HeaderSIS.jpg

Difference between revisions of "IS480 Team wiki: 2014T1 Code Blue Project Overview"

From IS480
Jump to navigation Jump to search
 
(18 intermediate revisions by 3 users not shown)
Line 16: Line 16:
  
 
! style="border-style: solid; border-width: 0 1px 1px 0; width: 12%; background-color:#F8F8F8"| [[IS480_Team_wiki:_2014T1 Code Blue_Project Resources| <span style="color:#000066;">Project Resources</span>]]
 
! style="border-style: solid; border-width: 0 1px 1px 0; width: 12%; background-color:#F8F8F8"| [[IS480_Team_wiki:_2014T1 Code Blue_Project Resources| <span style="color:#000066;">Project Resources</span>]]
 
! style="border-style: solid; border-width: 0 1px 1px 0; width: 12%; background-color:#F8F8F8"| [[IS480_Team_wiki:_2014T1 Code Blue_Team Reflections| <span style="color:#000066;">Team Reflection</span>]]
 
 
|}
 
|}
 
<!---------------END of header ---------------------->
 
<!---------------END of header ---------------------->
 +
<!---------------START of Sub-Header ---------------->
 +
{| style="border-collapse: separate; border-spacing: 0; border-width: 3px; border-style: none; border-color: #000066; padding: 0;  height: 30px;" width="100%"
 +
|-https://wiki.smu.edu.sg/is480/Link_title
  
{| class="wikitable" style="width:100%; background: #ffffff; text-align: left; border: 3px solid #0B2161;"
+
! style="border-style: none; border-width: 0 1px 1px 0; width: 12%; background-color:#A9BCF5 "| [[IS480_Team_wiki:_2014T1 Code Blue_Project Overview| <span style="color:#000000;">Project Description</span>]]
 +
 
 +
! style="border-style: none; border-width: 0 1px 1px 0; width: 12%; background-color:#F8F8F8"| [[IS480_Team_wiki:_2014T1 Code Blue_scope| <span style="color:#000000;">Project Scope</span>]]
 +
 
 +
! style="border-style: none; border-width: 0 1px 1px 0; width: 12%; background-color:#F8F8F8"| [[IS480_Team_wiki:_2014T1 Code Blue_tech| <span style="color:#000000;">Technologies</span>]]
 +
 
 +
! style="border-style: none; border-width: 0 1px 1px 0; width: 12%; background-color:#F8F8F8"| [[IS480_Team_wiki:_2014T1 Code Blue_Xfactor| <span style="color:#000000;">X-factor</span>]]
 +
 
 +
|}
 +
<!---------------END of Sub-Header------------------->
 +
<br>
 +
<div style="background: #000066; padding: 12px; font-weight: bold; font-size: 100%; line-height: 0.5em;"><font face="Arial" color="white">Goal & Motivation</font></div>
 +
{| style="width:100%; background: #ffffff; text-align: left;"
 
|-  
 
|-  
 
| [[Image:aim-icon.png|150px]]  
 
| [[Image:aim-icon.png|150px]]  
| The goal of this project is to develop a simulation tool which provides a graphical visualization of a dynamic queue management framework for the hospitals’ emergency department (ED). Focused on managing patient queue dynamically before doctor consultation, the project shall implement the dynamic patient-prioritization strategies. The strategies make use of several greedy algorithms such as Shortest-Consultation-Time-First (SCON) or Shortest-Remaining-Time-First (SREM) to improve on patients’ length of stay (LOS) in the ED.
+
| style="padding-left: 20px;" | The goal of this project is to develop a simulation tool which provides a graphical visualization of the dynamic queue management framework which can be implemented in the hospitals’ Emergency Department (ED). The framework addresses the complex challenges faced by hospital to achieve a desired service level for the patients (e.g. LOS of 90% of patients must be within x minutes). Focused on managing patient queue dynamically before doctor consultation, the project shall implement the dynamic patient-prioritization strategies. The strategies make use of several greedy algorithms such as Shortest-Consultation-Time-First (SCON) or Shortest-Remaining-Time-First (SREM) to improve on patients’ length of stay (LOS) in the ED.
  
 
The aim of this tool is firstly to allow healthcare practitioners to better understand and visualize the mechanism and effects of the proposed strategies; secondly to appreciate how existing data in the database can aid automation and improve operation; and finally to allow visual interaction with a simulation tool.
 
