Smartphone-based monitoring of heart rate variability and resting heart rate predicts variability in symptom exacerbations in people with complex chronic illness

Smartphone-based monitoring of heart rate variability and resting heart rate predicts variability in symptom exacerbations in people with complex chronic illness | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Smartphone-based monitoring of heart rate variability and resting heart rate predicts variability in symptom exacerbations in people with complex chronic illness Annie Aitken, Abbey Sawyer, Akiko Iwasaki, Harlan M. Krumholz, and 6 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5423422/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 24 Mar, 2026 Read the published version in npj Digital Medicine → Version 1 posted 10 You are reading this latest preprint version Abstract Background: Complex chronic conditions like Long COVID and Myalgic Encephalomyelitis/Chronic Fatigue Syndrome involve energy limitations and changes in heart rate variability (HRV) and resting heart rate (HR). Mobile health technologies now offer real-time, valid measurements of HRV and HR, advancing symptom monitoring and management. Using a high-density dataset from an observational longitudinal study, we aimed to describe, quantify, and predict within-person co-variations in daily biometric data and subsequent crash, fatigue, and brain fog symptom occurrences. Methods: Leveraging data collected through a mobile health app (n=4,244), we developed predictive models using mixed-effects linear regression and logistic regression to explore how within-person fluctuations in biometrics (HR, HRV, and respiratory rate) predict dynamic change in symptomology (crash, fatigue, and brain fog). Predictive performance was assessed using 5-fold stratified cross-validation and compared to a 20% holdout set to evaluate model generalizability to new observations and individuals. Results: Across all symptom domains, within-person changes in HRV and HR consistently emerged as key predictors of symptom change across all models, with higher HR and lower HRV conferring risk for crashes, fatigue, and brain fog. Moreover, 7-day biometric stability (or variable dispersion) was a robust predictor of symptom occurrence and severity. Models trained solely on biometric features achieved moderate predictive performance in the stratified cross-validation set; however, incorporating random effects to capture individual-specific variations and prior-day symptom reports substantially enhanced model accuracy, with AUC values reaching .91. Discussion and Conclusion: This study is the first to use data-driven models to predict everyday symptom experiences in individuals with complex chronic illnesses based on biometric fluctuations. Findings demonstrate the potential utility of mobile health tools for real-time monitoring of symptoms and highlight the need for further research to refine these predictive models and integrate them into clinical decision-making processes. Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Introduction Complex chronic illnesses such as Long COVID (LC) and Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS) can have a significant effect on the quality of life (1,2). The dynamic, often unpredictable nature of symptom severity and disability contributes to these declines in well-being. For example, approximately 85% of those affected by LC experience episodic symptoms that can rapidly fluctuate from periods of symptom stability to severe exacerbations, resulting in significant functional declines (3–5). These periods of severe worsening symptoms are often colloquially called “crashing” or “flare-ups” by those living with complex chronic illnesses. However, biological drivers of these symptom exacerbations have not been fully elucidated, and many times, people with complex chronic illnesses are unable to link a change in symptoms to a particular event. Overall, the inability of people with complex chronic illnesses to have measurable ways to predict periods of symptom exacerbation can lead to less effective disease management and, subsequently, reduced quality of life (5). Recent work suggests that heart rate variability (HRV) and other heart rate (HR) metrics may hold utility as potential biomarkers of complex chronic illness symptoms (6,7). Resting HR, or the number of successive heartbeats per minute, and HRV, the variability of R to R intervals measured linearly or in time/frequency domains, have been well identified as key metrics of overall health in healthy cohorts (8–10). Further, HRV is shown to be a strong indicator of cardiovascular health, physiological responsiveness to stress, and autonomic function (9,11,12). HRV, though generally non-specific, is influenced by multiple organ systems, including the cardiovascular, autonomic nervous, respiratory, and immune systems, making it a robust indicator of overall health. Given the diversity of organ systems and pathobiology implicated in the pathogenesis of complex chronic illnesses such as LC and ME/CFS, serial evaluation of HRV and HR