Like many clever ideas, RosterLab started as a way of helping someone before morphing into a fully-fledged tool – in this case, nursing rosters.
Isaac Cleland was working on his PhD in AI at the University of Auckland and helping his professor’s colleague create a roster. His approach was good, because it went on to make Cleland, Sunny Feng and Daniel Ge winners of the University of Auckland Velocity Challenge in 2020. Velocity described RosterLab as using “state-of-the-art artificial intelligence techniques to build exceptional rosters for hospital staff, much more quickly than is possible by hand. With a simple-to-use web application for roster creation and associated staff phone app, RosterLab can automate almost all of the rostering process”.
Rosters are like a 90x90 Sudoku…. time-consuming with so many daily complexities – trade-offs unpredictable demand drivers, weekly requirements for variable start and finish times, rotating versus fixed patterns, aligning staff resources to the level of patient care required, staff requests and availabilities, union regulations, nights, sleep periods for those doing night….the list goes on. And, more recently, rosters have had to cope with the prolonged impact of COVID-19 on staffing across hospitals and aged care.
Mentoring provided by the University’s Centre for Innovation and Entrepreneurship Convinced the team to continue pursuing AI to create better rosters. From this, Cleland, Feng and Ge formed RosterLab.
“We began by talking with charge nurses about the rostering tools they used. It was either pen and paper, or a spreadsheet. It took days and just one change meant they would have to redo the roster,” says Ge, co-founder and Head of Customer Experience for RosterLab.
Then there’s the challenge of creating a roster that meets every criterion against real world constraints. “Bad rosters lead to burnout, excess overtime, and resentment among the nurses and healthcare assistants, as some suffer from poor rostering more than others,” says Ge.
Rosterlab can now generate rosters for up to 300 nursing staff or up to 16 weeks in duration. “Those rosters take into account all the formal element such as number of staff needed, skills coverage, shifts, rules, history, leave information, and staff preferences, “says Feng.
RosterLab stores the data, so an unexpected request for leave means a new roster can be generated within five to 15 minutes.
Ge says RosterLab is designed to make the process as easy as possible. “We can show roster makers how to create rules and guidelines and then how to change them. For example, a rule might be the maximum number of nights in a row a nurse can work or smoothing out days when a facility is overstaffed or understaffed.”
There are also challenges around the use of locum nurses, retaining staff and the cost of replacing a nurse, including lost time on the ward and administration.
RosterLab is now in use around New Zealand, from a health clinic in Dargaville and aged care facilities to users across several district health boards. Over the next 12 months, the team hopes to expand in the New Zealand market to help improve the health system and staff shortages.
By the way, Cleland says you can use their app to solve any Sudoku puzzle. You have nine shifts, numbered 1-9, with nine employees over a nine-day roster. And you can model all the sudoku rules using RosterLab’s natural language rule builder.
Want to find out more? https://www.rosterlab.com/products