A research team led by Professor Nanako Tamiya and Associate Professor Ryuta Iguchi of the University of Tsukuba School of Medicine found that the emergency home visit service, in which a doctor visits a patient's home to perform an examination, determines the degree of urgency to be low when making a telephone inquiry in advance. To avoid undertriage, we built a machine learning model to predict which patients are likely to underestimate urgency.The researchers will apply this model to a real emergency room service to see if it reduces undertriage.
According to the University of Tsukuba, the research team analyzed data from approximately 2018 of the approximately 11 patients aged 2021 and over who used Fast Doctor's emergency home visit service from November 1 to January 16. did.
As a result, undertriage patients accounted for 1.6% of the total, with an average age of 38.4 years and 57.2% were male.The main complications were hypertension and chronic lung disease, with common cold symptoms and fainting.
Of the five machine learning models we created, we investigated which information in the model with the best performance would facilitate under-triage, and found that elderly patients had complications such as hypertension, diabetes, cerebral infarction, and dementia. Yes, it was confirmed that it is easy to be judged when complaining of cold symptoms, headaches, and allergic reactions.
When Fast Doctor receives a phone call from a patient, it determines the degree of urgency according to the criteria compiled by the Fire and Disaster Management Agency, and dispatches a doctor to the patient who needs to be seen within six hours and who finds it difficult to go to the hospital. , a certain number of under-triage had occurred.When under-triage occurs, there is a possibility that hospital visits will be delayed and subsequent illnesses will become more severe, which is a challenge for emergency home-call services.
Paper information:[Annals of Medicine] Machine Learning Models Predicting Undertriage in Telephone Triage