may represent a novel approach to identifying and predicting symptom exacerbations (13). In laboratory environments, HRV, HR, and respiratory rate (RR) are typically measured using electrocardiograms (ECG) and respiration belts. These tools have been clinically validated to provided research-reliable estimates of sympathetic (SNS) and parasympathetic nervous system (PNS) activity, but are generally limited to controlled lab or clinical settings (14–16). Using ECG and respiration belts for managing health conditions is less practical in real-world settings due to their cumbersome nature and reliance on specialized equipment, making them difficult to implement. However, recent advances in mobile technologies have enabled naturalistic and user-friendly collection of cardiac and respiratory biomarkers of autonomic nervous system (ANS) control via photoplethysmography (PPG) sensors on smartphone cameras or wearable devices(17,18). These advances in digital health allow for the acquisition of self-reported daily fluctuations in symptoms alongside biometric proxies of physiological fluctuation. For example, biometrics and longitudinal symptom reporting have been pivotal in predicting symptom exacerbation in people with chronic diseases. One example is in people with chronic obstructive pulmonary disease (COPD). Using the myCOPD app, daily self-report data used in machine learning models could predict respiratory exacerbation in the subsequent days with moderate accuracy (19). Other digital health applications for people with COPD such as COPDPredict™ accurately predicted respiratory exacerbations with a high degree of accuracy using both self report and biometric inputs (20). These models have the potential to influence early treatment of respiratory exacerbations, which is associated with improved clinical outcomes, including exacerbation recovery time. Similar platforms developed for cystic fibrosis have determined that the use of simple self-reporting platforms can lead to earlier detection of respiratory exacerbations and treatment in the form of oral antibiotics in adults (21) and pediatrics (22). While the pathophysiology is vastly different, these studies raise a question of the potential influence of earlier detection on crash length and severity or complex chronic illness characterized by energy limitation. The Visible application is one such mobile application that has recently been developed that provides people with LC, ME/CFS, and other complex chronic conditions the ability to self-monitor symptoms and potentially relevant biometrics. The present study aims to conduct a high-level descriptive analysis of retrospectively collected data from the Visible application and to predict both within-person and between-person variations in daily biometric data in relation to subsequent evening reports of crash, fatigue, and brain fog. Methods No personal identifiable information was collected from this dataset. Thus, the dataset was determined to be exempt from human subject research by the Mount Sinai Program for the Protection of Human Subjects. However, all data presented here was derived from individuals who consented to provide their data for research purposes on the Visible application. Participants People aged ≥ 18 years with self-identified complex chronic illnesses such as LC, ME/CFS, or people experiencing other causes of energy limitation who were using the Visible application and opted to share their data anonymously were eligible for inclusion. Visible is a commercially available application that can be downloaded from major smartphone app stores, including both iOS and Android platforms. People self-identifying with LC were asked to confirm that they met the World Health Organization (WHO) definition of LC. Participants were included if they provided at least 14 biometric readings. Data used in this analysis was collected between August 2022 and April 2024. Outcomes Biometrics Resting heart rate (HR), heart rate variability (HRV), and respiration rate (RR) were measured using Photoplethysmography (PPG). In the unpaid version of the Visible application, users provide PPG reading using their smartphone camera. PPG via smartphone camera detects blood volume changes in the skin. When the user places their fingertip over the smartphone camera, the device emits light that penetrates the skin. As blood pulses through the vessels, it modulates the amount of light absorbed and reflected. The camera captures these changes in light intensity, and the resulting signal is analyzed to derive heart rate and other biometric data (18,23). Participants using a subscription version of the application (Visible) had the option of measuring their HR and HRV with a Polar Verity Sense™ armband (Polar Electro, Inc, Kempele, Finland). The Polar Verity Sense is a PPG-based optical heart rate sensor designed for upper-arm wear. The device was positioned with the sensor placed on the inner side of the armband, ensuring firm contact with the skin. Users are instructed to complete the biometric recording sitting or lying down and in the same position for