The aim of this tool is firstly to allow healthcare practitioners to better understand and visualize the mechanism and effects of the proposed strategies; secondly to appreciate how existing data in the database can aid automation and improve operation; and finally to allow visual interaction with a simulation tool.
 +
 +
The project will focus on the demand perspective of the integrated framework.
 
|-   
 
|-   
 
| colspan="2" | <hr>
 
| colspan="2" | <hr>
 
|-
 
|-
 +
 
| [[Image:ed-waiting-time.png|150px]]
 
| [[Image:ed-waiting-time.png|150px]]
| Hospitals are facing increasing challenges in today's emergency department due to growth in patients demand for services and limited capacity in resource allocation. (D. J. Medeiros, Eric Swenson & Christopher DeFlitch). The framework addresses the complex challenges faced by hospital to achieve a desired service level for the patients (e.g. LOS of 90% of patients must be within x minutes).
+
| style="padding-left: 20px;" |Hospitals are facing increasing challenges in today's emergency department due to growth in patients demand for services and limited capacity in resource allocation. (D. J. Medeiros, Eric Swenson & Christopher DeFlitch, 2008). Long waiting times and a lack of doctors are often observed and experienced.
 +
 
 +
In the existing literature, addressing the issue of long waiting times in an ED often takes the form of single-faceted queue management strategies that are either from a demand perspective or from a supply perspective. From the demand perspective, there is work on queue design such as priority queues, or queue control strategies such as a fast-track system and demand restriction through ambulance diversion. However, they may not sufficiently leverage insights that can be derived from both historical and real-time data (Kar Way Tan, 2013).  
 +
 
 +
Therefore, an integrated framework that manages queues dynamically in the ED from both the demand and supply perspectives by leveraging historical data and real-time data is being proposed by our sponsor, Assistant Professor (Practice) Kar Way.
 
|-
 
|-
 
| colspan="2" | <hr>
 
| colspan="2" | <hr>
 
|-
 
|-
 
| [[Image:scope-icon.png|150px]]
 
| [[Image:scope-icon.png|150px]]
| The scope of the project includes the following feature set/module:
+
| style="padding-left: 20px;" | The high level scope of the project comprises of the following module:
 
<ul>
 
<ul>
    <li>
+
<li>Simulation Module</li>
      <b>Simulation Input Management</b>
+
The simulation model will be based on a M/M/S queueing theory. The time varying hourly arrival rate of the patients will be following a Poisson process with a lambda that is defined in the data file. The service rate will follow a exponential distribution with time varying service rate that will likewise be defined in the data file. The number of servers will be capped at 5.  
      <ul>
+
 
        <li>Managing simulation settings such as the probability of patient to go through laboratory procedure and probability of different laboratory result simulated</li>
+
 
        <li>Managing simulation decision parameters such as the type of queuing strategy chosen, the historical data file used and the type of simulation to run.</li>
+
There will be 2 assumptions to this simulation model - No Balking and No Reneging.
      </ul>
+
 
    </li>
+
No Balking: Patients left before being in the queue. (eg Patients came to the hospital and noticed that the queue is long and left without registering)
    <li><b>Queue Strategy Engine</b>
+
 
      <ul>
+
No Reneging: Patients registered, but somewhere along the process, left the hospital without being seen.
          <li>Implementation of the strategy Shortest-Consultation-Time-First (SCON). This strategy focuses on using the length of consultation period of the patient to determine his/her priority in the queue.</li>
+
 
          <li>Implementation of the strategy Shortest-Remaining-Time-First (SREM). This strategy focuses on using the patient's remaining length of stay period to determine his/her priority in the queue.</li>
+
<li>Reporting Module</li>
          <li>Implementation of the hybrid strategy Mixed Strategy. This strategy uses both aspects from the SCON and SREM strategy.</li>
 
      </ul>
 
    </li>
 
    <li><b>Visualization of Simulation</b>
 
      <ul>
 
          <li>Viewing the simulation in animation mode</li>
 
          <li>Viewing the simulation in play-back mode</li>
 
      </ul>
 
    </li>
 
    <li><b>Reporting</b>
 
      <ul>
 
          <li>Dashboard-like reporting tool which display the simulation outcome and evaluation of the strategies, such as No.of patient served within target length of stay</li>
 
          <li>Exporting of the report to PDF format by the user</li>
 
      </ul>
 
    </li>
 
 
</ul>
 
</ul>
 +
<p>For a more detailed scope, please refer to our [https://wiki.smu.edu.sg/is480/IS480_Team_wiki%3A_2014T1_Code_Blue_scope project scope].</p>
 
