Competitive Learning Environment Fuels Care Solutions and Engagement

Hosting a ‘hackathon’ kept team members connected and broadened learning while adapting to a remote work setting


Drew Chewning knows hackathons. With a cybersecurity background, he’s familiar with the concept. 

Advertising Policy

Cleveland Clinic is a non-profit academic medical center. Advertising on our site helps support our mission. We do not endorse non-Cleveland Clinic products or services Policy

“From my perspective, hackathons served as a platform for cybersecurity experts to test a new technology’s security,” says Chewning, cloud manager within Cleveland Clinic Enterprise Analytics. “Bright minds coming together to push the boundaries; basically, to break the new tool or program.”

Then Joseph Dorocak introduced him to a new hackathon concept.

A data scientist and Cleveland Clinic’s Enterprise Analytics manager, Dorocak explains, “When the onset of COVID-19 forced remote work, I looked for opportunities to keep my team engaged. We landed on a monthly week-long meeting to tackle backlog projects and apply hackathon concepts; except we weren’t trying to break things.”

These sessions connected the team in a competitive learning environment that broadened their knowledge and honed their capabilities. But, there was more to learn.

Seek experts to propel learning

To enhance their skillset, Dorocak invited Chewning, Santino Rizzo, a cloud architect and lead for the Cloud Center of Excellence as well as external partners to learn cloud computing and Databricks.

“They understand the technical capabilities that we needed to bring our project into reality,” says Dorocak. “We leaned on their expertise to expand our analytic reach with the cloud environment and Databricks.”

Databricks is a cloud-based tool used to process, transform and explore large quantities of data. 

Rizzo explains, “Expanding our data and analytic resources to meet our caregivers’ needs pointed to Cloud software. Artificial Intelligence (AI) and Machine Learning (ML) capabilities available through Databricks are of the most interest.”

Advertising Policy

Cloud technologies provide vast access to more resources to scale projects efficiently. Using Databricks is ideal for large data problems with significant complexity.

“Technologies in the cloud are quite extensive,” says Rizzo. “During a hackathon, there’s a greater potential to need additional resources to keep the work moving. The cloud allows those resources to be available on demand.”

Rizzo continues, “At the time, we had limited experience with Databricks, so their initial ask piqued our interest as an opportunity to get better at the product. The fast-paced hackathon session spurred quick learning to overcome hurdles, to iterate the process and mature the technology.”

Conquering the cloud

Hackathon week involved 10 data scientists vying to develop the best, fiscally viable care model.

“Through a clinical lens, we’re trying to improve processes with AI and ML.” Dorocak adds, “From a financial perspective, we’re taking a holistic view at stratifying data that support clinical decisions and optimizes patient care.”

The goal was to have the best-performing model by the end of the week. Competitors borrowed ideas, shared insight and quickly adapted to new capabilities.

Dorocak’s team chose to create a predictive model for earlier identification of sepsis patients. Sepsis, a severe bloodstream infection, can potentially cause tissue damage, organ failure and death. Symptoms often mimic other conditions, making diagnosing difficult; delayed treatment increases sepsis mortality.

“Sepsis is a national concern that several data science groups across the country are trying to develop predictive modeling to more quickly identify at-risk patients,” notes Dorocak. “It’s a big data, big compute problem, which is ideal for Databricks.”

Advertising Policy

During the hackathon, the training and model-building used historical data. This provides a good proxy, but only real patients and encounters will validate its accuracy.

“A strong validation process is essential before moving the model into production,” explains Dorocak. “We’re getting close to scoring real patients. The next phase is a three-month evaluation on the backend to confirm the accuracy and determine best case use.”

Extended learning brings strong partnerships

Staying connected in a remote work setting can be challenging. This hackathon fostered camaraderie and collaboration.

Dorocak agrees:  “This experience created nice touch points to learn what others are working on and how they approach their work. The knowledge transfer is really beneficial.”

“While we have been working with these teams in other small capacities, this was a continuation of our work and the ability to expand available resources,” Rizzo adds.

For Chewning, “This session strengthened our departmental partnerships while expanding and evolving our knowledge. And from a cybersecurity perspective, we created a replicable model that protects sensitive data in a secure environment while fostering collaboration — and that’s a very good thing.”