Enhancing Cardiopulmonary Resuscitation Quality Using a Smartwatch: Neural Network Approach for Algorithm Development and Validation

BACKGROUND

Sudden cardiac arrest is a major cause of mortality, necessitating immediate and high-quality cardiopulmonary resuscitation (CPR) for improved survival rates. High-quality CPR is defined by chest compressions at a rate of 100-120 per minute and a depth of 50-60 mm. Monitoring and maintaining these parameters in real time during emergencies remain a challenge.

OBJECTIVE

This study introduces a neural network model designed to predict and assess CPR quality using accelerometer data from a smartwatch.

METHODS

The study involved 83 participants performing CPR on mannequins, with accelerometer data collected via smartwatches worn by the participants. These data were aligned with gold-standard data from the mannequins. The accelerometer-derived compression data were segmented into 5-second intervals for training the neural network models. A total of 1226 neural network models were developed, incorporating variations in hyperparameters and dataset configurations to optimize performance.

RESULTS

The optimal model demonstrated the capability to accurately predict the number of compressions and the average compression depth within a 5-second interval. The model achieved an accuracy of ±3.8 mm for compression depth and an average deviation of 0.8 compressions. The results indicated that the neural network model could accurately assess CPR quality metrics, surpassing other models discussed in the literature. The large and diverse dataset used in this study contributed to the robustness and reliability of the model.

CONCLUSIONS

This study validates the efficacy of a neural network model in accurately predicting CPR metrics using smartwatch accelerometer data. The model outperforms previous methods and shows promise for real-time feedback during CPR. Future work involves deploying the model directly on smartwatches for real-time application, potentially improving sudden cardiac arrest survival rates through immediate and accurate feedback on CPR quality.

Impact of a Smart-Ring-Based Feedback System on the Quality of Chest Compressions in Adult Cardiac Arrest: A Randomized Preliminary Study

This study aimed to assess the effectiveness of a novel chest compression (CC) smart-ring-based feedback system in a manikin simulation. In this randomized, crossover, controlled study, we evaluated the effect of smart-ring CC feedback on cardiopulmonary resuscitation (CPR). The learnability and usability of the tool were evaluated with the System Usability Scale (SUS). Participants were divided into two groups and each performed CCs with and without feedback 2 weeks apart, using different orders. The primary outcome was compression depth; the proportion of accurate-depth (5–6 cm) CCs, CC rate, and the proportion of complete CCs (≤1 cm of residual leaning) were assessed additionally. The feedback group and the non-feedback group showed significant differences in compression depth (52.1 (46.3–54.8) vs. 47.1 (40.5–49.9) mm, p = 0.021). The proportion of accurate-depth CCs was significantly higher in the interventional than in the control condition (88.7 (30.0–99.1) vs. 22.6 (0.0–58.5%), p = 0.033). The mean SUS score was 83.9 ± 8.7 points. The acceptability ranges were ‘acceptable’, and the adjective rating was ‘excellent’. CCs with smart-ring feedback could help achieve the ideal range of depth during CPR. The smart-ring may be a valuable source of CPR feedback.

Effect of smartphone applications on cardiopulmonary resuscitation quality metrics in a mannequin study: A randomized trial

OBJECTIVE:

The aim of this randomized, cross-over trial is to reveal the effect of smartphone cardio-pulmonary resuscitation (CPR) feedback applications (App) on a group of lay rescuers' chest compression-only CPR quality metrics. Quality metrics is measured initially and after 3 months.

METHODS:

A floor-based Resusci Anne mannequin (Laerdal Medical, Stavanger, Norway) was used. Three scenarios (CPR with device App-on [scenario-a], CPR with device App-off [scenario-b], and hands-only CPR [scenario-c]) were randomly allocated to all participants. All the participants performed 2 min of hands only-CPR for each scenario. Data of mean chest compression rate, mean chest compression depth, and recoil were recorded and compared for each scenario.

