Schedule for MDs and separate patient stream

[vinayak_n2]

[ FlexSim 21.1.2 ]

test.fsm

Hi Team,
There are two things I want to implement in my model and I'm not sure what is the best way to go about it. Could you give me some direction?

1) I have two MDs (MD1 and MD2) and I want to give them two separate working schedules (example MD1 = Monday and Tuesday 8 AM to 5 PM, MD2 = Tuesday and Friday 9 AM to 6 PM)

2) I want to create separate patient streams for these two MDs who show up for appointments of variable duration only when the MD is available.



@Matthew Gillespie

[eric_m3 · Accepted solution]

Hi @Vinayak N2, for the first question, you will want to create a time table for each doctor. There's a tutorial that goes over how it works and how to set it up.

For the second question, you can create two different patient flows, one for each doctor. Then you can set the schedule of the sources to create tokens during the times you know the doctor is working. The variability in time is something you can control in the "Process" activity in process flow.

test-em.fsm

[vinayak_n2]

test-em.fsm

Thanks Eric! Great ideas!

For the first question, I was able to create the 'Time tables'. Thanks for the tutorial.

For the second question, I'm trying to model it from a capacity allocation point of view.

For example, for MD1's schedule below, for Tuesday, I want to allocate patients to the schedule only for 3 hours between 9 to 12. It could be 3 patients taking 1 hour each or 6 patients taking 0.5 hours each. I can assign the duration of appointment using a distribution via the process time field. But is there any way to model the arrivals based on the time table?

[eric_m3]

I don't know of a way to directly change the arrivals based on a time table. You can always just make the arrivals schedule follow the same hours as the time table for the doctor. The other option I can think of is to include a decide activity before the patient is created to see if the doctor is working or not. If they are off schedule, the token goes to a sink. If they are on schedule, the token proceeds to create a patient. That's not directly manipulating the arrivals, but it is filtering them out when the doctor is not working. I did this for MD1 so you can see how it might work.

test-em.fsm

[vinayak_n2]

Thanks Eric, that's a great idea too, but I thought of another direction that I think is more representative of how I want my model to behave.

I want to create a list with number of slots (which varies based on the weekday) available for patients to book from. And patients should be able to book slots (1 slot or 2 slots based on a percentage) if they are available in the list and arrive at the clinic until the number of slots for the day are completely booked. If no slots are available, then they should book slots on the next day's list of slots.

A) I created a global table called 'Slotsperweekday' to create slots into the list. But I am not able to push the number of slots to the list based on the Day of the model run time. Can you help me with that?



B) I then went to create an arrival source that creates the demand per day based on a distribution and pulls these from the list and processes. I'm struggling with choosing an appropriate source object to do this. And if the daily demand exceeds the number of slots available, I want the demand (or patient) to carry over to the next day. Any ideas or examples with regard to this would be super helpful. test3.fsm

[eric_m3]

Great, thanks for the added explanation. I've added a bit to your model that should do what you're wanting. The general idea is that a token is created for each time slot every day (repeating weekly). That token will have a specific day and time assigned to it that can be pushed to a list. Tokens are also created for patients that can pull from the list (either one or two slots). The token is delayed until the scheduled time at which point it moves on to create a patient in the model. If there are more tokens created in one day than slots available, the token will pull a value from the next day.

I gave the slot token some labels to keep track of things. They include the day in the week (1-7: based on the order of the global table), total day since the model was running (starting at 1 and incrementing each day), and the time slot that it will fall in (hours) both by total model time and time of day. Total model time is used to sort the list and to make sure the patient isn't created until the delay is up. Time of day is just used as a reference to see what time in the day that patient is assigned (ex: 9-5, this label has no logic dependent on it, I just thought it was convenient when looking at the values on the list). The push to list activity has a timer in case a slot is never scheduled. That way those time slots are pulled off the list and you can keep track of how many appointments are going unscheduled.

For the source creating the patients, I decided to use a Date Time Source. This allows you to have a different distribution of patients for each day of the week. Right now, the distribution is a uniform distribution with a lower bound of half the slots and an upper bound of the total slots for each day. You can change this to whatever distribution you'd like.

I made some notes in the process flow of things to be aware of that the logic uses to work (in case you make changes). You might want to create a model that uses hours since the seconds can get big fast. I would recommend that, but just be aware that you would need to change anything referencing time to be consistent with the time unit change. Let me know if you have any questions.

test3-em_1.fsm

[vinayak_n2]

Thank you, @Eric M! This is exactly what I wanted to do!

One question:

Could you explain what the value in the slotDuration column would return? To me it looks like it is referencing the same cell?

[eric_m3]

Yes, the SlotDuration was made to determine the length of each appointment. You could have just put 30 minutes, but I thought I'd make it semi-dynamic in case you change the slots for a certain day. It just calculates the number of working seconds in that day divided by the number of slots in that day. So the whole denominator is just a reference to the Slots column in the table.

[vinayak_n2]

Thank you, Eric! That makes perfect sense. I actually confused myself because while I was looking at the table, the first column was hidden from view.

[vinayak_n2]

Hi Eric, one more question if I may,

I want to use this flow for multiple MDs with their own slots and patient arrivals.

So, I split the Patient source and MD)_Days source into two and gave them two separate labels MDnumber and MD (which equals to 1 or 2). I want to pull the slots such that patient arrivals with MDnumber = 1 pulls only slots with respect to slots with MD =1. Could you show me how to do this? Is this the correct approach?

@Eric M test3-em-1.fsm

[eric_m3]

You could consolidate it all into one block like what you have above, but it may be simpler to copy the entire block for each new MD. The reason being they will each have their own list and their own table (I'm assuming). I would just copy the whole process flow, create a new table and then update references to the table, list, and Patient Flow the patient is created in. Also, you could probably get rid of the "Assign Labels: SlotsToPull" since that sounds like it will now be MDnumber. You can then update the Pull from List activity to reference that label for the quantity to pull.

If you're planning on doing this with a lot of MDs, I would assign a label at the beginning of the block referencing that MD's table and replace all references to that table with the new label. That way when you copy the flow for a new MD, you only need to update the assign labels one time to the new table instead of looking through each activity and making sure it's updated to what it should be.

[vinayak_n2]

oh yeah, that makes sense! It's definitely cleaner that way. Thanks.