The healthcare industry has been a champion of whole-person care for decades, yet we seem no closer to a solution for proactive health management than we were in the early days of nurse advice lines and disease management programs.

In the past, public and private healthcare organizations looked toward Maslow’s Hierarchy of Needs to help individuals with severe health issues better manage chronic disease. The industry reached out to underserved populations to assess and address basic needs, such as food, water, housing and transportation. After all, if people don’t know where their next meal is coming from, there’s little chance they’ll be thinking about improving their health.

Today, the social determinants of health (SDoH) focus on many of the same concepts found in Maslow’s hierarchy but in more concrete terms. Described by the Centers for Disease Control and Prevention, these factors include unstable housing, low income, unsafe neighborhoods or substandard education, each of which has the potential to negatively impact healthcare and patient satisfaction if left unresolved.

SDoH assessments are becoming easier to accomplish thanks to artificial intelligence and algorithms; the hard part is acting on the information culled from the data. Often, few processes and cost-effective resources are in place to resolve issues, especially for underserved populations.

Parsing the Data

In the early days of disease management, government-run health plans typically provided details about patients with chronic conditions to the organizations providing care. However, it wasn’t until nurses contacted these individuals that the health organization would learn of the many issues keeping patients from receiving care.

Thanks to steadily maturing technology, we now can gather this information concurrently and automatically from a variety of patient records, providing caregivers with the information necessary to address SDoH.

SDoH data, for example, is often recorded in progress notes, admission notes, procedure notes, discharge summaries and consultation notes in an electronic medical record (EMR). It’s challenging, however, to cost-effectively and efficiently extract this data. 

To meet these goals, healthcare organizations can turn to AI and natural language processing (NLP) algorithms to identify SDoH factors for individuals on a very large scale. AI and NLP systems can review existing clinical notes in the EMR to identify patients at risk for SDoH factors, with no need for additional surveys, phone calls or clinical workflow steps.

To start the process, an ontology needs to be created for the NLP algorithm. The algorithm then compares the ontology’s list of words and phrases relating to SDoH with those found in the clinical notes. The initial ontology should include common words and phrases used by physicians and clinicians to describe SDoH, such as “food insecurity” and “homeless.”

The algorithm’s accuracy can be ensured by using SDoH terms standardized by the National Institutes of Health and included in the Unified Medical Language Library. One study shows an SDoH algorithm with 88% accuracy. Accuracy increases as the ontology set grows: As the algorithm is exposed to more records, it continues to learn the terms most commonly related to SDoH and automatically refines its ontology.

Because the AI systems and algorithms are automated, physicians and clinicians are free to focus on patient care. The information could be supplied to care coordinators who work directly with patients to improve SDoH-related conditions with the objective of improving health outcomes.

Taking Action

Collecting and analyzing the data is only the first – and arguably the easiest – step. A host of complications arise once a plan is devised, as the plan must be made actionable and financially tenable. But despite the complexities, it’s incumbent on the industry to begin making progress on helping underserved populations receive regular healthcare services and manage their health.

As providers assume increased responsibility for whole-person care under value-based care programs, payers, caregivers and the healthcare industry will need to join together to address SDoH holistically.

The Robert Wood Johnson Foundation has created an exhaustive library of research, analysis and news reports highlighting the various ways in which SDoH  could be addressed in the community.  But it will be difficult to make improvements without the investment of significant resources from a number of public and private entities. Additionally, any programs or changes must be balanced with the patient’s privacy and dignity. At the end of the day, better health management can’t be forced on the individual.

In the meantime, the industry can take small, effective steps to affect positive change. Social prescribing is one way to make progress with patients of all types. Rather than – or in addition to – prescribing medication, a caregiver may write out a prescription for a daily walk through the park, spending more time with loved ones or taking an art class. Such activities alleviate challenges to patients’ physical health, psychological well-being, social isolation or financial stress. In addition to positive side-effects, many of these activities are often available at low or no cost. A study published in BMJ Open suggests positive results can be had when addressing these issues.

Many questions remain when it comes to eliminating the challenges caused by SDoH factors. It’s a broad issue that demands a far-reaching response from government agencies, non-profit organizations and private industry. While technology can help identify those in need of assistance, we must work together to create a solid set of processes and develop deep resources and programs, all the while understanding that the ROI of improved healthcare and decreased treatment costs will be necessary for long-term success.

Mark Miller

Mark Miller

Mark Miller is AVP, Marketing, Provider and Health Systems within Cognizant’s Healthcare Business unit and leads the strategic and tactical marketing for... Read more