Out-of-Hospital Cardiac Arrest Detection by Machine Learning Based on the Phonetic Characteristics of the Caller's Voice.

INTRODUCTION: Out-of-hospital cardiac arrest (OHCA) is a major public health issue. The prognosis is closely related to the time from collapse to return of spontaneous circulation. Resuscitation efforts are frequently initiated at the request of emergency call center professionals who are specifically trained to identify critical conditions over the phone. However, 25% of OHCAs are not recognized during the first call. Therefore, it would be interesting to develop automated computer systems to recognize OHCA on the phone. The aim of this study was to build and evaluate machine learning models for OHCA recognition based on the phonetic characteristics of the caller's voice. METHODS: All patients for whom a call was done to the emergency call center of Rennes, France, between 01/01/2017 and 01/01/2019 were eligible. The predicted variable was OHCA presence. Predicting variables were collected by computer-automatized phonetic analysis of the call. They were based on the following voice parameters: fundamental frequency, formants, intensity, jitter, shimmer, harmonic to noise ratio, number of voice breaks, and number of periods. Three models were generated using binary logistic regression, random forest, and neural network. The area under the curve (AUC) was the primary outcome used to evaluate each model performance. RESULTS: 820 patients were included in the study. The best model to predict OHCA was random forest (AUC=74.9, 95% CI=67.4-82.4). CONCLUSION: Machine learning models based on the acoustic characteristics of the caller's voice can recognize OHCA. The integration of the acoustic parameters identified in this study will help to design decision-making support systems to improve OHCA detection over the phone.

Machine learning is the key to diagnose COVID-19: a proof-of-concept study.

The reverse transcription-polymerase chain reaction (RT-PCR) assay is the accepted standard for coronavirus disease 2019 (COVID-19) diagnosis. As any test, RT-PCR provides false negative results that can be rectified by clinicians by confronting clinical, biological and imaging data. The combination of RT-PCR and chest-CT could improve diagnosis performance, but this would requires considerable resources for its rapid use in all patients with suspected COVID-19. The potential contribution of machine learning in this situation has not been fully evaluated. The objective of this study was to develop and evaluate machine learning models using routine clinical and laboratory data to improve the performance of RT-PCR and chest-CT for COVID-19 diagnosis among post-emergency hospitalized patients. All adults admitted to the ED for suspected COVID-19, and then hospitalized at Rennes academic hospital, France, between March 20, 2020 and May 5, 2020 were included in the study. Three model types were created: logistic regression, random forest, and neural network. Each model was trained to diagnose COVID-19 using different sets of variables. Area under the receiving operator characteristics curve (AUC) was the primary outcome to evaluate model's performances. 536 patients were included in the study: 106 in the COVID group, 430 in the NOT-COVID group. The AUC values of chest-CT and RT-PCR increased from 0.778 to 0.892 and from 0.852 to 0.930, respectively, with the contribution of machine learning. After generalization, machine learning models will allow increasing chest-CT and RT-PCR performances for COVID-19 diagnosis.

Evaluation of HEAR score to rule-out major adverse cardiac events without troponin test in patients presenting to the emergency department with chest pain.

BACKGROUND AND IMPORTANCE: Current guidelines for patients presenting to the emergency department with chest pain without ST-segment elevation myocardial infarction (non-STEMI) on electrocardiogram are based on troponin measurement. The HEART score is reportedly a reliable work-up strategy that combines clinical evaluation with troponin value. A clinical rule that could select very low-risk patients without the need for a blood test (HEAR score, being the HEART score without the troponin item) would be of great interest. OBJECTIVES: To prospectively assess the safety of a HEAR score <2 to rule-out non-STEMI without troponin measurement. Secondary objective was to assess the safety of a sequential strategy that combines HEAR score and HEART (defined as two-step HEART strategy). DESIGN, SETTINGS AND PARTICIPANTS: Prospective observational study in six emergency departments. Patients with nontraumatic chest pain and no alternative diagnosis were included and followed up for 45 day. Patients were considered at low-risk if the HEAR score was <2 or, for the two-step HEART strategy, if the HEART score was <4. OUTCOMES MEASURE AND ANALYSIS: The primary endpoint was the 45-day rate of major adverse cardiac events (MACE) in patients with a HEAR score <2. A HEAR score based strategy was consider safe if the rate of the primary endpoint was below 1%, with an upper margin of the 95% confidence interval (CI) below 3%. RESULTS: Among 1452 patients included, 1402 were analyzed and 97 (7%) had a MACE during the follow-up period. The HEAR score was <2 in 279 (20%) patients and one presented a MACE [0.4% (95% CI: 0.01-1.98)]. The two-step HEART strategy classified low-risk an additional 476 patients (34%) and one of these 476 patients had a MACE [0.3% (95% CI: 0.03-0.95)]. The two-step HEART strategy would have theoretically avoided 360 troponin measurements (19%). CONCLUSIONS: In our prospective multicenter study, a HEAR based work-up strategy was safe, with a very low risk of MACE at 45 day. We also report that a two-step HEART-based strategy may safely allow significant reduction of troponin measurements in patients presenting to the emergency department with chest pain.

