUESTIONS?

AppendicesProcess flow

1) Emergency• We describe in detail what the adjacent diagram represents for 911 calls:• Step 1) Emergency incident occurs, caller calls 911 EPS (EPS primary, AHS secondary)

Step 2) Call evaluator verifies location of caller, while location information from TELUS and phone companies is populatedStep 3) Call evaluator prealerts dispatcher and paramedicsStep 4) Conduct ProQA (roughly 45sec) while paramedics are getting readyStep 5) Information populates into CADStep 6) After scenario is confirmed and acuteness identified, paramedics are notifiedStep 7) Time stamp recorded

Note: Children stay on the phone for the entire durationRoughly 10% of the time the call evaluator stays on the line to conduct pre-arrival instructions

2) IFT (Inter-facility transfer)• Step 1) Call or fax from AHS entity or other contracting company• - Fax is pre-booked days in advance, Calls are within hours or the same day• Step 2) Multiple calls can be made and modifications to facility transfer route• -IFT transfer planning is one of the most cognitively demanding positions

Notes:-Dispatching for IFT is not linear and static like 911 calls, can be pushed backed and modified-Seven or more radio calls are used for each IFT (inter-facility transfer call), CCC deals with roughly 150 IFT calls per-day-One person is designated for time-stamping and another person is designated for radio receiving.

3) Air Ambulance• The flight portion of the incoming calls are also a diverse entity. Flight call evaluators have to be fluent in both inter facility transfer coordination as well as emergency

because the incoming flight calls could be either. The IFT’s are pre-booked and the info waits in CAD for 3-4 days before the transfer takes place. Flight calls are also much longer than the normal emergency call and can be overly demanding.

• • Emergency Flight Calls:

Step 1) Helicopter takes off within 30min of callStep 2) Many more calls are made to coordinate activities between ground crew, paramedics and critical care teams, multiple events must be coordinated

Notes: -Time of a call may be double, triple or even longer than a regular ground 911 call-Two flight call evaluators, also take regular calls (one dispatcher)-Heavy call demand for time stamping from multiple areas ( STARS etc)