Virtual Triage: A New Way of Identifying Patients in Need of Emergency Care
By Piotr Orzechowski
Patients sometimes underestimate the acuity of their conditions and forgo emergency care when it’s needed, resulting in higher mortality rates and higher-acuity-care utilization that is more costly to the U.S. health system in the long run.
However, technology application has shown the ability to help patients significantly reduce avoidable health risks and healthcare costs by enhancing early detection and care alignment. Virtual triage (VT) engines can be thought of as “symptom checkers” that leverage artificial intelligence to provide access to information services beyond physician office hours on an anywhere, anytime basis from any internet-connected device.
A recent study based on more than 3 million patient VT interviews demonstrated that the technology offers a new vehicle toward early detection of severe evolving pathology by providing a way of engaging patients who believe their symptoms are not serious. Additionally, VT was shown to be a beneficial tool in accelerating care referral and delivery for life-threatening conditions when patient misunderstanding of risk results in care delay.
Background on virtual triage
About 14% of U.S. emergency department (ED) visits involve life-threatening illness. In some cases, had these patients come to the ED earlier, outcomes would have improved as they would have received prioritized treatment and their conditions would not have deteriorated as quickly, thus reducing the length of hospital stays and lowering overall care costs. Similarly, when patients delay seeking care for serious conditions, they may increase their risk of avoidable mortality and need higher-cost services and treatments.
VT can alleviate these issues in numerous ways. For example, the technology can enable patients to avoid long waits in the ED plus associated travel time, as well as when patients are uncertain whether symptoms warrant medical attention or what kind of care to seek, and, further, if symptoms worsen.
Here’s how VT works: The technology interviews the patient, starting with basic questions about age, sex, symptoms, medical history, risk factors, and medications. Then, based on those answers and Bayesian probabilities, asks the patient the next-most-relevant question, like the way a human clinician processes information and interacts with patients.
Once the VT AI has evaluated all patient data, it computes the probabilities of likely conditions using a statistical algorithm for advanced symptom assessment informed by an underlying medical knowledge base. Lastly, the technology advises the patient of probable causes, illness severity, and recommended acuity level of medical care.
VT has shown itself to be far superior to internet browser search because it evaluates patients systematically to identify and communicate the appropriate level of care acuity based on a clinical AI algorithm that has been validated by physicians. Further, the technology offers greater benefits than traditional clinical triage protocols and decision trees because it can dynamically calculate the probability of possible conditions and adjust the patient interview accordingly.
However, it is important to note that results from VT interviews are not diagnostic and not medical consultations. The technology provides information and guidance only. When the VT engine determines that a threat to life is present, it advises the patient to seek immediate medical assistance.
Impact and opportunities: 3 benefits of virtual triage
To conduct the study, researchers reviewed a dataset of more than 3 million VT interviews conducted over 16 months. Researchers then narrowed those records to more than 12,000 interviews in which patients reported symptoms of five potentially life-threatening conditions (myocardial infarction, stroke, asthma exacerbation, pneumonia, and pulmonary embolism) but indicated that they did not intend to seek urgent care.
When they analyzed the data further, researchers found that 38.5% of individuals whose VT indicated a condition requiring emergency care had no pre-triage intent to consult a physician. Furthermore, 61.5% intending to possibly consult a physician reported that they had no intent to seek emergency medical care.
The VT study showcased the following three ways that the technology can benefit patients, providers, and other major healthcare stakeholders:
- Influencing behavior: VT AI has a strong influence on modifying care-seeking behavior, demonstrating its critical role in healthcare pathways.
- Cost reduction: By ensuring early detection and appropriate care referral, VT technology helps avoid costly emergency care and long-term treatments, effectively reducing healthcare expenses.
- Improved patient satisfaction: Patients benefit from shorter wait times, more efficient care, and the convenience of accessing healthcare services remotely, all of which contribute increased satisfaction with their healthcare experiences.
Evidence has shown that, in many cases, patients who need urgent care do not seek it, often because they are unaware of the severity of their conditions. By using AI to better identify patients who may need urgent care, virtual triage technology offers a cost-effective, patient-friendly means of early detection and care alignment.
Piotr Orzechowski is the founder and CEO of Infermedica, a digital health company specializing in AI-powered solutions for symptom analysis and patient triage, which he founded in 2012. He began his professional career in the gaming industry as an engineer and software developer and worked for several smaller businesses and startups before founding Infermedica with the goal of making healthcare more accessible and convenient for everyone.