Exclusive ambulatory management of spontaneous pneumothorax with pigtail catheters, a prospective multicentric study.

OBJECTIVE: Spontaneous pneumothorax occurs most frequently in young active patients. Published guidelines do not all agree about its initial management; most patients are hospitalised and treated with chest tube. This prospective multicentric cohort study was designed to assess the potential of ambulatory management. METHODS: We included all consecutive patients with large spontaneous primary (PSP) and secondary pneumothoraces (SSP) presenting at the Lorient, Vitré and Rennes hospitals between December 2013 and July 2016. They were treated with a small-bore pigtail catheter and one-way valve and managed as outpatients following a specific protocol. When this failed, patients were hospitalised on day 4 for suction and surgical pleurodesis was envisaged on day 6. Patients were followed-up for one-year to assess relapse. RESULTS: Of the 148 patients included (129 PSP, 19 SSP), 122 (82⋅4%) were managed exclusively as outpatient with success in 84⋅5% of PSP and 68⋅4% of SSP patients. There were few complications: 13 vaso-vagal episodes and 3 minor bleedings. The one-year recurrence rates were 33⋅1% for PSP and 52⋅6% for SSP (p = 0⋅114 Hazard Ratio = 0⋅538; IC95% [0⋅249-1⋅161]). CONCLUSION: These results are consistent with our previous study and confirm that this exclusive ambulatory management of spontaneous pneumothoraces can be successfully implemented in new centres with a high success rate and few complications.

New considerations on pathways involved in acute traumatic coagulopathy: the thrombin generation paradox.

Background: An acute traumatic coagulopathy (ATC) is observed in about one third of severely traumatized patients. This early, specific, and endogenous disorder is triggered by the association of trauma and hemorrhage. The early phase of this condition is characterized by the expression of a bleeding phenotype leading to hemorrhagic shock and the late phase by a prothrombotic profile leading to multiple organ failure. The physiopathology of this phenomenon is still poorly understood. Hypotheses of disseminated intravascular coagulation, activated protein C-mediated fibrinolysis, fibrinogen consumption, and platelet functional impairment were developed by previous authors and continue to be debated. The objective of this study was to observe general hemostasis disorders in case of ATC to confront these hypotheses. Method: Four groups of 15 rats were compared: C, control; T, trauma; H, hemorrhage; and TH, trauma and hemorrhage. Blood samples were drawn at baseline and 90 min. Thrombin generation tests, platelet aggregometry, and standard hemostasis tests were performed. Results: Significant differences were observed between the baseline and TH groups for aPTT (17.9 ± 0.8 s vs 24.3 ± 1.4 s, p < 0.001, mean ± SEM), MAP (79.7 ± 1.3 mmHg vs 43.8 ± 1.3 mmHg, p < 0.001, mean ± SEM), and hemoglobin (16.5 ± 0.1 g/dL vs 14.1 ± 0.3 g/dL, p < 0.001, mean ± SEM), indicating the presence of an hemorrhagic shock due to ATC. Compared to all other groups, coagulation factor activities were decreased in the TH group, but endogenous thrombin potential was (paradoxically) higher than in group C (312 ± 17 nM/min vs. 228 ± 23 nM/min; p = 0.016; mean ± SEM). We also observed a subtle decrease in platelet count and function in case of ATC and retrieved an inversed linear relationship between fibrinogen concentration and aPTT (intercept, 26.53 ± 3.16; coefficient, - 3.40 ± 1.26; adjusted R2: 0.1878; p = 0.0123). Conclusions: The clinical-biological profile that we observed, combining normal thrombin generation, fibrinogen depletion, and a hemorrhagic phenotype, reinforced the hypothesis of activated protein C mediated-fibrinolysis. The key role of fibrinogen, but not of the platelets, was confirmed in this study. The paradoxical preservation of thrombin generation suggests a protective mechanism mediated by rhabdomyolysis in case of major trauma. Based on these results, we propose a new conception concerning the pathophysiology of ATC.