each recording. HRV score is derived from calculating the root mean square of successive differences between heartbeats and scaling the score to receive an interpretable value between 0-100. Resting HR is calculated by measuring the number of heartbeats per minute. RR is calculated by detecting changes in heart rate related to breathing, where the heart rate increases during inhalation and decreases during exhalation. By leveraging these heart rate variations, the respiration rate is estimated. Additionally, the coefficient of variation (CoV) of an individual’s biometric values over the previous seven days to assess physiological stability and variability. Incidence of Crash A crash was defined within the app as periods of time when an individual's illness is significantly worse. The following explanation was provided to users: “Crashes usually occur as a part of post-exertional symptom exacerbation and affect your ability to carry out your usual activities. They normally last a few days. People use different words to describe these, such as flare-ups''. Each evening, users are prompted to respond with either a “✓” or an “X” to the item labeled “Crash” under the question “What else happened today?” This prompt allows users to indicate whether they experienced a crash that day. Check-marks were coded as 1 and Xs were coded as 0. Fatigue Each evening, users are prompted to log the degree of severity in fatigue. Fatigue was self-reported by participants on a 0 to 3 scale (0 = no impairment; 1 = mild; 2 = moderate; 3 = severe). Brain Fog Each evening, users are prompted to log the degree of severity in brain fog. Brain fog was self-reported by participants on a 0 to 3 scale (0 = no impairment; 1 = mild; 2 = moderate; 3 = severe). Figure 2 Description: Adults with complex chronic illnesses completed a biometric assessment (heart rate, HR; heart rate variability, HRV; respiratory rate, RR) each morning, followed by a self-report symptom survey in the evening. The diagram illustrates the model structure, where morning biometric scores (e.g., HRV) predict same-day evening symptom severity. This approach allows for within-day analysis of the relationship between physiological measures and symptom changes. Analysis Analyses were performed in R version 4.3.0(24) using the lme4 package (25) for multilevel modeling and tidymodels (26) for data preprocessing for predictive modeling. We conducted a multilevel regression analysis to predict same-day associations between morning biometrics and evening symptom reports. We disaggregated between-person (across participants) and within-person (within each participant) estimates of tonic and dynamic associations between biometrics and symptoms by partitioning predictor variances into between-person (group-level averages) and within-person (individual-level changes). Time-invariant, between-person predictors were grand-mean centered while time-varying, within-person predictors were group-mean centered. At the between-person level, we included all time-invariant predictors such as grand-mean biometric levels, stability scores, age, gender, and sensor modality. On the within-person level, we included all time-varying predictors, including the group-mean centered biometric level and stability scores, observation date, and autoregressive outcome value of the day prior. The autoregressive variable is included in the models to control for serial correlation effects of for when an individual experiences multiple days of elevated or diminished symptomology. Given the number of regression models and large sample size, only p-values < 0.001 are interpreted as significant to help control for inflated p-values that can occur in large samples. Model: $$\:{Y}_{i,j}={\gamma\:}_{00}+{u}_{0j}+{\gamma\:}_{10}*H{R}_{ij}+{\gamma\:}_{1j}*H{R}_{ij}+{\gamma\:}_{20}*HR{V}_{ij}+{\gamma\:}_{2j}*HR{V}_{ij}+{\gamma\:}_{30}*R{R}_{ij}+{\gamma\:}_{3j}*R{R}_{ij}$$ $$\:+{\gamma\:}_{40}*HR\_Co{V}_{ij}+{\gamma\:}_{4j\:}*HR\_Co{V}_{ij}\:+\:\:{\gamma\:}_{50}*HRV\_Co{V}_{ij}+{\gamma\:}_{5j\:}*HRV\_Co{V}_{ij}\:+\:{\gamma\:}_{90}*RR\_Co{V}_{ij}\:+\:{\gamma\:}_{6j}*RR\_Co{V}_{ij}$$ $$\:+{\gamma\:}_{7j}*dat{e}_{ij}+{\gamma\:}_{80}*ag{e}_{ij}+{\gamma\:}_{90}*gende{r}_{ij}+{\gamma\:}_{10j}*{Y}_{i,1-j}+{\gamma\:}_{110}*source\_devic{e}_{ij}+{\epsilon\:}_{ij}$$ In this multilevel model, \(\:Y\) represents the outcome variable (e.g., crash, fatigue, brain fog) for observation 𝑖 within participant 𝑗. The group-level intercept, denoted as \(\:{\gamma\:}_{00}\) ​, represents the average outcome across all participants when all predictors are at their average values. The term \(\:{u}_{0j}\) is the random intercept for participant 𝑗, capturing deviations in baseline outcomes for each participant compared to the overall average. This accounts for individual differences that are not explained by the fixed effects. The model includes both within-person and between-person components. The within-person effects are captured by the linear terms \(\:{\gamma\:}_{10}\) - \(\:{\gamma\:}_{110}\) . These terms represent the relationship between the fluctuations in each