By: Josh Morgan, PsyD
SAS Analytics
In our field of work, there are many calls to reduce suffering. Seems reasonable, right? It’s even in California’s Mental Health Services Act (MHSA), where public systems are called to “reduce subjective suffering.” And as we broadly focus more on outcomes in health, measuring suffering (and hopefully its reduction) is crucial.
In order to measure something, we have to define it.
While some definitions of suffering simply refer to the presence of symptoms, does the presence of illness alone necessarily mean suffering? Have you ever seen someone with an illness who is suffering? It’s painful, and we want to help stop it. In contrast, have you ever seen someone with an illness who is not suffering?
Ever since my Dialectical Behavior Therapy (DBT) training, I’ve preferred a more whole person approach to suffering. Dr. Marsha Linehan, the founder of DBT, defines suffering as non-acceptance of our situation. Think back to people you have known with illnesses. Does their acceptance or non-acceptance of their situation impact their ability to cope with it and therefore their suffering? How does this affect quality of life?
Does this really matter, though? Isn’t it easier to just focus on symptoms?
To measure a reduction in suffering, sure, it’s easier to just look at symptoms. Are there negative consequences of this? As I talked about last year, we can unintentionally contribute to stigma and discrimination by only measuring and talking about negatives.
At a broader level, what happens when we assume that people with behavioral health conditions suffer? Does that help give any hope of living in recovery and resilience, even while symptoms are present?
But a more whole person approach to evaluating suffering can pose challenges. Here are two suggestions on how to tackle this subject.
Natural Language Processing
The words we use matter and express a lot about our cognitive and emotional states. When we talk about things like subjective suffering, as framed in the MHSA, a qualitative approach is virtually required. It can be burdensome to conduct a robust qualitative analysis (Believe me–my dissertation was qualitative), but advances in technology, like Natural Language Processing (NLP) can speed up the process while also helping ensure all voices are heard.
As an example, many organizations already get consumer (and family member) feedback via written responses, grievances, compliments and focus groups. Well-established NLP includes sentiment analysis, which provides a quick quantitative sense of how people feel about something. A common tool in retail, sentiment analysis can be useful for stakeholder feedback, public comment periods and experiences of care. Diving deeper, NLP can pull out themes and trends that do not depend upon a person catching the right phrases and interpreting the feedback. Frankly, it can be easy to accidentally skip over a part of a response, misinterpret it, or not catch a subtlety that advanced analytics can assist in identifying. Pair those results with human wisdom in interpreting the meaning of the themes and trends, and more voices have been heard in their own words for greater impact!
In today’s quantitative world, we often shy away from the qualitative for many reasons. NLP can help bridge the gap and give rich life to our understanding of people’s lives. It’s one of the best ways, in my view, of seeing the whole person.
Whole Person Analytics
As the Chief of Behavioral Health Informatics at the San Bernardino County Department of Behavioral Health, I led systemwide strategy to evaluate outcomes. We spent many hours talking about how to tackle subjective suffering. Our solution was to not focus on just a single metric, but at least two data points. Symptom reduction could be one, but there had to be another metric along with it, such as improvements in hope. If someone had improvements in hope AND improved symptoms, for instance, the chances of reducing suffering is likely.
Oftentimes, we focus on a single data point as our metric. There’s good reasons for this. But it can be limiting and inaccurate, especially when we try to get at concepts like suffering. Combining data points together to get a more whole person perspective will give us a better sense of what’s really going on in our communities and with the people we serve.
A major question with these suggestions is how to get the data I suggest. Head over to my LinkedIn article, “Data sources to assess whole person suffering” for initial thoughts on potential data sources. Stigma and discrimination reduction are major themes in the work we do. Let’s use data for good to tell a more complete, accurate story of people’s lives, suffering, recovery, resilience, and wellness!
This is a very interesting point about some of the most cutting edge possibilities in analyzing the needs of children and families. One of the most important ideas articulated, is that successful measurement in human services require not just measuring how one variable moves, but how multiple variables move, and in relation to each other. The recourse to language analysis makes sense — if a computer could assess a conversation in a room for idetic themes, etc., could it facilitate a fuller understanding of a reasonable treatment plan faster? The power of a NLP styled machine learning would be bigger still. Consider that communimetric data is always to a certain extent linguisitic, because it is all built on communicating what a client’s needs and strengths are. As such, analysis of a person’s CANS or ANSA, especially over time, is a story — can we identify themes in that story like we might in a person’s narrative? Can we predict the next phase of the story, like we can with a person’s statements (much like Alexa or auto-correct does)? Humans are complicated, but machines thinking in new ways can help us better grasp some essential features, and that can help us revolutionize our care systems.