The Keck School of Medicine of USC: Accelerating clinical trials using ML
About The Keck School of Medicine of USC
Keck School of Medicine of USC, a University of Southern California school, is a prestigious institution focused on preparing healthcare professionals to improve patient care in the U.S. and worldwide.
Tell us your challenge. We're here to help.
Contact us
About Pluto7
Pluto7, the 2019 Google Cloud Specializations Partner of the Year in Data and Analytics, provides best-in-class AI and ML consulting services and solutions for businesses across industries.
The Keck School of Medicine of USC worked with Pluto7 and Google Cloud to leverage machine learning in order to streamline clinical trial management.
Google Cloud results
- Accelerates Medicare Coverage Analysis from days to milliseconds
- Improves efficiency of clinical trial administration to free up staff and budget
- Expedites clinical trial activation and completion through further ML deployments
Significantly reduces clinical trial activation time
As research and life science organizations around the world devote tremendous energy to drug development, highly efficient processes and cost-effective technology are essential within clinical trials to get therapies into the hands of patients. Machine learning (ML) has proven its ability to transform clinical research, but it can also expedite the clinical trial administration process through automation and establishing predictive analytics.
The University of Southern California (USC) Keck School of Medicine and Southern California Clinical and Translational Science Institute (SC CTSI) embarked on a digitization journey to modernize USC’s Medicare Coverage Analysis (MCA) processes managed in the Clinical Trials Office.
MCA processes are a critical component of clinical trial administration, since they dictate where budget will be pulled from. Traditionally, this highly manual procedure was a bottleneck in clinical trials but is a critical component of the process of launching a study. The school felt as though ML could provide the automation it needed to expedite the process.
“We work with teams on promising clinical trials to remove roadblocks for researchers at USC and across the U.S.,” says Allison Orechwa, Ph.D., Director of Programmatic Development at SC CTSI. “We need to be as efficient as possible in our approach to empower our researchers, and taking a more advanced approach to analytics stood out as one of the best ways to accomplish this.”
Allison and her team decided to work with partner Pluto7 to take advantage of powerful, healthcare-ready Google Cloud analytics solutions.
“We needed access to real-time data but didn’t have a way to access it, which helped prompt our broader digital transformation. When we started to explore technologies, Google Cloud looked like a game changer, and we knew Pluto7 could further align our processes with the right technologies.”
—Allison Orechwa, Ph.D., Director, Programmatic Development, SC CTSI (Southern California Clinical and Translational Science Institute)Laying a solid foundation for advanced analytics
After bringing in a lean process improvement expert to evaluate administrative workflows, the school identified many areas of improvement within regulatory, contracting, and hospital administration units. Over six months, the school eliminated 30 percent of the steps needed to approve clinical trials, while accelerating activation times by 50 percent. The lean process improvement effectively streamlined the steps leading up to patient enrollment through more automated administrative activities.
Once those lean processes were implemented, the lean expert, a member of the USC Marshall School of Business, noted that advanced analytics and machine learning could yield even more dramatic results. The school had a relatively advanced clinical trial management system in place, but it was not easy to customize to keep pace with changing needs.
“We needed access to real-time data but didn’t have a way to access it, which helped prompt our broader digital transformation,” says Allison. “When we started to explore technologies, Google Cloud looked like a game changer, and we knew Pluto7 could further align our processes with the right technologies.”
“At our first meeting, USC made it clear they were seeking to use ML to solve previously unsolvable problems. We needed the best technology, and Google Cloud stood out from the beginning.”
—Manju Devadas, founder and CEO, Pluto7Streamlining clinical trial activation with ML
Pluto7, the 2019 Google Cloud Specialization Partner of the Year for Data and Analytics, was perfectly positioned to help the school with its transformation. Pluto7 had extensive experience applying advanced analytics to streamline processes.
