CASE STUDY: ADVERSE EFFECTS REPORTING
89% ACCURACY
ROI IN 15 MONTHS
MACHINE LEARNING
OVERVIEW
An Adverse Event (AE) is described as “any untoward medical occurrence” in a patient
receiving a medicine. These events include everything from so called “non-serious” side
effects such as nausea to more serious events such as reports of malignancy. Our client dealt
with more than 100,000 of these per year, which is a mandatory legal responsibility. The
pharmaceutical company had a CRM in place to track input from various devices but this
was reliant on humans (doctors, clinical specialists, etc) entering the right amount of detail in
cases. This information then needed to be evaluated by specialists every day to determine if a
case was deemed to have an adverse effect.
WE DELIVERED
We were able to develop a framework that included keywords often associated with AE
cases, as well as background factors such as age, sex, underlying conditions, additional
medication, medical history. Each day the bot would extract the information and highlight
the cases to the expert that were most likely to be adverse events.
OUR RESULTS
The initial success rate of the bot was 70%, but as more information was fed into the system
and the parameters could be adjusted this rose to 89% and meant more of the specialist
time was spent dealing with the right cases.
The firm saw a return on investments in 15 months, but this doesn’t convey the true benefits
and savings the firm would have seen had their reporting been late and fines applied. This
implementation took a large amount of pressure from the team and ensured the most
important cases were dealt with first.