Mix and match: Powerful software takes the hassle out of scheduling
December 2000 Mark Uehling
Suppose your lab runs 20 workstations and employs 25 employees
to operate them. How many possible schedules could properly match
up the people to the machines?
The answer will be obvious only to a math wizard: 1023 possible schedules. Put another way, there
are countless ways to fill the 20 slots in the lab from a list of
25 techs. Given the complexity of the job, it’s amazing mere mortals
have managed scheduling as well as they have.
Of course, no math wizard would know that Bob works nights or
that Susan is the lead technologist doing coagulation. Any good
laboratory supervisor would know that and draw up a schedule accordingly,
taking still more time to tinker with the lineup when someone calls
in with a soccer game to attend or a sick cat to take to the vet.
Now there’s a better way. It combines raw computational power
and nuanced human insight into a lab’s workforce. The program is
called Lab Scheduler. It’s Windows software that knows most of what
any supervisor does about the strengths of each technologist and
technician on the roster. Lab Scheduler scans tens or hundreds of
thousands of so-so schedules. In a few minutes, all but one of the
theoretically possible schedules have been cast aside.
That one remaining schedule has survived a literally Darwinian
selection process inside the computer. The software was designed
by an academic pathologist and crack programmer at the University
of Virginia. Then it was licensed and retooled by a small Virginia
company called Medical Automation Systems, or MAS.
Judging from a few testimonials, the silicon-based schedules are
uncannily good. Perhaps Lab Scheduler’s most appealing trick is
that it automatically rotates staff through their workstations often
enough to maintain their competencies-but not so fast they are not
in their most crucial spot when they need to be.
For people like Ed Bush, chief operating officer of Medex Regional
Laboratories, Bristol, Tenn., the Lab Scheduler package has been
a lifesaver. He guesses it will save half an FTE, or between $15,000
and $25,000 annually, once it’s fully implemented. "I’m convinced
it’s the slickest thing since sliced bread," he says. "I get excited
about it because it will help us save significant time not only
in scheduling but in staff utilization."
Scheduling is always hard, he concedes, but it has become especially
difficult in the era of TN-Care, a statewide managed care plan,
which decimated his head count. "We lost 16 FTEs in two weeks, affecting
25 people. We reduced staffing from 84 to 59. It was awful," Bush
recalls of the period in 1994. Today, as a newly formed consolidated
reference lab with more than 300 employees, "the software will become
even more important," Bush says.
Filling out his second and third shifts, he reports, was agonizing.
Now, with Lab Scheduler, major or minor staffing glitches are easier
to deal with. "If someone calls in sick, you can double click on that
workstation, and it lists all the employees who have skills on that
station. Then you can do an employee profile for availability."
In rotating technologists to maintain their competencies, Bush and
other Lab Scheduler customers quickly see a double payoff. They
no longer have to think so hard about scheduling issues. And the
software’s record-keeping and report-generating functions can help
create the paperwork to expedite inspections and accreditations
that are related to maintaining employee competencies.
Some of the other benefits are less tangible but equally intriguing.
Lab Scheduler’s output can be tweaked to reflect Susan’s and Bob’s
departmental preferences. But the larger idea is that the software
can put an end to some of the bickering or complaining about one
employee getting preferential treatment over another.
Says Bush: "It is almost impossible to be fair during the scheduling
process. The program improves efficiencies and lets us treat everyone
equally unfairly. It removes favoritism: ’I like Susie so I’ll work
her here and I’ll give the odd shifts to this other person.’
Some of that [personal chemistry] may go back 20 to 25 years."
Indirectly, Bush says, Lab Scheduler addresses his concern about
the demographic trends in laboratories: "We’re all getting older.
It’s frightening to look at where we’re going as an industry. Our
applicant pool is drying up. We have to use the available resources
in the most efficient way."
The story at the Mayo Clinic is similar. More than 70 technologists
and technicians are being scheduled by Lab Scheduler in Rochester.
Most of the lab workers are scheduled by a few lead techs, one of
whom is Janice Krahn, lead technician at the Mayo Clinic’s hospital
clinic laboratory.
Krahn is as pleased as Bush. The software knows which techs prefer
to work at which of four geographical locations, and which techs
rotate through all of the sites. "A one-month schedule is posted
at all times, so everyone can check on their lab assignment well
in advance, which avoids a drive across town at the last minute,"
says Krahn.
"I like it a lot," she says of the software. "I think it’s better
[than a human scheduler]. You don’t have the factor of people coming
to me and complaining that ’I’m always over there.’" Krahn’s
computer skills may not be robust enough to qualify her for a job
at IBM, but she says that doesn’t matter: With one click of her
mouse, using Lab Scheduler’s AutoLink function, she can generate
a schedule. Then when her practiced eye discerns a few minor problems, she can
just as easily improve on it. "We’re big on our head count. Our minimum
on the day shift is 12. If we’re down to 11, I get a flag. It’ll be
a different color and tell me what area needs a different person."
Soon, Krahn reports, the software will also be connected to the
payroll system. That means administrators will be able to verify
overtime, time off, and hours worked. "Our management staff will
be able to verify employee timecards. It will be a lot faster for
them," Krahn says, "as far as total hours worked and so on, as long
as the leads keep it updated."
Her only complaint-and she’s not alone-is that while the schedule
on the screen is in color, the printed versions are not. Lab Scheduler’s
product manager, Kathryn Kelly, says color printing is something
that MAS programmers are working on.
The Charlottesville, Va., company, founded in 1994 and probably
best known for RALS, its point-of-care product, has been assiduously
incorporating suggestions from the real world. A newcomer to the
lab scene, Kelly has personally demonstrated or sent Lab Scheduler
to some 700 laboratories.
First impressions are positive. "Most reactions are incredible,"
says Kelly. She notes "people tend to think that I have been doing
laboratory scheduling myself for many years." Sample reactions from
potential customers: "I have been looking for this product for over
10 years" and "I can breathe again."
As with any innovation, however, there can be resistance to closing
the sale. So far, 10 separate laboratory sites have purchased Lab
Scheduler. Twenty-nine users have been trained to operate the program
at those sites. An additional 20 labs are evaluating the program
as part of trials in which the company sets up Lab Scheduler using
phone lines or the Internet. At least 1,200 techs nationwide are
being scheduled with the program.
For all the power of Lab Scheduler, Kelly graciously points out
that the Internet has any number of do-it-yourself scheduling Web
sites. Such inexpensive choices may well suffice for small labs
with 10 or fewer techs, Kelly says. But she notes that the plain-vanilla
online schedulers are not designed for laboratories.
Her product’s ability to rotate employees to maintain their competencies
is far beyond some hypothetical Web site like any.old.calendar.com.
"Maintaining the proficiencies alone points you toward Lab Scheduler
rather than a generic scheduling program," says Kelly. "It’s maintaining
the skills, it’s taking the bias out of the schedule, it’s the best
fit for the laboratory, it’s the reporting features."
One of the big selling points, it turns out, is the variety of documents
the program can generate. Says Kelly: "Some people want to know more
about the reports than about the scheduling." Some customers, she
reports, tell her "’all of the information I want is right there on
the screen. Make it print out.’ You can do it for an individual, for
a department, for the whole lab. Or just totals. By workstation, by
employee."
With half a dozen staffers working on Lab Scheduler,
Kelly reports, MAS has been able to mine the experiences of early
customers to improve the program. As an example, she cites phlebotomy
routes. In the early days Lab Scheduler would send someone who returned
at 4:00 PM back out the door to do another route until 6:00 PM.
"We decided there must be a way for us to ensure there is only one
physical location per employee per day." Done.
Another feature in the works is the ability to run the program
over the Internet. That could be a payoff for laboratories because
the software could be updated automatically. And it might reduce
informational technology costs, since MAS would be maintaining a
lab’s scheduling data on its own computers in Virginia.
By the first day of 2001, several customers will operate Lab Scheduler
over the Internet. These labs’ schedules will not be stored on site
but on Lab Scheduler’s well-protected, backed-up computers. Many
new customers, like Walter Reed Army Medical Center, will access
their schedules over long distances-and be spared the hassle of
maintaining a computer for scheduling. That should allow MAS to
troubleshoot the schedule more quickly.
As for security or privacy risks? Few customers worry about that.
The company takes all the standard precautions. "Once they realize
there is no patient data, people tend to relax quite a bit," Kelly
says of the Web-enabled version of Lab Scheduler.
In general, Kelly says, the company’s goal has been to let the
program adapt to the customer-not the other way around. At one sales
call in a northerly U.S. location, she asked those assembled for
a reason that a technologist might not show up for work. Someone
shouted: "Ice fishing."
No problem, said Kelly. "We put that in the program as a reason
people were out. It shows up on the report as ’ice fishing.’ Lab
Scheduler is user definable and people love that. We don’t want
people to have to change how they run their laboratory. We just
want to enhance what they’ve got."
The company spends a fair amount of time examining detailed setup
information supplied by the laboratory. That includes basic personnel
information about the competencies of each technologist, technician,
and supervisor, and the types of workstations and existing rotations.
Almost every item in the program-every person, skill, and machine-gets
an easily changed numerical value, a weighting of its urgency or
priority, that helps the computer create the schedule.
MAS then tries to examine the information for minor conflicts
or illogic that might make Lab Scheduler’s output less than perfect.
"We try to figure out-hey, what’s this about," says Kelly. "Do you
really want that weekend workstation to be operating on a Tuesday?"
After the basic issues are ironed out, MAS sends out a team to
instruct the laboratory in question. The length of the training
varies, but typically one to three people need to be taught the
program over a period of four days.
For all its technical complexity, much of the program’s attraction
has to do with its ability to handle the softer side of scheduling.
"The human chemistry is a big part of it," Kelly concedes. "That’s
why we need these human resources features in the program. The program
will optimize the schedule for the laboratory. But when Susie complains
to Ann that she’s near Bob again, you can manually change anything.
You’re not stuck with this automated outcome." As time passes, Kelly says she’s more aware of the importance of keeping
not only technologists and technicians but also supervisors as stress-free
as possible. She recounts a lab that calculated the cost of finding
a new supervisor at $7,000 before the first paycheck is ever cut.
Which brings up the cost of Lab Scheduler.
There are two bills. The first covers entering the right information
into the program’s database and training key employees. Those setup
fees are $3,500-$6,000.
The second Lab Scheduler fee is an annual licensing fee. That
number is based on the total number of warm bodies (full time, part
time, per diem, supervisory) being scheduled. The annual fee is
calculated in tiers, with bigger labs paying a lower fee per scheduled
employee than smaller labs.
The company might charge $500 a year for a lab with 10 techs,
or $7,000 for a laboratory of 100 techs. To schedule 250 people,
MAS might charge $9,000 annually. As steep as those numbers might
seem, Kelly says they are half of the cost of doing a schedule manually.
As Kelly tallies it up, a laboratory that schedules 30 techs may
spend $3,456 annually if the time of a scheduler is valued at $24
per hour and the task takes 144 hours a year, or 12 hours a month.
In contrast, Kelly notes, MAS might charge $1,800 to use Lab Scheduler
for a lab with 30 employees.
One unanticipated bonus is that the program can provide just as
much data about nonbiological resources in the labs. That’s because
the program also monitors workstations, the lapses of which can
be tracked as exhaustively as that of any human.
As Kelly explains: "If you have a workstation that is going down
quite a bit, you can track that. You can print out a report and
show every hour that it’s been down. It’s about managing resources."
It’s a tool, then, to be used when negotiating with vendors in the
future.
What Lab Scheduler is intended to bring to the party, basically,
is a reduction in the daily level of frustration that accompanies
scheduling. Says Kelly: "Most of it is saving time, saving money,
but a lot of it is saving frustration. It’s ’take this bull’s eye
off my chest.’ Let the computer do it."
The man who figured out how to let the computer do it is
pathologist James Boyd, MD, of the University of Virginia. Dr. Boyd
is associate professor, director of laboratory, systems engineering,
and specimen support services, and associate director of clinical
chemistry and toxicology.