RESULTS:

One hundred and thirty-seven first-year students from the Vocational School of Health Services in Turkey participated in this study to mimic lay rescuers. Difference in the initial mean rate of chest compressions was statistically significant when CPR was performed with device App-on (scenario-a) compared to scenarios b and c (P < 0.001, P < 0.001). Furthermore, difference in the mean chest compression rate at the 3^rd month was statistically significant among the scenarios when CPR was performed with device App-on (scenario-a) (P = 0.002, P = 0.001). The difference in initial and 3^rd month mean compression depth and the percentage of recoil was not statistically significant among the scenarios.

CONCLUSION:

This study shows that the mean chest compression rate and percentage of compressions with adequate rate improved with smartphone App-on, and these results were persistent up to 3 months.

Development of Smart-Ring-Based Chest Compression Depth Feedback Device for High Quality Chest Compressions: A Proof-of-Concept Study

Recently, a smart-device-based chest compression depth (CCD) feedback system that helps ensure that chest compressions have adequate depth during cardiopulmonary resuscitation (CPR) was developed. However, no CCD feedback device has been developed for infants, and many feedback systems are inconvenient to use. In this paper, we report the development of a smart-ring-based CCD feedback device for CPR based on an inertial measurement unit, and propose a high-quality chest compression depth estimation algorithm that considers the orientation of the device. The performance of the proposed feedback system was evaluated by comparing it with a linear variable differential transformer in three CPR situations. The experimental results showed compression depth errors of 2.0 ± 1.1, 2.2 ± 0.9, and 1.4 ± 1.1 mm in the three situations. In addition, we conducted a pilot test with an adult/infant mannequin. The results of the experiments show that the proposed smart-ring-based CCD feedback system is applicable to various chest compression methods based on real CPR situations.

Smartwatch feedback device for high-quality chest compressions by a single rescuer during infant cardiac arrest: a randomized, controlled simulation study

OBJECTIVE

According to the guidelines, rescuers should provide chest compressions (CC) ~1.5 inches (40 mm) for infants. Feedback devices could help rescuers perform CC with adequate rates (CCR) and depths (CCD). However, there is no CC feedback device for infant cardiopulmonary resuscitation (CPR). We suggest a smartwatch-based CC feedback application for infant CPR.

PARTICIPANTS AND METHODS

We created a smartwatch-based CC feedback application. This application provides feedback on CCD and CCR by colour and text for infant CPR. To evaluate the application, 30 participants were divided randomly into two groups on the basis of whether CC was performed with or without the assistance of the smartwatch application. Both groups performed continuous CC-only CPR for 2 min on an infant mannequin placed on a firm table. We collected CC parameters from the mannequin, including the proportion of correct depth, CCR, CCD and the proportion of correct decompression depth.

RESULTS

Demographics between the two groups were not significantly different. The median (interquartile range) proportion of correct depth was 99 (97-100) with feedback compared with 83 (58-97) without feedback (P = 0.002). The CCR and proportion of correct decompression depth were not significantly different between the two groups (P = 0.482 and 0.089). The CCD of the feedback group was significantly deeper than that of the control group [feedback vs. control: 41.2 (39.8-41.7) mm vs. 38.6 (36.1-39.6) mm; P=0.004].

CONCLUSION

Rescuers who receive feedback of CC parameters from a smartwatch could perform adequate CC during infant CPR.

Effect of smart devices on the quality of CPR training: A systematic review

AIM OF THE REVIEW

Use of smart devices to provide real-time cardiopulmonary resuscitation (CPR) feedback in the context of out-of-hospital cardiac arrest (OHCA) has considerable potential for improving survival. However, the findings of previous studies evaluating the effectiveness of these devices have been conflicting. Therefore, we conducted a systematic review of the literature to assess the utility of smart devices for improving the quality of CPR during CPR training.

DATA SOURCES

Thirteen electronic databases were searched. The articles were reviewed according to the eligibility criteria. CPR quality was evaluated based on the rates and depths of chest compression, and the proportion of adequate depth of chest compressions.

