Solving the scheduling problem

One of the things I want Ankhos to accomplish is to simplify appointment scheduling. I’ve got some free time during a down-period of our launch so I’m writing this blog post as a brainstorm session for myself.

There are currently four fundamental goals that we are any scheduling system is trying to solve:

1. Be able to tell the patient when to arrive

2. Be able to spread patient load throughout or across workdays

3. Be able to show employees where time-critical workflow elements (busy times) are in their day to help department administrators plan their workday. For example, It would be nice for the chemo nurses to know what drugs they need for the day and how much.

4. Keep patient waiting time to a minimum.

These are probably no-brainers, but stating them explicitly helps the thought process.  In my math and computer science training, we studied complex algorithms and thought of ways to make them either faster, cheaper or less complex (Hint: You can only pick two). I want to try this approach to scheduling, also…

Think of  “Appointment Scheduling” as an algorithm… one that has likely been well-published and analyzed to death.  In computer science, a lot of major breakthroughs are not made by bashing your head against a problem, but by looking at it in a different light, or by noticing that we aren’t trying to solve the “Generic Scheduling Problem”, but the “Clinical Oncology Scheduling Problem”.

Well, what’s different about Oncology scheduling than general scheduling? What information do we have that we can use to simplify the problem? to make it cheaper, faster and less complex? Can we have all three if we think of it as an easier problem?

With general scheduling, a doctor has an opening and the appointment is made. At COS, the patient gets blood drawn at the in-clinic lab every time they come in. Some may also see the doctor, and some may also get treatment. That’s already three things to schedule simultaneously. Don’t forget that there are multiple doctors whose schedules may vary; whose hospital rounds may run late.

Currently, the scheduler sits with the patient and enters three appointments (due to limitations of current software). This is very awkward and time-consuming. I studied the schedulers and counted the clicks required to schedule a patient. It takes nearly 30 clicks to schedule a patient for all three phases of their appointment.

Sliders:

One potential solution we’ve come up with uses a ‘slider’ widget, four of them to be specific (I’ll post a concept screenshot when the idea is more developed). The first three sliders will each determine the length of each phase of the appointment, the lab, exam and treatment phases.

The fourth slider will determine the patient arrival time. Below the sliders will be a series of graphs, showing population of each area during the day vs appointment times.

As the sliders change, the graphs below will change to reflect the new values.  The goal is simple: Keep the graphs level and keep the height of the graphs below the maximum capacity of each department.

In four clicks, the scheduler would be able to completely lay out and visually assess the effect of scheduling a patient at a certain time and duration.

Waiting room:

Another concern with scheduling is latency of patient flow. The latency manifests itself in the waiting room.  This idea is un-proven, but perhaps if we embrace the waiting room time by adding it to the appointment schedule, we can more accurrately and flexibly determine patient flow.

So now we have 6 sliders, three for each of the lab, exam and treatment phases, two waiting room times sandwiched between these and a final slider to determine the appointment start time.

Will this approach be simpler and easier than the current scheduling software? Only testing will tell.

Side note:

By embracing waiting room times, we might even be able to tell a patient what their waiting time during the day is. Maybe we can more accurrately tell when the end of an appointment will occur if rides need to be scheduled or lunch needs to be had between appointment phases.

Tags: , , , , , ,

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s


%d bloggers like this: