Expecting the Unexpected: Predicting Panic Attacks From Mood, Twitter, and Apple Watch Data

OBJECTIVE

Panic attacks are an impairing mental health problem that affects 11% of adults every year. Current criteria describe them as occurring without warning, despite evidence suggesting individuals can often identify attack triggers. We aimed to prospectively explore qualitative and quantitative factors associated with the onset of panic attacks.

RESULTS

Of 87 participants, 95% retrospectively identified a trigger for their panic attacks. Worse individually reported mood and state-level mood, as indicated by Twitter ratings, were related to greater likelihood of next-day panic attack. In a subsample of participants who uploaded their wearable sensor data (n = 32), louder ambient noise and higher resting heart rate were related to greater likelihood of next-day panic attack.

CONCLUSIONS

These promising results suggest that individuals who experience panic attacks may be able to anticipate their next attack which could be used to inform future prevention and intervention efforts.

The Reliability of the Apple Watch's Electrocardiogram

Background An electrocardiogram (ECG) is a standard tool used to detect various cardiovascular abnormalities. Detection sensitivity for atrial fibrillation (AF) was recently shown to be greatly increased by using short, intermittent ECG recordings. Modern mobile ECG recording devices that can monitor patients' heart activities around the clock have made this a reality. The Apple Watch is one of these portable ECG devices that can detect heart rhythms and is approved by the American FDA for screening and detecting AF. Objectives We compared the results of the Apple Watch I lead ECG with conventional ECG results to assess the sensitivity and specificity of the Apple Watch I lead ECG. We then determined the abnormalities that can be detected by the Apple Watch I lead ECG. Methods This study was conducted on outpatient cardiac clinics at King Abdullah bin Abdulaziz University Hospital (KAAUH), and Prince Sultan Cardiac Center (PSCC), from May to October 2021. A standard 12-lead ECG was recorded and compared with the Apple Watch I lead ECG. A total of 469 ECG comparisons were included in this study and evaluated by two investigators. The data on patient demographics, medical and medication history, and ECG data were reviewed and analyzed using IBM SPSS Statistics for Windows, Version 23 (Released 2015; IBM Corp., Armonk, New York, United States). Results No significant differences were seen between the Apple Watch and the 12-lead ECG in terms of the studied ECG characteristics. A significant and strong positive correlation between the heart rate measurements in the 12-lead ECG and Apple Watch ECG was documented. The most commonly found abnormalities in the Apple Watch ECG were AF in 37 (7.9%), followed by first-degree atrioventricular (AV) block in 32 (6.8%). The sensitivity of Apple Watch's automated interpretation to detect an AF was 99.54%, while the manual interpretation yielded a sensitivity of 100%. Conclusion The results of this study demonstrated a robust relationship between the 12-lead ECG and Apple Watch ECG in the diagnosis of arrhythmias. Consequently, cardiac patients may consider the Apple Watch ECG a trustworthy remote monitoring technique.

Prophylactic relationship between mental health disorder symptoms and physical activity of Royal Canadian Mounted Police Cadets during the cadet training program

Objective

Royal Canadian Mounted Police report experiencing extremely frequent potentially psychologically traumatic events (PPTE). In a recent study, approximately half of participating RCMP screened positive for one or more mental disorders, which is approximately five times the diagnostic proportion for the general Canadian population. Increased reporting of mental health symptoms been linked to PPTE exposures. Programs promoting physical activity may be useful interventions to supplement or pair with mental health interventions, providing anxiolytic, antidepressant, and stress-buffering effects. The current study was designed to assess the relationship between physical activity behaviors and reported mental health disorder symptoms of cadets during the Royal Mounted Canadian Police (RCMP) Cadet Training Program (CTP). The current study also examined the relationship between exercise and mental health disorder symptoms of cadets during the CTP.

Methods

The study included data from 394 cadets (76.1% male). An analysis of variance (ANOVA) and a series of t-tests were used to assess several differences across sociodemographic groups. Bivariate Spearman's Rank correlations were performed between the average number of active calories burned per day, as recorded by Apple Watches, and changes in self-reported mental health disorder symptoms (i.e., Generalized Anxiety Disorder [GAD], Major Depressive Disorder [MDD], Posttraumatic Stress Disorder [PTSD], Social Anxiety Disorder [SAD]. Alcohol Use Disorders [AUD], Panic Disorder [PD]) from pre-training (starting the CTP) to pre-deployment (completing the CTP) 26 weeks later.

Results

There were statistically significant correlations between physical activity and self-reported mental health disorder symptom scores during CTP. Cadets who performed more physical activity from pre-training to pre-deployment had statistically significantly greater decreases in symptoms of GAD (ρ = -0.472, p < 0.001), MDD (ρ = -0.307, p < 0.001), PTSD (ρ = -0.343, p < 0.001), and AUD (ρ = -0.085, p < 0.05). There was no statistically significant relationship between physical activity and changes in PD symptoms (ρ = -0.037, p > 0.05). There were also no statistically significant relationships between pre-CTP mental health disorder symptom scores and the volume of physical activity performed during CTP.

Conclusion

There was evidence of a significant relationship between reductions in mental health disorder symptom scores and physical activity during the 26-week CTP. The results highlight the role that exercise can play as an important tool for reducing mental health disorder symptoms, considering there was no relationship between pre-CTP baseline mental health scores and physical activity performed during CTP. Further research is needed to understand differences in physical activity behaviours among cadets and serving RCMP.

Accuracy of the Apple Watch Oxygen Saturation Measurement in Adults: A Systematic Review

The purpose of this review is to summarize the research on the accuracy of oxygen saturation (spO2) measurements using the Apple Watch (Apple Inc., Cupertino, California). The Medline and Google Scholar databases were searched for papers evaluating the spO2 measurements of the Apple Watch vs. any kind of ground truth and records were analyzed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. The five publications with 973 total patients that met the inclusion criteria all used the Apple Watch Series 6 and described 95% limits of agreement of +/- 2.7 to 5.9% spO2. However, outliers of up to 15% spO2 were reported. Only one study had patient-level data uploaded to a public repository. The Apple Watch Series 6 does not show a strong systematic bias compared to conventional, medical-grade pulse oximeters. However, outliers do occur and should not cause concern in otherwise healthy individuals. The impact of race on measurement accuracy should be investigated.

"suMus," a novel digital system for arm movement metrics and muscle energy expenditure

Objective: In the field of non-treatable muscular dystrophies, promising new gene and cell therapies are being developed and are entering clinical trials. Objective assessment of therapeutic effects on motor function is mandatory for economical and ethical reasons. Main shortcomings of existing measurements are discontinuous data collection in artificial settings as well as a major focus on walking, neglecting the importance of hand and arm movements for patients' independence. We aimed to create a digital tool to measure muscle function with an emphasis on upper limb motility. Methods: suMus provides a custom-made App running on smartwatches. Movement data are sent to the backend of a suMus web-based platform, from which they can be extracted as CSV data. Fifty patients with neuromuscular diseases assessed the pool of suMus activities in a first orientation phase. suMus performance was hence validated in four upper extremity exercises based on the feedback of the orientation phase. We monitored the arm metrics in a cohort of healthy volunteers using the suMus application, while completing each exercise at low frequency in a metabolic chamber. Collected movement data encompassed average acceleration, rotation rate as well as activity counts. Spearman rank tests correlated movement data with energy expenditure from the metabolic chamber. Results: Our novel application "suMus," sum of muscle activity, collects muscle movement data plus Patient-Related-Outcome-Measures, sends real-time feedback to patients and caregivers and provides, while ensuring data protection, a long-term follow-up of disease course. The application was well received from the patients during the orientation phase. In our pilot study, energy expenditure did not differ between overnight fasted and non-fasted participants. Acceleration ranged from 1.7 ± 0.7 to 3.2 ± 0.5 m/sec² with rotation rates between 0.9 ± 0.5 and 2.0 ± 3.4 rad/sec. Acceleration and rotation rate as well as derived activity counts correlated with energy expenditure values measured in the metabolic chamber for one exercise (r = 0.58, p < 0.03). Conclusion: In the analysis of slow frequency movements of upper extremities, the integration of the suMus application with smartwatch sensors characterized motion parameters, thus supporting a use in clinical trial outcome measures. Alternative methodologies need to complement indirect calorimetry in validating accelerometer-derived energy expenditure data.

Evaluation of a machine learning approach utilizing wearable data for prediction of SARS-CoV-2 infection in healthcare workers

Objective

To determine whether a machine learning model can detect SARS-CoV-2 infection from physiological metrics collected from wearable devices.

Materials and Methods

Health care workers from 7 hospitals were enrolled and prospectively followed in a multicenter observational study. Subjects downloaded a custom smart phone app and wore Apple Watches for the duration of the study period. Daily surveys related to symptoms and the diagnosis of Coronavirus Disease 2019 were answered in the app.

Results

We enrolled 407 participants with 49 (12%) having a positive nasal SARS-CoV-2 polymerase chain reaction test during follow-up. We examined 5 machine-learning approaches and found that gradient-boosting machines (GBM) had the most favorable validation performance. Across all testing sets, our GBM model predicted SARS-CoV-2 infection with an average area under the receiver operating characteristic (auROC) = 86.4% (confidence interval [CI] 84-89%). The model was calibrated to value sensitivity over specificity, achieving an average sensitivity of 82% (CI ±∼4%) and specificity of 77% (CI ±∼1%). The most important predictors included parameters describing the circadian heart rate variability mean (MESOR) and peak-timing (acrophase), and age.

Discussion

We show that a tree-based ML algorithm applied to physiological metrics passively collected from a wearable device can identify and predict SARS-CoV-2 infection.

Conclusion

Applying machine learning models to the passively collected physiological metrics from wearable devices may improve SARS-CoV-2 screening methods and infection tracking.

Accelerometry-Based Step Count Validation for Horse Movement Analysis During Stall Confinement

Quantitative tracking of equine movement during stall confinement has the potential to detect subtle changes in mobility due to injury. These changes may warn of potential complications, providing vital information to direct rehabilitation protocols. Inertial measurement units (IMUs) are readily available and easily attached to a limb or surcingle to objectively record step count in horses. The objectives of this study were: (1) to compare IMU-based step counts to a visually-based criterion measure (video) for three different types of movements in a stall environment, and (2) to compare three different sensor positions to determine the ideal location on the horse to assess movement. An IMU was attached at the withers, right forelimb and hindlimb of six horses to assess free-movement, circles, and figure-eights recorded in 5 min intervals and to determine the best location, through analysis of all three axes of the triaxial accelerometer, for step count during stall confinement. Mean step count difference, absolute error (%) and intraclass correlation coefficients (ICCs) were determined to assess the sensor's ability to track steps compared to the criterion measure. When comparing sensor location for all movement conditions, the right-forelimb vertical-axis produced the best results (ICC = 1.0, % error = 6.8, mean step count difference = 1.3) followed closely by the right-hindlimb (ICC = 0.999, % error = 15.2, mean step count difference = 1.8). Limitations included the small number of horse participants and the lack of random selection due to limited availability and accessibility. Overall, the findings demonstrate excellent levels of agreement between the IMU's vertical axis and the video-based criterion at the forelimb and hindlimb locations for all movement conditions.