participant's biometric variables and their symptom outcome specific occasion. The between-person effects, represented by \(\:{\gamma\:}_{1j}\) - ​ \(\:{\gamma\:}_{10j}\) , denote the relationships between each participant's overall (between-person) average levels of time-invariant variables and their average symptom outcomes. These terms capture how individual differences in long-term average biometric states relate to average symptom reporting. In addition to the biometric variables, the model adjusts for other factors, including the observation date, age, gender, previous day's symptom reporting ( \(\:{Y}_{i,1-j}\) ), and the device used to collect biometric data. Finally, the residual error term \(\:{\epsilon\:}_{ij}\) ​ accounts for unobserved variability in symptom outcomes at the individual level that is not explained by the model's predictors. Predictive Model Evaluation We performed stratified blocked 5-fold cross-validation (CV) to assess model predictive model performance. We compared models with only biometric feature variables and models that included biometric and autoregressive outcome variables within our linear mixed model framework. We also compared prediction models that included the random effect term in the prediction model to those without the random effect term to evaluate how random intercepts for each user account for individual variance in an individual’s probability value. Classification was the primary goal, so fatigue and brain fog were recoded such that “no or mild symptoms” were coded as 0, and “moderate to severe symptoms” were coded as 1. The 5-fold CV with stratified partitions was based on individual participants and a 20% holdout set for external validation. In the stratified CV, each participant's data contributed to all folds, ensuring that performance estimates reflect the model's ability to generalize to new observations from individuals included in the training set. The holdout set consisted of participants not present in the training set, allowing assessment of the model's generalizability to entirely new individuals. Stratification and holdout procedures are presented in Fig. 3 . The primary performance metric was the mean of CV area under the receiver operating characteristic curve (ROC-AUC). Figure 3 Description: Figure illustrating the 5-fold cross-validation process for three example participants. The colored sections (blue and gold) represent training/testing for each stratified fold within the larger training set. The holdout set is not pictured. Results Descriptive Statistics Data from n = 4,244 Visible application users are presented. Table 1 presents descriptive statistics on the Visible dataset to reflect the breadth of the data used in the subsequent analyses. Descriptive graphics of the biometrics are presented in Fig. 3 . Of the full sample, 3,514 users tracked crashes, 3,915 users tracked fatigue, and 3,525 users tracked brain fog. Table 1 Dataset Information Per-User Characteristic Per User n = 4,244 1 Number of Biometric Readings Provided 125 (55, 257) Length of Sequential Biometric Readings Provided (without missing a day) 20 (10, 44) Total Reported Crash Episodes 15 (5, 42) Total Reported Fatigue Episodes 76 (32, 171) Total Reported Brain Fog Episodes 27 (6, 83) Duration of Crash Episodes (days) 4 (2, 10) Duration of Fatigue Episodes (days) 18 (8, 49) Duration of Brain Fog Episodes (days) 6 (2, 19) Days between Crash Episodes 5 (3, 11) Days between Fatigue Episodes 1.68 (1.17, 2.78) Days between Brain Fog Episodes 3 (1, 6) 1 Median (IQR) Figure 4 Description. Figure 4 a: A scatterplot displaying individual data points for heart rate (HR), heart rate variability (HRV), and respiratory rate (RR), illustrating the spread and distribution of the data across these variables. Figure 4 b: A bar plot showing the distribution of symptom outcomes (Crash, Fatigue, and Brain Fog). Fatigue and Brain Fog are dichotomously coded (0 = no or mild symptoms, 1 = moderate to severe symptoms). Figure 4 c: A correlation matrix displaying the correlation coefficients between HR, HRV, and RR, with corresponding color gradients representing the strength and direction of the relationships. Multilevel Model Regressions 1. Crash Outcome We first used a logistic MLM regression with random participant intercepts to model how fluctuation in daily biometrics predicted the likelihood of a reported crash that evening (Fig. 2 a). On the within-person level, decreases in HRV, increases in HR, and decreases in 7-day HRV and HR stability (increases in CoV) were associated with an increased likelihood of a reported crash that evening. On a between-person level, the likelihood of experiencing a crash was significantly associated with lower average HRV and lower long-term stability in HR (increased CoV). There was also a significant main effect of time, such as reports of crashes decreased across app usage. The autoregressive crash fixed effect was a highly significant predictor in the model. The full MLM model results for crashes are available in SI Table 1 . 