Pluto7 first surveyed USC researchers to identify challenges they faced. They found that administrative burdens, disparate systems, lack of process standards, and long activation times were the most common.
Working with USC and building off lean process improvements already in motion at the university, Pluto7 developed a workflow to accelerate clinical trial administration. The budgeting process is a particularly complex area that the school believed could be dramatically improved, especially for cancer studies.
“At our first meeting, USC made it clear they were seeking to use ML to solve previously unsolvable problems,” says Manju Devadas, founder and CEO of Pluto7. “We needed the best technology, and Google Cloud stood out from the beginning.”
Typically, a highly skilled research administrator has to look at each study procedure to decide if it should be charged to the research sponsor or if it is a standard of care that should be charged to insurance, which would take days.
To streamline this manual process, Pluto7 created an algorithm to read standard of care guidelines, match a given procedure with those guidelines, and assign it to the correct billing category in milliseconds. Google Cloud solutions such as Document AI and Cloud Vision further accelerate analysis.
“The process was highly collaborative right from the start, as one of our MCA experts worked with Pluto7 engineers to ensure the algorithms were trained properly,” says Allison.
“This is about saving lives. Getting new drugs developed and to market can greatly improve patient outcomes, so the faster we are, the better the impacts for healthcare providers and their patients. With Pluto7’s help and Google Cloud ML solutions, we have a foundation for accelerating clinical trials and supporting better patient outcomes.”
—Allison Orechwa, Ph.D., Director, Programmatic Development SC CTSIComparing end-to-end processing time and given limitations within the proof of concept scope, the model saves considerable time by automating today’s complex human-dependent decision-making process. After successfully completing a proof of concept for the MCA process, the school significantly shortened the budgeting process for breast cancer studies. The model also predicted billing designation with accuracy ranging from 70–90 percent, an improvement from the previously manual process. Now the school is looking into a range of other projects to accelerate clinical trials across other diseases.
USC completes roughly 200 clinical trials annually, and each one takes up to 20 people more than 80 hours to activate. The more that time can be reduced, the faster critical results will be available within USC and the broader research community around the globe.
“This is about saving lives,” says Allison. “Getting new drugs developed and to market can greatly improve patient outcomes, so the faster we are, the better the impacts for healthcare providers and their patients. With Pluto7’s help and Google Cloud solutions, we have a foundation for accelerating clinical trials and, in the end, supporting better patient outcomes.”
Constant improvements aided by ML
When the project first started, Pluto7 used Google Cloud solutions to create ML algorithms tailor-made to USC’s specific needs. The school is now working hand in hand with Pluto7 to further explore large-scale analytics and ML models out-of-the-box over serverless infrastructure with BigQuery, Google Cloud’s enterprise data warehouse.
“We solved one problem in a very large, complex workflow,” says Manju. “We’re now beginning to align more advanced ML solutions to other workflows for even higher-impact results.”
To date, the Keck School mostly uses its own datasets to support the analytics to improve clinical research informatics. BigQuery presents an opportunity to pull and combine that data with public datasets for even more powerful results, including running blazing-fast queries on petabyte-scale future data for enriched insights about clinical trials and population health research.
“We are focused on putting this work into action in our own school, but the more we share the proof of concept, the clearer it becomes how Google Cloud mixed with ML algorithms can improve a range of processes,” says Allison. “We are one of 60 academic U.S. medical centers charged with translating science into better healthcare for everyone [through the CTSA program]. The work we are doing now helps us improve research and enhance patient outcomes today and tomorrow.”
Tell us your challenge. We're here to help.
Contact us
About The Keck School of Medicine of USC
Keck School of Medicine of USC, a University of Southern California school, is a prestigious institution focused on preparing healthcare professionals to improve patient care in the U.S. and worldwide.
About Pluto7
Pluto7, the 2019 Google Cloud Specializations Partner of the Year in Data and Analytics, provides best-in-class AI and ML consulting services and solutions for businesses across industries.