RESULTS

Ultimately, 11 studies (5 randomised controlled trials, 1 randomised trial, and 5 randomised cross-over trials) were selected for this systematic review. Eight of these studies used smartphones and three used smartwatches. This review did not find an apparent benefit from smart device use during CPR in terms of maintaining the recommended compression rates and depths of chest compressions. However, all three smartwatch studies reported that the proportion of chest compressions of adequate depth was significantly improved with smartwatch use (smartwatch group vs. non-smartwatch group in the three studies: 65.01% vs. 45.15%, p = 0.01; 64.6% vs. 43.1%, p = 0.049; 98.7% vs. 79.3%, p = 0.002).

CONCLUSION

This review does not find durable evidence for usefulness of smart devices in CPR training. However, the smartwatches may improve the accuracy of chest compression depth. Future studies with larger sample sizes might be necessary before reaching a firm conclusion.

Smartwatches as chest compression feedback devices: A feasibility study

BACKGROUND

Recently, there have been attempts to use smartphones and smartwatches as the feedback devices to improve the quality of chest compressions. In this study, we compared chest compression depth feedback accuracy between a smartphone and a smartwatch in a hands-only cardiopulmonary resuscitation scenario, using a manikin with a displacement sensor system.

METHODS

Ten basic life support providers participated in this study. Guided by the chest compression depths displayed on the monitor of a laptop, which received data from the manikin, each participant performed 2min of chest compressions for each target depth (35mm and 55mm) on a manikin while gripping a smartphone and wearing a smartwatch. Participants had a rest of 1h between the instances, and the first target depth was set at random. Each chest compression depth data value from the smartphone and smartwatch and a corresponding reference value from the manikin with the displacement system were recorded. To compare the accuracy between the smartphone and smartwatch, the errors, expressed as the absolute of the differences between the reference and each device, were calculated.

RESULTS

At both target depths, the error of the smartwatch were significantly smaller than that of the smartphone (the errors of the smartphone vs. smartwatch at 35mm: 3.4 (1.3) vs. 2.1 (0.8) mm; p=0.008; at 55mm: 5.3 (2.8) vs. 2.3 (0.9) mm; p=0.023).

CONCLUSION

The smartwatch-based chest compression depth feedback was more accurate than smartphone-based feedback.

Effectiveness of chest compression feedback during cardiopulmonary resuscitation in lateral tilted and semirecumbent positions: a randomised controlled simulation study

Feedback devices have been shown to improve the quality of chest compression during cardiopulmonary resuscitation for patients in the supine position, but no studies have reported the effects of feedback devices on chest compression when the chest is tilted. Basic life support-trained providers were randomly assigned to administer chest compressions to a manikin in the supine, 30° left lateral tilt and 30° semirecumbent positions, with or without the aid of a feedback device incorporated into a smartphone. Thirty-six participants were studied. The feedback device did not affect the quality of chest compressions in the supine position, but improved aspects of performance in the tilted positions. In the lateral tilted position, the median (IQR [range]) chest compression rate was 99 (99-100 [96-117]) compressions.min(-1) with and 115 (95-128 [77-164]) compressions.min(-1) without feedback (p = 0.05), and the proportion of compressions of correct depth was 55 (0-96 [0-100])% with and 1 (0-30 [0-100])% without feedback (p = 0.03). In the semirecumbent position, the proportion of compressions of correct depth was 21 (0-87 [0-100])% with and 1 (0-26 [0-100])% without feedback (p = 0.05). Female participants applied chest compressions at a more accurate rate using the feedback device in the lateral tilted position but were unable to increase the chest compression depth, whereas male participants were able to increase the force of chest compression using the feedback device in the lateral tilted and semirecumbent positions. We conclude that a feedback device improves the application of chest compressions during simulated cardiopulmonary resuscitation when the chest is tilted.