2. Fatigue Outcome Next, we used MLM regression with random participant intercepts to model how daily biometrics fluctuation predicted the fatigue level that evening (Fig. 2 b). On the within-person level, decreases in HRV, increases in HR, and decreases in 7-day HRV, HR, and RR stability (increases in CoV) were associated with higher fatigue that evening. On the between-person level, fatigue was significantly associated with higher average HR and lower long-term stability in HRV and HR (increased CoV). Similarly, individuals who did not identify as male or female reported higher average levels of fatigue. The autoregressive fatigue fixed effect was a highly significant predictor in the model. The full MLM model results for fatigue are available in SI Table 2 . 3. Brain Fog Outcome Finally, we used MLM regression with random participant intercepts to model how daily biometrics fluctuation predicted the brain fog level reported that evening (Fig. 2 c). On the within-person level, decreases in HRV, increases in HR, and decreases in 7-day HRV and HR stability (increases in CoV) were associated with increased brain fog. On the between-person level, brain fog was significantly associated with higher average HR and lower long-term stability in HR (increased CoV). Individuals who did not identify as male or female reported higher average brain fog levels. The autoregressive brain fog fixed effect was a significant predictor in the model. The full MLM model results for brain fog are available in Fig. 5 a-c and SI Table 3 . Model fit statistics for predictors of crash, fatigue, and brain fog are presented in Table 2 . Figure 5 Description. Forest plot representing the standardized coefficients from multilevel regression models. 5a plots the standardized odds ratios when crash is modeled as the dependent variable; 5b plots the standardized beta coefficients when fatigue is modeled as the dependent variable; 5c plots the standardized beta coefficients when brain fog is modeled as the dependent variable. Table 2 Model Statistics for Predictors of Crash, Fatigue, and Brain Fog Crash Fatigue Brain Fog Intraclass Coefficient 0.29 0.29 0.36 Marginal R 2 / Conditional R 2 0.122 / 0.468 0.148 / 0.396 0.269 / 0.531 N (Total Obs) 399,010 540,564 474,047 Predictive Model Performance Next, we compared the predictive performance of multilevel random effects models with and without autoregressive variables in model training. To facilitate classification analysis, we dichotomized fatigue and brain fog into no or mild symptoms (coded as 0) and moderate to severe symptoms (coded as 1). In the stratified cross-validation (CV), performance was evaluated on new observations, enabling us to generalize results to new data from the same sample. We compared model performance under two conditions: (1) when autoregressive features were included during model training and (2) when random effects were incorporated during prediction. When random effects were excluded from the prediction process (i.e., predictions were made without accounting for individual-specific variation in intercepts), AUC values ranged from .60 to .61 when only biometric features were used and from .81 to .87 when both biometrics and autoregressive variables were included in the training. When random effects were included in prediction (i.e., individual-specific variations in intercepts were accounted for in predictions), AUC values ranged from .83 to .88 for models trained on biometrics alone and from .87 to .91 for models trained on both biometrics and autoregressive outcomes. Across all models, stratified CV demonstrated superior performance compared to holdout-set testing. This is likely because the holdout set could not use participant-specific estimates of average biometrics during prediction, leading to lower performance metrics. Table 3 Model Performance Metrics (Mean ROC-AUC) for Crash, Fatigue, and Brain Fog Using Stratified Cross-Validation and Hold Out Set Outcome Stratified CV (no-RE) Stratified CV (RE) Holdout Set (no-RE) Biometrics Only Crash .60 .83 .58 Fatigue .61 .86 .57 Brain Fog .61 .88 .57 Biometrics + Day Prior Outcome Crash .81 .87 .82 Fatigue .79 .87 .77 Brain Fog .85 .91 .84 Table 3 note . RE = Random Effects; CV = Cross-Validation Discussion This study analyzed a large (n = 4,244), longitudinal, high-frequency dataset of individuals self-reporting LC, ME/CFS, or other energy-limiting chronic conditions. Our findings revealed individual-level associations between morning biometric fluctuations and evening symptom reports, with model AUCs ranging from .60 to .91, depending on the inclusion of individual intercepts and previous-day symptom data. These results underscore the potential for improving care for individuals with complex chronic conditions through the targeted development of personalized, evidence-based remote physiological monitoring systems. Overall Predictive Performance Models trained solely on biometric features produced moderate predictive accuracy, but including random effects—capturing individual-specific intercepts—led to substantial improvements, with AUC values rising from .60-.61 to .83-.88. This suggests that