Balancing Out the Peaks and Valleys: How Stanford Health Care Infusion Centers Used Data Science to Overcome Scheduling Complexity

Oct. 5, 2016
A common challenge with healthcare in general, and cancer care in particular, is the peaks and valleys in patient appointment scheduling, which often results in inefficient use of resources, unpredictable wait times for patients and overtime hours for staff.

A common challenge with healthcare in general, and cancer care in particular, is the peaks and valleys in patient appointment scheduling, which often results in inefficient use of scarce and expensive resources, unpredictable wait times for patients and overtime hours for staff.

Leadership at Stanford Cancer Center, a part of Stanford Health Care, faced these scheduling complexities in the operations of its Infusion Centers. The infusion treatment units perform more than 65,000 infusions annually and that number is growing steadily.

According to Sridhar Seshadri, vice president of Cancer Services at Stanford Health Care, patients were facing long wait times in the middle of the day, between 11 am and 2pm and the infusion chairs were underutilized during other times of the day. 

Sridhar Seshadri

 “There is a classic healthcare phenomenon where there is a valley in the early part of the week and then again in the later part of the week and patients build up in the middle of the week, and then you see the same pattern happening during the day, when it’s slow in the early morning and evening, and patient volume peaks in the afternoon. You will see this in most production flows in healthcare, so basically areas such as chemotherapy labs, radiation therapy and the operating room. And, we were seeing this same problem in our infusion centers,” Seshadri says. 

To complicate matters, at Stanford Cancer Centers there are at least five types of infusion appointments depending on what chemotherapy treatments patients need – one hour, two hours, three to four hours, six to eight hours and nine-plus hours. And each appointment slot has a different daily volume of patients with a maximum of four patients that can start treatment at any given appointment time, and this results in a total of 256 possible appointment slots each day.

The mathematical complexity of balancing that complicated scheduling results in a staggeringly large solution set, much too complex for schedulers to calculate, according to Seshadri, who has an  engineering background with advanced engineering degrees and experience running Healthcare Solutions for General Electric’s Medical Systems.

The classic answer to this problem is patient scheduling done on an ad hoc basis, with patients slotted into whatever appointment time is open or requested, yet this leads to suboptimal sequencing of appointments because it doesn’t take into account the patient volume and mix levels, staff scheduling and equipment availability.

“So you end up with this bad racking and stacking of appointments, with long and complicated appointments put back to back, and if there is a complication, then it disrupts the flow for that infusion chair for the rest of the day,” he says.

With his engineering background, including an expertise in Six Sigma/Lean business principles, Seshadri says he began looking at the problem from an operations management standpoint. Seshadri says he was familiar with LeanTaaS, a healthcare software-as-a-service (SaaS) provider and the company’s founder Mohan Giridharadas as Stanford Health Care has been using various products from LeanTaaS since 2010.

“Mohan has a lean background and he and I got to talking and out of this we had some ideas, born on the back of a napkin, so to speak, about using mathematics to solve the problem,” Seshadri says. As a result, Stanford Cancer Center began beta-testing LeanTaaS’ iQueue data analytics software in 2014.

iQueue was designed to solve scheduling and operational performance problems using data science and optimization algorithms, according to Giridharadas.

Mohan Giridharadas

Since adopting the iQueue data analytics, Stanford Cancer Center has reduced patient wait times by 30 percent at peak times. At the same time, the infusion centers have been able to see 25 percent more patients with 15 percent lower costs, according to Giridharadas.

Giridharadas says the iQueue healthcare suite uses data science and predictive analytics to optimize the infusion chair usage, which creates a more balanced patient appointment schedule as well as a balanced staffing schedule. LeanTaaS worked with Stanford’s chemotherapy leadership, including Seshadri and Donna Healy, director of clinical operations at the Cancer Center, to deploy the analytics platform and then eventual integration with their existing electronic health record (EHR).

“We took two years of historical data and pumped that into the analytic engine as well as operating constraints, how many infusion chairs are available, the hours when the chairs are open and the staff that’s available and that’s translated into mathematical equations into iQueue and out comes as a production schedule,” Seshadri says.

Recognizing the operational challenges in healthcare, Giridharadas, who has a background in manufacturing and service operations, says he decided to use mathematics and lean thinking to tackle these challenges by founding the company five years ago and then developing iQueue for healthcare organizations.  

Comparing each patient to a card in a deck of cards, Giridharadas says iQueue mathematically shuffles the deck so patients are filled up in the right order. “So the magic that this unlocks is that wait times come down, you can see more patients, your unit costs go down and nurses are satisfied because the work flow is more flat,” he says.

Along with Stanford Health Care, many other healthcare organizations, such as UCSF’s Helen Diller Family Comprehensive Cancer Center, the University of Colorado Hospital and Wake Forest, have field-tested the software as well. Giridharadas says other healthcare organizations have seen similar increases in patient volume and reductions in patient wait times as Stanford while also seeing 50 to 70 percent reductions in staff overtime and 15 to 20 percent reductions in the unit cost of service.

“I’ve talked with more than 50 infusion centers and they all have the same two problems: patients wait a long time in the middle of the day and chairs in the infusion centers are underutilized in the morning and at night,” he says.

With the success of the iQueue data analytics platform at its infusion centers, Stanford Health Care is now focused on making similar improvements to scheduling optimization in other areas of its operations.

“We’re using the same iQueue platform algorithm to look at optimizing clinic visits, as we do about 110,000 visits a year, so that’s our second journey. And if that works out, we can then take it to look at in-patient flow, or perhaps the operating room. So this is a progression beyond just the infusion centers,” Seshadri says.

The iQueue for Infusion Centers platform is actually the first in the LeanTaaS iQueue Suite, as iQueue for Clinics is currently being field tested and iQueue for Surgery is in active development.

LeanTaaS iQueue case study results before implementation

LeanTaaS iQueue caes study results after implementation