The CognitiveScale healthcare team recently returned home from an exciting week in Vegas for the one and only HIMSS 2016. Among the 1500 health IT vendors, and 100s of education sessions, there were a few themes that boldly dominated the conference: risk analytics, telehealth, and, most importantly, population health.
You can think of population health in terms of three overarching questions.
- Whom do we engage? Identify a population that needs to be targeted by aggregating clinical, administrative, and consumer data. Stratify them through various analytics.
- How do we engage them? Find out what insights are relevant to engage and motivate a person at an individual and contextual level. Nudge them towards the right behaviors, while simultaneously learning from their actions to adjust these insights.
- What outcomes are we managing to improve their health and care? Determine tangible outcomes from the quality objectives defined by value-based payment contracts or Medicare payment incentives.
At the 2016 HIMSS conference, there was an explosion of solutions answering the question “Whom do we engage?” The healthcare innovation space has seen a proliferation of vendors, both big and small, focused on identifying a particular population through a variety of approaches, from finding gaps in care against standard quality measures, to employing predictive analytics that identify future risk. The companies we saw ranged from general purpose solutions, to specific solutions hyper-focused on one area (e.g. risk of sepsis).
Nevertheless, while numerous companies showcased the who, there was a noticeably limited number of vendors with tangible approaches to answering the how. #Engage4Health was an official hashtag of the conference, yet the majority of patient engagement solutions focused only on pushing messaging and notifications to customers, and provided no insight into a multitude of nagging questions integral to achieve true personalization.
Do we really know everything about the people we are trying to engage?
For example, what environment and socio-economic condition are they in? What are their dominant behaviors? And beyond their current conditions, how do we motivate them to engage in managing their health? What insights will nudge the right behavior in the long-term? Finally, how do we learn from past results what works and what doesn’t in order to adjust our approaches continuously?
You might now be wondering how on earth one could learn so much about a population. Enter cognitive technologies. We at CognitiveScale are focused on answering the questions that enable engagements to be contextual and actionable. Using machine learning and natural language processing (NLP) technologies, we have started to make more healthcare data meaningful for delivering care — particularly through harnessing the enterprise dark data, which includes physician notes, customer service interactions, environmental data, signals from social feeds, and other forms of data that are traditionally difficult for a computer to digest. Furthermore, deploying cognitive solutions allows for continuous learning from interactions and events. Adapting our models in real time will allow for a transformation of customer engagement models in healthcare from traditional models, which rely on extending the clinical workforce (not very scalable), or highly impersonal messaging (not very engaging).
Population health will fail to deliver value in the long-term if this gap in patient engagement is not addressed. To truly achieve change, vendors in population health must broaden their focus to the entire population — not just the high utilizers — and engage and motivate them at a level that leads to meaningful, contextual change. It’s time we move beyond the whom, and answer the how.