We might not be able to observe the progressive loss of cognitive and intellectual abilities someone with dementia is experiencing from the outside, but healthcare clinicians can detect it when they observe their ability to bathe, groom or dress themselves deteriorate. Minitab consultant and Insights 2019 speaker David Patrishkoff is researching how to help with the aid of Minitab software.
Activities of Daily Living (ADLs)
The healthcare industry calls basic self-care tasks like one’s ability to properly feed themselves, move around or go to the bathroom “Activities of Daily Living,” or ADLs. Since the 1950s, healthcare professionals have scored ADLs with pre-set criteria (see this worksheet from the National Palliative Care Research Center for example).
After populating a worksheet like this, a healthcare professional can flag the functional capabilities of older adults and use the results to assess their ability to live independently.
What if symptoms could be caught earlier?
Enter Machine Learning and Predictive Analytics
There is evidence that deterioration of ADL scores are preventable. Screening can greatly help as the first step in the process too. For example, preventing elderly patients from falling has been shown to reduce the use of home healthcare, and the associated costs.
Building off of related research, Minitab consultant David Patrishkoff set out to use Machine Learning to help detect ADL deterioration earlier in the process and address it accordingly.
In healthcare, interventions are activities or strategies (such as screenings or vaccinations) to assess, improve, maintain, promote or modify health of individuals or groups. David uses Minitab Statistical Software and Salford Predictive Modeler (SPM) to examine 1,200+ interventions and therapies that nurses and home care workers provide to people across the country, and select the best ones to maintain or improve their independence and their ADLs.
“I start in Minitab with data visualizations and clean up the data, then jump into SPM for the really heavy lifting of very complex data sets,” David said. “I have columns of data where I have one of any 83,000 prescriptions that are prescribed to people. There are 43,000 diagnosis codes too. The algorithms in SPM can deal with highly dimensional data.”
A Master Black Belt who began his career in the automotive industry, David has consulted in about 60 different industries worldwide and trained nearly 30,000 professionals in Lean Six Sigma and patient safety.
Applying Predictive Analytics to Problem-solving
David first began using TreeNet in SPM to enhance his research into causes of traffic accident injuries and deaths, and he is applying some of the same methods now to ADLs and home care.
There is a belief that you have to be a data scientist coding in Python and R to handle these kinds of problems, he notes, but that’s not necessarily true. David recommends learning to use predictive analytics software like SPM to see how you can do better root cause analysis.
David also credits the 64-bit version of Minitab 19 with helping him with larger data sets he was unable to work with in previous versions.
“It helped me tremendously,” he said. “I had old files that were too big and then with the 64-bit Minitab 19 it further helped my analysis.”
David has been speaking at conferences about his research and how classic Six Sigma and operational excellence practitioners can build on their knowledge of statistical methods to take the next step into the data science revolution. He plans to present and publish further findings next year on how to provide home healthcare clinicians a stable methodology to improve patient outcomes.