Using behavioral science and AI to meet the needs of patients and providers

Why do people make decisions that are not in their best interests? Why do people forego opportunities that will improve their physical or economic well-being? Thanks to the groundbreaking work of researchers like Nobel Prize winner Richard Thaler, we know why people often don’t act rationally. They can be affected by biases like the “availability heuristic,” “bounded rationality,” or “loss aversion.” These technical terms simply explain why people make ‘bad’ or suboptimal decisions. Behavioral science takes psychology and real-world data into account when analyzing and predicting human behavior.

We believe that when it comes to healthcare, leveraging the insights provided by behavioral science along with data and AI can improve both financial and clinical outcomes.

Patient access and revenue cycles: defining the problem

Problems with patient access to healthcare is a big problem for health systems. Take, for example, a Medicaid-eligible patient. We know that patients who present as self-pay can be a significant source of financial strain on health systems. These patients can undergo significant acute services; but if they're uninsured, the likelihood of the health system being reimbursed for the services provided is very low. Therefore, helping these patients enroll in Medicaid, disability, and other types of federal, state, and local programs is important, for the patient herself and for the provider.

Medical claim denials are also on the rise. According to Change Healthcare data, denials have risen 23% since 2016, accounting for over 11% of all claims. About half of these denials are caused by front-end revenue cycle issues like authorization/eligibility.

COVID has disrupted the authorization, eligibility, and enrollment process. Between attrition due to burnout and the unintended consequences of vaccine mandates, hospitals are chronically short-staffed. Sometimes, hospitals are unable to have a specialist meet with a potential applicant in the emergency department to help determine if they are eligible for Medicaid. Fast-forward and you end up with a patient in debt and a hospital that isn’t reimbursed for the care it provided.

Lack of adequate clinical triage also creates problems for both patient and provider.Providers can sometimes struggle with identifying, triaging, and segmenting patients to the most effective care setting. When providers are unable to triage patients appropriately, patients often default to using the ER or don’t seek treatment at all. 

“Transitions of care”—which refers to the movement of patients between healthcare practitioners, settings, and home as their condition and care needs change―is another source of friction and inefficiency. Ineffective care transition processes lead to adverse events and higher hospital readmission rates and costs.

Empowering better decisions

So how can behavioral science help address these and other pernicious problems plaguing our healthcare system? Take, for example, a patient’s capacity to weigh all available information and options. When faced with a medical problem, the patient can be buffeted by many stresses, including fear, anxiety, financial insecurity. Behavioral science teaches us that this can lead to “cognitive depletion.” Simply put, the mind is overwhelmed by too much information and burdened by too much stress to properly process all the signals.

Therefore, patient communication can and should be simplified and coupled with sentiments of empathetic support. Just because a given action will benefit the patient, behavioral science tells us that this doesn’t mean he/she will necessarily take the action. But by breaking up the tasks in simplified “calls to action,” and making common cause with the patient (“Let’s finish this together”), we can nudge the patient toward the desired outcome.

Behavioral economics can also be used in efforts to enroll patients for medical coverage. The fear of loss―“loss aversion”―is twice as powerful psychologically as the possibility of gain. So, as we engage patients, we can make them more likely to begin an enrollment process if it promises to allow them to “not miss out,” rather than “make an improvement.”

Smaller slices: coupling behavioral science with data science

Using behavioral science in patient engagement is intrinsically a personalized process. One size does not fit all. Data science and AI can help us match the message to the patient and the appropriate platform. It allows us to recognize far more specific slices of the population and moves us away from broad segmentation.

We can identify those most likely to be eligible for disability programs or Medicaid. It can also help us understand the right method of communicating, and to whom, at each step in the patient journey.

For example, one might assume that a grandmother in her 70s would prefer phone calls and written communications from her providers. However, we know from our data that many seniors are far more digitally savvy than one might expect. They’ve taught themselves how to text and interact on social media so they can stay in touch with their grandchildren perhaps. Communicating with them digitally for certain transactions and events may be the most appropriate, something that traditional broad segmentation would not allow us to recognize.

Data and AI can also predict those patients who are most likely to be “no-shows.” We know that no-shows are a drag on efficiency and resource management. So, if we know who is most likely to miss an appointment, we can remind them of their appointments more frequently, remove the friction points that are causing them to miss the appointment, and deploy resources with the expectation that a certain percentage of people will not show up despite our best efforts.

Transforming the patient and provider experience

Behavioral science coupled with data and AI can inform a nuanced approach, which helps us understand the motivations and needs of patients and helps indicate how best to focus patient-access strategies. It helps us better understand how to more effectively engage and how to use an omnichannel communications approach more effectively to engage with the entire patient population.

Above all, combining behavioral science, data, and AI leads to better fiscal and clinical outcomes for patients and providers.

Jason Lee, Vice President of Product Management

B.A., Economics, Harvard University; MBA, Healthcare and Corporate Social Responsibility, Walter A. Haas School of Business at UC Berkeley
Jason Lee oversees offerings across the revenue cycle to ensure Change Healthcare’s portfolio of RCM service offerings address key customer needs. Jason is passionate about developing solutions that help patients navigate effortlessly through their healthcare journey.

Follow Jason on LinkedIn

Tabitha Hillman-Burcham, Behavioral Science Researcher and Senior Manager in AI

B.A., International Journalism, The Ohio State University; PhD Behavioral Psychology, The Ohio State University
Tabitha Hillman-Burcham is a Behavioral Scientist at Change Healthcare. She leads a team of Behavioral Science experts within the AI organization who aim to understand and defend the motivations, intentions, feelings, and needs of selected populations. She is currently an adjunct professor at Indiana University.