accounting for individual differences improves model performance. The additional improvement in model performance when combining both biometric and previous-day self-report features underscores the impact of self-report data and the autocorrelative nature of symptom flair-ups. These findings highlight the importance of personalized approaches in health monitoring, where a one-size-fits-all model may fall short. Within-Person Biometric Sources of Variation in Symptoms Across all outcomes, within-person predictors of symptom changes were the strongest. We calculated daily fluctuations from baseline averages for heart rate variability (HRV), heart rate (HR), and respiration rate (RR), along with weekly changes in stability (CoV) for each biometric. RR was less predictive, contributing minimally to symptom prediction. Daily HRV and HR changes, along with 7-day biometric stability, emerged as key predictors, with higher HRV and lower HR associated with increased risk of crashes, fatigue, and brain fog. These findings are in line with past research that has found that individuals with chronic illness, including LC and ME/CFS, show alterations in HRV and HR compared with normative populations (14,27–29). More short-term variability in HRV and HR was linked to a higher probability of experiencing worsening symptoms, suggesting that fluctuations in cardiovascular dynamics over several days may destabilize daily symptom patterns. These results suggest that short-term fluctuations in HRV and HR dynamics, not just single-day metrics, are crucial for predicting symptom exacerbations. Between-Person Biometric Sources of Variation in Symptoms To a lesser extent than within-person predictors, between-person baseline biometric patterns also demonstrated predictive value for symptom experiences. Among these, the most consistent predictor was long-term variability in HR, where individuals with more stable HR patterns experienced fewer symptoms on average, including crashes, fatigue, and brain fog. This finding may reflect the role of HR stability as a marker of overall physiological resilience for people with complex chronic illnesses (30), where greater consistency in autonomic functioning reduces vulnerability to stressors that can trigger symptoms. Lower average HRV scores were linked to an increased likelihood of crashes but not fatigue or brain fog, suggesting that while HRV may be critical for predicting periods of acute stress (such as crashes), its influence on more chronic symptoms like fatigue and brain fog may be less pronounced. Conversely, higher average HR scores were associated with increased fatigue and brain fog, indicating that sustained elevations in HR could be a sign of prolonged physiological stress, contributing to chronic symptom experiences. Interestingly, we observed a decrease in reported crashes and the average level of brain fog reported by Visible app users over time, suggesting potential evidence of the mobile application's efficacy in helping users manage their condition better. Previous research has demonstrated that digital health tools can empower individuals to monitor their health patterns and make informed behavioral decisions, leading to improved health outcomes (20,31). However, future controlled studies are necessary to validate these findings and confirm the effectiveness of the Visible app. Speculation of mechanisms driving predictive HR and HRV dynamics Mechanistic inferences are difficult given the uncontrolled nature of the dataset and the multisystem nature of illnesses such as LC and ME/CFS but exploring potential mechanisms behind the observed associations between low HRV, high HR, and symptom crashes remains important. HRV and HR are commonly recognized as proxy measures of autonomic nervous system function: increased SNS activity raises HR and reduces HRV, while PNS activity lowers HRV and increases HR (33). One plausible mechanism involves the vagus nerve, which plays a key role in regulating inflammation and modulating central nervous system responses (34). A meta-analysis found a consistent negative relationship between HRV and markers of inflammation (35), thought to be mediated through the cholinergic anti-inflammatory pathway(36). Chronic inflammation can be regulated by the autonomic nervous system via this same cholinergic pathway and has been shown to decrease HRV and elevate resting HR. In conditions such as LC and ME/CFS, persistent pathogens, reactivation of latent viruses, onset of autoimmunity, dysregulation of cortisol and other hormones, mitochondrial dysfunction and endothelial dysfunction have all been reported and can all lead to chronic pro-inflammatory responses with the potential to cause significant daily fluctuations in HR and HRV (37–41). The potential for these mechanisms to not only cause chronic activation of inflammatory pathways, but also daily fluctuations in these activations could indeed explain how subsequent fluctuations in HRV and HR could be used to predict the emergence of crashes and other debilitating symptoms. More research to better understand how these various