In the medical field, triage is classifying patients according to their levels of urgency in order to provide priority to the ones who need care the most. Undertriage is the practice of placing patients in more urgent categories than they actually require. It can be fatal. Currently, Japanese researchers have created a computerized technique for anticipating undertriage in a telephone triage system. Their research, which was recently published in Annals of Medicine, may enhance phone triage outcomes everywhere.
Triage protocols – why use them and how do they work?
Triage protocols are regularly modified to enhance their effectiveness and lower the percentage of undertriage. One promising method for doing this is to employ machine-learning models, which are computer-based programs that use a training set of data to find patterns and then make predictions about fresh data based on the patterns that they have learned to recognize. Recently created such programs are more accurate at determining the right levels of urgency than older techniques.
Patients may also receive triage via phone services, where they are informed of the urgency with which they should seek in-person medical attention. In order to solve this problem, Japanese researchers decided to construct machine-learning models for phone-based triage.
They used data from a private organization that uses a phone-based triage system to provide after-hours medical treatment. Two of the five programs they created were particularly effective at predicting undertriage in the data, according to the research.
Notably, the two machine-learning models that performed the best shared certain crucial characteristics. Both models revealed the same risk factors for undertriage: a higher age, being a man, having other conditions like hypertension or diabetes, and certain categories of complaints including common cold symptoms. To improve patient outcomes, these risk indicators can be employed to update phone-triage procedures. Additionally, phone-triage techniques can be evaluated and enhanced internationally using this machine-learning approach.
The medical community will appreciate any factors that help shorten the process and/or increase its accuracy because over-the-phone triage can be time-consuming and challenging to perform well. The rates of undertriage will be reduced and the phone-triage process may be sped up by giving the risk variables listed in the current study priority. Improved protocols could save many lives and lead to better patient outcomes because phone-based health care is becoming more and more popular, especially in the wake of COVID-19.