|}
 
|}
 +
 +
== <div style="background: #000066; padding: 12px; font-weight: bold; font-size: 60%; line-height: 0.5em;"><font face="Arial" color="white">Process in the Emergency Department</font></div> ==
 +
[[Image:codeblue_ed_process.png|left|500px]]
 +
<p>To set the context, it is important to understand the process that is happening in the Emergency Department of a hospital. </p>
 +
<p>Patients will first make the necessary registration upon their visit and an initial assessment at the triage. Subsequently, they would proceed to queue for the consultation with the doctor.The doctor would determine if further tests and treatment is needed for the patient after the initial consultation. If not, the patient is immediately discharged or admitted into one of the wards. If they do need additional treatments, they would receive the necessary treatments and rejoin the queue to consult the doctor for a final review before being discharged/admitted.</p>
 +
<br><br>
 +
 +
<div style="background: #000066; padding: 12px; font-weight: bold; font-size: 100%; line-height: 0.5em;"><font face="Arial" color="white">How do this 3 strategies work?</font></div><br>
 +
 +
<p>The 3 strategies overall aims to optimally shortened the overall patient's length-of-stay across the board. </p>
 +
<b>SCON Strategy (Shortest Consultation Time first)</b><br>
 +
 +
<p>SCON seeks to arrange the queue dynamically by putting the patients with the shortest consultation duration to the front of the queue.
 +
Each patient's consultation duration is derived from an exponential distribution, with the ' lambda' being the doctor service rate. The numerical value is generated by using the random  'seed'  used for consultation. Each computation will be triggered by an 'Event"  (eg,  StartDoctorConsult).</p>
 +
 +
If the consultation is a review consultation (consultation with the doctor after being sent for investigative test/treatments), the consultation duration will be based on the tests that the Patient is sent to and the Probability of the seriousness of test result that is generated at random. The numerical value will be derived from the sum of exponential distribution with the mean as the average time assigned due to the probability of seriousness of test result.
 +
 +
The numerical value for priority of each generated 'Event' is then sorted by the following method, arranging all the patients with high priority(low consultation duration) to the front of the queue.
 +
 +
<b>SREM (Shortest Remaining Time first)</b><br>
 +
 +
SREM strategy would first retrieve the "Length-Of-Stay" duration that is entered by the user at the simulation page. It then computes the remaining time left for each using several other parameters such as the "Simulation elapsed time" and "Patient Start Journey Time".
 +
With the computed remaining time, a priority is generated for the patients. Patient with a relatively larger negative remaining time value will be assigned a higher numerical priority value while a Patient with a positive remaining time value will be assigned a lower numerical priority value.
 +
Likewise as the SCON strategy, the patients will be rearranged dynamically within the queue based on priority value with the following method,
 +
 +
<b>MIXED Strategy</b><br>
 +
 +
The mixed strategy computes the priority based on sum of weighted SCON and SREM strategy. The optimal weights to be assigned to both strategies is derived from a local search algorithm.
 +
 +
<div style="background: #000066; padding: 12px; font-weight: bold; font-size: 100%; line-height: 0.5em;"><font face="Arial" color="white">Existing Software tools?</font></div>
 +
 +
There are popular animation tools such as Arena and ExtendSim in the market.However,the drawback with exsiting ones are the static queue priority system inbuilt in them.This limited function only allows users to set objects priority at the start and only once.
 +
 +
As much our strategies are concerned,the queues have to be re-adjusted every time a re-entrant patient joins back in.Because of the fact that queues would not be able handle the dynamic side of it. Hence, it is not viable to use a commercial simulation software for this particular simulation and we decided to built a simulator from scratch.
 +
 +
== <div style="background: #000066; padding: 12px; font-weight: bold; font-size: 60%; line-height: 0.5em;"><font face="Arial" color="white">References</font></div> ==
 +
Papers referenced:
 +
<ul>
 +
<li>Kar Way Tan (2013) Dynamic Queue Management for Hospital Emergency Room Service.</li>
 +
<li>D. J. Medeiros, Eric Swenson & Christopher DeFlitch (2008) Improving Patient Flow in a Hospital Emergency Department, Proceedings of the 2008 Winter Simulation Conference.</li>
 +
</ul>
 +
 +
Images in this page taken from:
 +
<ul>
 +
<li>http://www.softicons.com/social-media-icons/blue-jelly-social-icons-by-webtreatsetc/aim-icon</li>
 +
<li>http://www.straitstimes.com/sites/straitstimes.com</li>
 +
<li>http://www.clker.com/clipart-crosshairs.html</li>
 +
</ul>

Latest revision as of 22:22, 20 November 2014

Code Blue 1.jpg
Home The Team Project Overview Project Management Project Documentation Project Resources
Project Description Project Scope Technologies X-factor


Goal & Motivation
Aim-icon.png The goal of this project is to develop a simulation tool which provides a graphical visualization of the dynamic queue management framework which can be implemented in the hospitals’ Emergency Department (ED). The framework addresses the complex challenges faced by hospital to achieve a desired service level for the patients (e.g. LOS of 90% of patients must be within x minutes). Focused on managing patient queue dynamically before doctor consultation, the project shall implement the dynamic patient-prioritization strategies. The strategies make use of several greedy algorithms such as Shortest-Consultation-Time-First (SCON) or Shortest-Remaining-Time-First (SREM) to improve on patients’ length of stay (LOS) in the ED.

