Machine learning-based analysis of regional differences in out-of-hospital cardiopulmonary arrest outcomes and resuscitation interventions in Japan

Abstract

Refining out-of-hospital cardiopulmonary arrest (OHCA) resuscitation protocols for local emergency practices is vital. The lack of comprehensive evaluation methods for individualized protocols impedes targeted improvements. Thus, we employed machine learning to assess emergency medical service (EMS) records for examining regional disparities in time reduction strategies. In this retrospective study, we examined Japanese EMS records and neurological outcomes from 2015 to 2020 using nationwide data. We included patients aged ≥ 18 years with cardiogenic OHCA and visualized EMS activity time variations across prefectures. A five-layer neural network generated a neurological outcome predictive model that was trained on 80% of the data and tested on the remaining 20%. We evaluated interventions associated with changes in prognosis by simulating these changes after adjusting for time factors, including EMS contact to hospital arrival and initial defibrillation or drug administration. The study encompassed 460,540 patients, with the model's area under the curve and accuracy being 0.96 and 0.95, respectively. Reducing transport time and defibrillation improved outcomes universally, while combining transport time and drug administration showed varied efficacy. In conclusion, the association of emergency activity time with neurological outcomes varied across Japanese prefectures, suggesting the need to set targets for reducing activity time in localized emergency protocols.

Figure 1

Overview of data splitting and stratified cross-validation methods and the neural network-based machine learning model. The model was developed using the stratified cross-validation method with CPC1/2. The machine learning model consisted of a five-layer neural network. AUROC—area under the receiver operating characteristic; BN—batch normalization; CPC—cerebral performance category.

Figure 2

Associations of changes in EMS activity time with predicted CPC1/2 in 47 prefectures. No adjustments are represented by square boxes. The color bar indicates the increase or decrease in predicted CPC1/2 relative to the unadjusted case. The left panel displays adjusted results for EMS arrival to hospital arrival time and to first defibrillation time. The right panel presents adjusted results for EMS arrival to hospital arrival time and EMS arrival to first drug administration time. In both scenarios, shorter activity times improved prognosis, while longer activity times worsened it. However, the changes ranged from − 20 to + 30 and from − 10 to + 5 for each factor. CPC—cerebral performance category; EMS—emergency medical service

Figure 3

Example of the association of changes in EMS arrival to hospital arrival time and defibrillation time with predicted CPC1/2. No adjustments are represented by square boxes. The color bar indicates the increase or decrease in predicted CPC1/2 relative to the unadjusted case. The figure demonstrates a consistent observation across all 47 prefectures that a decrease in the time intervals between EMS arrival to hospital arrival time and to first defibrillation time is anticipated to enhance patient prognosis. The observed changes spanned from − 20 to + 30 and − 10 to + 5. EMS—emergency medical service; CPC—cerebral performance category.

Figure 4

Example of the association of changes in EMS arrival to hospital arrival time and administration time with predicted CPC1/2. No adjustments are represented by square boxes. The color bar indicates the increase or decrease in predicted CPC1/2 relative to the unadjusted case. In the prefecture shown in the left panel, shortened EMS arrival to hospital arrival time was associated with improved prognosis; no association was seen with EMS arrival to drug administration time. However, in the prefecture shown on the right, earlier drug administration improved prognosis more than shorter EMS arrival to hospital arrival time. In contrast to defibrillation, different associations were observed in different prefectures. Changes ranged from − 7 to + 5 and − 10 to + 15 for each factor. EMS—emergency medical service; CPC—cerebral performance category.