One of his quasi-official roles in his department is to advise
colleagues on statistical and technology concerns. He majored in
applied mathematics, which may explain why-upon hearing of the scheduling
complexities-he took it upon himself to write a program to resolve
them.
"I do a lot of programming in SAS, statistical analysis system
programming language," he says, "and do a fair number of statistical
analyses for various colleagues. It’s always been a big part of
what I do." So when colleague John Savory, PhD, briefed him on the Virginia lab’s
scheduling headaches, Dr. Boyd was intrigued. He toyed with simple
schedulers, but none could properly rotate the techs to maintain their
competencies. As Dr. Boyd recounts, he was briefly stymied.
But then he attended a medical conference and learned about genetic
algorithms. These computational engines randomly generate potential
"legal" solutions to a problem like how to minimize the distance
a sales executive must travel on his or her rounds. Each such solution,
in a metaphorical sense, can be thought of as an "offspring" or
"organism." Then, generating new offspring, the program uses mathematic
analogs of Darwinian selection principles to pit the organisms against
each other.
After untold eons of evolution inside the computer, only the fittest-like
hardy rats or humble bacteria-survive. Rather than duplicate a previous
schedule, Dr. Boyd realized, he could tell the computer to slog
through tens or hundreds of thousands of possible schedules for
the same lab and-using a numerical score built into each schedule-pick
the most highly rated organism.
Genetic algorithms, Dr. Boyd points out, are available commercially.
His contribution was realizing that he could create a skill- and
tech-rating scheme that would solve the problem of when to take
a particular tech off a particular workstation. The computer does
that automatically.
But by tinkering with the numerical values for a particular person
or workstation, the human operator of Lab Scheduler can also ensure
that if Mary is learning to do CBCs she will stay at that station
long enough to master them-but not so long that she loses proficiency
in anything else.
Though he seems to have made several novel conceptual and mathematical
leaps, Dr. Boyd is ever the programmer: resolutely modest. "I’m
not saying my work is all that unique," he insists. "I’m sure this
scheduling problem has been solved elsewhere, in industry. It’s
just not commercially available for some reason."
But Dr. Boyd goes on to explain that the program does not pick
the supremely, certifiably best of the trillions of possible schedules.
It does not have to. It uses tens of thousands of iterations, gadzillions
of years of simulated evolution, to do the choosing for him. Thus
Lab Scheduler just chooses an extremely good schedule, one better
than all the others it evaluates.
"The schedules that get developed are sometimes not perfect,"
Dr. Boyd concedes. "But the technologist can immediately see that
’if I just swap this person for this person,’ that would fix it
up and make it better. Minor tweaks often will make a schedule that
is good nearly perfect. The techs have found that they can spot
these readily and can fix them very easily."
Dr. Boyd knows that firsthand, because an early precursor to Lab
Scheduler was used at his own institution, the University of Virginia.
When Dr. Boyd realized he did not have the time to be a one-man
technical support operation for the product, he had the campus lawyers
license it to MAS. The program was then overhauled to make it easier
to use for those less mathematically inclined than Dr. Boyd. The
first commercial version of Lab Scheduler was released in mid-1999.
The company keeps him informed on the development of his brainchild,
and Dr. Boyd does seem to take a small amount of pleasure in seeing
his idea loosed upon the world. "My general impression is that people
are happy with it," he says. "The technologists felt the scheduling
was fairer when it was done by the computer. It was less biased.
The fact that we can take schedules out far in advance does give
them notice about when they can take vacation. Those things are
valuable to people-they give them more sense of control in their
lives."
Some of those lives are in South Bend, Ind., where the
South Bend Medical Foundation, a large regional lab, is using Lab
Scheduler. Carola Cummings, manager of centralized transfusion services,
is scheduling the day shift for two hospital labs, a stat lab, and
a transfusion service, with as many as 60 people in the program
at one time. "I really like how I don’t have to handle my vacation
requests several times,"Cummings says. "The minute I get the request,
I can log it into the computer."