The aim of this tool is firstly to allow healthcare practitioners to better understand and visualize the mechanism and effects of the proposed strategies; secondly to appreciate how existing data in the database can aid automation and improve operation; and finally to allow visual interaction with a simulation tool.

The project will focus on the demand perspective of the integrated framework.


Ed-waiting-time.png Hospitals are facing increasing challenges in today's emergency department due to growth in patients demand for services and limited capacity in resource allocation. (D. J. Medeiros, Eric Swenson & Christopher DeFlitch, 2008). Long waiting times and a lack of doctors are often observed and experienced.

In the existing literature, addressing the issue of long waiting times in an ED often takes the form of single-faceted queue management strategies that are either from a demand perspective or from a supply perspective. From the demand perspective, there is work on queue design such as priority queues, or queue control strategies such as a fast-track system and demand restriction through ambulance diversion. However, they may not sufficiently leverage insights that can be derived from both historical and real-time data (Kar Way Tan, 2013).

Therefore, an integrated framework that manages queues dynamically in the ED from both the demand and supply perspectives by leveraging historical data and real-time data is being proposed by our sponsor, Assistant Professor (Practice) Kar Way.


Scope-icon.png The high level scope of the project comprises of the following module:
  • Simulation Module
  • The simulation model will be based on a M/M/S queueing theory. The time varying hourly arrival rate of the patients will be following a Poisson process with a lambda that is defined in the data file. The service rate will follow a exponential distribution with time varying service rate that will likewise be defined in the data file. The number of servers will be capped at 5. There will be 2 assumptions to this simulation model - No Balking and No Reneging. No Balking: Patients left before being in the queue. (eg Patients came to the hospital and noticed that the queue is long and left without registering) No Reneging: Patients registered, but somewhere along the process, left the hospital without being seen.
  • Reporting Module

For a more detailed scope, please refer to our project scope.

Process in the Emergency Department

Codeblue ed process.png

To set the context, it is important to understand the process that is happening in the Emergency Department of a hospital.

Patients will first make the necessary registration upon their visit and an initial assessment at the triage. Subsequently, they would proceed to queue for the consultation with the doctor.The doctor would determine if further tests and treatment is needed for the patient after the initial consultation. If not, the patient is immediately discharged or admitted into one of the wards. If they do need additional treatments, they would receive the necessary treatments and rejoin the queue to consult the doctor for a final review before being discharged/admitted.



How do this 3 strategies work?


The 3 strategies overall aims to optimally shortened the overall patient's length-of-stay across the board.

SCON Strategy (Shortest Consultation Time first)

SCON seeks to arrange the queue dynamically by putting the patients with the shortest consultation duration to the front of the queue. Each patient's consultation duration is derived from an exponential distribution, with the ' lambda' being the doctor service rate. The numerical value is generated by using the random 'seed' used for consultation. Each computation will be triggered by an 'Event" (eg, StartDoctorConsult).

If the consultation is a review consultation (consultation with the doctor after being sent for investigative test/treatments), the consultation duration will be based on the tests that the Patient is sent to and the Probability of the seriousness of test result that is generated at random. The numerical value will be derived from the sum of exponential distribution with the mean as the average time assigned due to the probability of seriousness of test result.

The numerical value for priority of each generated 'Event' is then sorted by the following method, arranging all the patients with high priority(low consultation duration) to the front of the queue.

SREM (Shortest Remaining Time first)

SREM strategy would first retrieve the "Length-Of-Stay" duration that is entered by the user at the simulation page. It then computes the remaining time left for each using several other parameters such as the "Simulation elapsed time" and "Patient Start Journey Time". With the computed remaining time, a priority is generated for the patients. Patient with a relatively larger negative remaining time value will be assigned a higher numerical priority value while a Patient with a positive remaining time value will be assigned a lower numerical priority value. Likewise as the SCON strategy, the patients will be rearranged dynamically within the queue based on priority value with the following method,

MIXED Strategy

The mixed strategy computes the priority based on sum of weighted SCON and SREM strategy. The optimal weights to be assigned to both strategies is derived from a local search algorithm.

Existing Software tools?

There are popular animation tools such as Arena and ExtendSim in the market.However,the drawback with exsiting ones are the static queue priority system inbuilt in them.This limited function only allows users to set objects priority at the start and only once.

As much our strategies are concerned,the queues have to be re-adjusted every time a re-entrant patient joins back in.Because of the fact that queues would not be able handle the dynamic side of it. Hence, it is not viable to use a commercial simulation software for this particular simulation and we decided to built a simulator from scratch.

References

Papers referenced:

  • Kar Way Tan (2013) Dynamic Queue Management for Hospital Emergency Room Service.
  • D. J. Medeiros, Eric Swenson & Christopher DeFlitch (2008) Improving Patient Flow in a Hospital Emergency Department, Proceedings of the 2008 Winter Simulation Conference.

Images in this page taken from: