In Resuscitation ; h5-index 66.0
BACKGROUND : A machine-learning model trained to recognize emergency calls regarding Out-of-Hospital Cardiac Arrest (OHCA) was tested in clinical practice at Copenhagen Emergency Medical Services (EMS) from September 2018 to December 2019. We aimed to investigate emergency call characteristics where the machine-learning model failed to recognize OHCA or misinterpreted a call as being OHCA.
METHODS : All emergency calls were linked to the dispatch database and verified OHCAs were identified by linkage to the Danish Cardiac Arrest Registry. Calls with either false negative or false positive predictions of OHCA were evaluated by trained auditors. Descriptive analyses were performed with absolute numbers and percentages reported.
RESULTS : The machine-learning model processed 169,236 calls to Copenhagen EMS and suspected 5,811 (3.4%) of the calls as OHCA, resulting in 84.5% sensitivity and 97.1% specificity. Among OHCAs not recognised by machine-learning model, a condition completely different from OHCA was presented by caller in 31% of the cases. In 28% of unrecognised calls, patient was reported breathing normally, and language barriers were identified in 23% of the cases. Among falsely suspected OHCA, the patient was reported unconscious in 28% of the cases, and in 13% of the false positive cases the machine-learning model interpreted calls regarding dead patients with irreversible signs of death as OHCA.
CONCLUSION : Continuous optimization of the language model is needed to improve the prediction of OHCA and thereby improve sensitivity and specificity of the machine-learning model on recognising OHCA in emergency telephone calls BACKGROUND: Survival after Out-of-Hospital Cardiac Arrest (OHCA) has increased in several countries after the implementation of initiatives to improve early recognition, dispatch protocol, bystander action and activation, as well as post-resuscitation care.[1-11] Steps to improve survival from OHCA have been summarized in the chain of survival and the Global Resuscitation Alliance '10 steps to improve survival'.[12] One of the central elements emphasised is the important role of early recognition of OHCA and the role of the dispatcher[4] who play a central role in early recognition and allocating rapid response to increase survival. In 15-20% of calls to Copenhagen Emergency Medical Services (EMS) regarding OHCA, resuscitation was initiated prior to the call.[3,13] For the remaining calls, it has been the dispatcher who establish clear recognition of OHCA and initiates a rapid response and dispatcher assisted cardiopulmonary resuscitation (DA-CPR).[3,14] The role of the dispatcher in recognizing OHCA has been proven to be challenging, as several aspects of telecommunication provide only limited insights to what is happening at the scene. Due to the inherent limitations of cognitive processing whilst communicating the dispatcher might miss recognizing a substantial number of OHCA or give delayed responses.[15] Several characteristics of OHCA have previously shown to present challenges in recognition with lower frequency of recognition and delayed or absent rapid response. While some characteristics has proven to ease recognition as suicide and trauma, other characteristics have shown to increase the difficulties of recognition namely difficulties in assessing breathing patterns and consciousness, language and linguistic challenges and the presence of seizures.[1,2,15-24]. To improve OHCA recognition, a machine-learning model was implemented into clinical practice in August 2018 at Copenhagen EMS.[14] The machine-learning model deciphered the conversation between dispatcher and caller and assisted the dispatcher in recognizing OHCA in the conversation in real-time. From September 2018 until December 2019 the machine-learning model alerted dispatchers in case of an ongoing emergency call showing a high probability of OHCA.[25] The aim of this study was to investigate and describe calls where the machine-learning model was unable to recognise OHCA or misinterpreted a call as being OHCA (false negatives and false positive OHCA calls).
METHODS : Copenhagen EMS serves 1.8 million inhabitants with an annual OHCA incidence of 80 per 100,000 persons and receives approximately 130,000 calls to the Danish emergency number (1-1-2), annually.[26] From September 1'st 2018 to December 31'st 2019, a machine-learning model was implemented that analysed all 1-1-2 emergency calls, deciphering the audio real time and alerting the dispatchers in case of a suspected OHCA[14,25]. The machine-learning model analysed all calls to 1-1-2, and information was stored on time of call, time of suspected OHCA and call-length. All calls were linked to the Danish Cardiac Arrest Register[11] to identify and verify calls as true or false OHCAs.
Nikolaj Blomberg Stig, Jensen Theo W, Porsborg Andersen Mikkel, Folke Fredrik, Kjær Ersbøll Annette, Torp-Petersen Christian, Lippert Freddy, Collatz Christensen Helle
2023-Jan-09