Cummings also admires the built-in head count feature, which allows
her to check the staffing for a particular day and learn who’s scheduled
to work and at which workstation. But she hasn’t yet decided whether
to recommend that her administrators buy the software. Using the
old-fashioned method, she could do six weeks of scheduling on paper.
Now, with Lab Scheduler, though she can schedule for longer periods,
she can print out at one time no more than a two-week schedule.
Cummings makes clear that, because of turnover within her department,
adding each new employee’s skills, days off, vacation time, and
other personal data has proved time-consuming. She says it’s hardly
fair to judge the program given those circumstances. But she also
sounds hesitant about whether to give it an unqualified recommendation.
She needs more time. The underlying database of employee skills and rotations, for starters,
was somewhat cumbersome to set up. Beyond that, "there’s a little
more interaction at the end after the AutoLink than I thought," Cummings
adds, referring to the software function that matches techs and workstations.
She’s hoping to train a clerk to run the day-to-day schedule adjustments,
subject to supervisory approval, once the program has all the basic
information it needs.
Her purchasing recommendation is riding on whether the program
will be too difficult for the clerk to use. "If the clerk can handle
it, then I would think definitely it would be a time-savings for
everybody," says Cummings.
One reservation about the software, Cummings says, is the way
it awkwardly accommodates her techs’ staggered starting times. "The
computer likes definite start and stop times. But our staff comes
in anytime from 3, 4, 5, 6, 7, 7:30 in the morning. I have a whole
bunch of random start times. That leaves little bits and pieces
all over on the computer program; when I print out the schedule,
I go in and delete some of those half-hour periods."
In its early days in Indiana at least, Lab Scheduler also had
difficulty assigning a particular tech to one of four physical locations.
That problem was solved by the ace programmers at MAS company headquarters
in Virginia. But the software still needs her eye for now. "It still
requires that last check by a human, and some of the combinations
may not be correct. You move people around at the end."
As it turns out, they do the same thing in the labs at the University
of Virginia, but there the experience has been uniformly positive.
Carolyn Smith-Lee, operations supervisor for the clinical labs at
University of Virginia Medical Center, is not going back to pencil
and paper scheduling. For her and the other shift supervisors, scheduling
60 people or so typically ate up 90 minutes every day. "We could
never forecast," she laments of the pre-Lab Scheduler era. "We could
never plan. We could never predict."
Now she’s done with scheduling in 20 minutes or so per day. And
with Lab Scheduler, negotiating staff time off, even months into
the future, is possible. "I can see my numbers," she says. "I can
move people and talk to that employee if necessary. I can say, ’Next
March, what about Thursday instead of Friday?’ People are more apt
to work with you because they don’t have any plans yet."
Smith-Lee has discovered another use for Lab Scheduler: to figure
out the best time to install new equipment, since every apparatus
seems to be replaced on a two- to three-year cycle. "You are able
to look at your grid and say, ’These are good times. Not this time.’"
For technologists and technicians in high-pressure situations,
Smith-Lee says, the program is more likely to ensure that no one
gets worked to the bone. "With trauma," she says, "before you get
it they already wanted it. It’s good that I can give technologists
the ability not to be in that environment three days in a row, or
a week at a time. They’d be a puddle of water on the floor. A little
intensity can go a long way."
Do laboratorians understand the mammoth Darwinian selectivity
under the hood of Lab Scheduler? Probably not, says Smith-Lee. But
they do appreciate its objectivity. "They understand it in the sense
of ’my God, I’m not on urine and fluids three days in a row.’ I
have fewer people saying ’I’m on that bench all the time.’"
Lab Scheduler can even shed light on broader hiring and training
needs. Smith-Lee suspects, for example, that she would be able to
identify training needs on her own. But she says the program shows
her visually-in color-where she needs to beef up the skills of her
team. Such questions as "Do you need more individual technologists?"
or "Do you need more technologists trained in a particular area?"
would get answered under the old scheduling system, but not as fast
as they are with Lab Scheduler. The Einsteinian program makes her
day a bit more, well, Boydian.
Mark Uehling is a freelance writer in Chicago.
|