Rich Jolly, VP Analytics & Strategy, Avamere Family of Companies
Iroutinely speak with C-Suite executives who complain of the high cost and limited (or no) business benefit from data scientists. This feedback, along with the daunting task of determining how to get started, might deter even the most motivated executive from launching an advanced analytics program. In this article we’ll look at four principles to help you jump-start your data science endeavor and discuss the catalysts, challenges, and path forward to realizing EBITDA from your advanced analytics.
In my role as VP of Analytics & Strategy at Avamere Health Services, I have been tasked with building predictive analytics to support clinical decision support and management resource planning. Healthcare is challenged with maximizing patient outcomes while minimizing costs all within the envelope of a complex reimbursement environment.Avamere provides a spectrum of post-acute care across skilled nursing, home care, long-term care, hospice and therapy - getting business value from analytics was no easy task.
Start small and start now!
It is an important time in the evolution of machine learning as several catalysts are radically changing the landscape. Cloud vendors are making high capacity computing and advanced managed services immediately available, with low technical barriers. Open source software such as Python, is bringing democratic capability to analytics. And, perhaps most importantly, advanced open source machine learning libraries such as Pandas, TensorFlow and Scikit-learn bring the most advanced models to all data scientists in an easy to use format.
Although the technology is readily available there are still great challenges to creating business value from it. For example, it is difficult reconciling the multidisciplinary nature of the task which ranges from working with clinicians to understand their processes for care delivery, working with software tools to develop models, understanding statistics to interpret the model’s performance and to pure software engineering in the delivery and maintenance of the tool. To start, assemble a team and work with them to find the low hanging fruit. Don’t ignore technology like Robotic Process Automation (RPA), which may not be cutting edge data science (being more biased to software engineering) but can deliver strong bottom line business value with relatively small resource requirements.
Build upon the successes of your Business Intelligence (BI) team
Moving to advanced analytics and machine learning is an evolutionary process for an organization. It is very important to first master the capabilities in descriptive and diagnostic analytics (KPIs, reporting, dashboards, etc.) before moving to predictive and prescriptive programs. To pinpoint areas in your business to apply advanced analytics and gain business value use the Lean technique of Value Stream Mapping to identify key points of information leverage.
Finding good data scientists is a challenge. First, let’s be clear, there is no shortage of people who want the job. Even as a healthcare firm we are flooded with job applicants. But the industry has limited ability to take on aspiring data scientists. Accomplished professionals are hard to find, and often gravitate to research or very high demand roles. Consequently, finding leaders who can assemble talent and navigate this space are also scarce.
Bridge the gap between Technology and Business
Bridging the gap between the business world and technical space is a real challenge. I truly consider this to be a cultural difference, and the bridge, which is often the data science manager, needs to be a translator between the two worlds. Being able to understand the business problem while also encompassing the technical possibilities is perhaps the most critical skill in landing analytics that create business value.
You not only need to build up a capability, but also build up the understanding and commitment within the organization. Don’t be surprised if you have some other executives who don’t see the need. Use your successes to build the case for them. It’s all part of the process.
Build solid processes for consistency and quality
As you grow, think broader and build solid processes for consistency and quality. These processes include Analytics Center of Excellence, Data Governance, etc.