biological pathways can influence daily HRV and HR data could further validate how to use daily HRV and HR monitoring strategies to better manage symptom burden in conditions such as LC and ME/CFS. Limitations & Future Directions This study has several notable limitations, which offer valuable insights for future research directions. Due to the retrospective study design, limited information was collected on demographics, making the generalizability of the findings uncertain. Future studies should aim to gather more detailed demographic information to assess the applicability of findings across different populations. Similarly, participants in this study reported that they met the WHO criteria for LC, but there were no standardized criteria for reporting ME/CFS. Additionally, this study included individuals with other energy-limiting conditions, which introduces variability and challenges in defining the sample population. This led us to refer to the dataset as representing individuals with complex chronic illnesses rather than a specific condition. Participants measured biometrics using either smartphone cameras or armbands, which could introduce inconsistencies. While short heart-rate measurements via smartphone PPG recordings have been found reliable (42,43), future studies should standardize biometric data collection methods to ensure consistent and comparable results. Factors like time of day, temperature, skin tone, device type, software version, and recording length may have affected data accuracy. Participants were instructed to take measurements upon waking, but these factors could introduce variability in results. Despite these limitations, this study is the first to leverage data-driven, large-scale assessments of common symptoms among individuals with complex chronic illnesses. Conclusion Leveraging a natural intensive longitudinal data design from mobile health technologies, this study highlighted the potential of daily biometric monitoring to predict symptom fluctuations in individuals with complex chronic conditions. Within-person deviations in daily HRV and HR from a person’s baseline and changes in their biometric weekly stability were robust predictors of crash, fatigue, and brain fog. While these models demonstrated promising predictive performance among existing users, further work is needed to enhance applicability to new populations. These findings underscore the potential of digital health tools to improve real-time symptom tracking and management, offering valuable insights for the future development of personalized care strategies and remote health monitoring symptoms for LC, ME/CFS and other complex chronic illnesses. Abbreviations BMI: Body Mass Index COPD: Chronic obstructive pulmonary disease ECG: Electrocardiogram HR: Heart Rate HRV: Heart Rate Variability LC: Long COVID ME/CFS: Myalgic Encephalomyelitis/Chronic Fatigue Syndrome MLM: Multilevel Model Analysis PNS: Parasympathetic Nervous System POTS: Postural Orthostatic Tachycardia Syndrome RMSE: Root Mean Square Error SNS: Sympathetic Nervous System RR interval: Intervals between R waves in an ECG Declarations Conflicts of Interest Rory Preston is a Lead Data Scientist at Visible Health Inc. Harry Leeming is a Co-Founder at Visible Health Inc. Annie Brandes-Aitken received consulting fees at Visible Health Inc. References Thompson EJ, Williams DM, Walker AJ, Mitchell RE, Niedzwiedz CL, Yang TC, et al. Long COVID burden and risk factors in 10 UK longitudinal studies and electronic health records. Nat Commun. 2022 Jun 28;13(1):3528. Falk Hvidberg M, Brinth LS, Olesen AV, Petersen KD, Ehlers L. The health-related quality of life for patients with myalgic encephalomyelitis / chronic fatigue syndrome (ME/CFS). PLoS One. 2015 Jul 6;10(7):e0132421. Davis HE, Assaf GS, McCorkell L, Wei H, Low RJ, Re’em Y, et al. Characterizing long COVID in an international cohort: 7 months of symptoms and their impact. EClinicalMedicine. 2021 Aug;38(101019):101019. Humphreys H, Kilby L, Kudiersky N, Copeland R. Long COVID and the role of physical activity: a qualitative study. BMJ Open. 2021 Mar 10;11(3):e047632. O’Brien KK, Brown DA, McDuff K, St Clair-Sullivan N, Solomon P, Chan Carusone S, et al. Conceptualising the episodic nature of disability among adults living with Long COVID: a qualitative study. BMJ Glob Health. 2023 Mar;8(3):e011276. Van Campen CLM, Visser FC. higher resting heart rate myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) patients compared healthy controls: relation stroke volumes. Medical Research Archives. 2022;(10). Fang S-C, Wu Y-L, Tsai P-S. Heart rate variability and risk of all-cause death and cardiovascular events in patients with cardiovascular disease: A meta-analysis of cohort studies. Biol Res Nurs. 2020 Jan;22(1):45–56. Shaffer F, Ginsberg JP. An Overview of Heart Rate Variability Metrics and Norms. Front Public Health. 2017 Sep 28;5:258. Thayer JF, Ahs F, Fredrikson M, Sollers JJ 3rd, Wager TD. A meta-analysis of heart rate variability and neuroimaging studies: implications for heart rate variability as a marker of stress and health. Neurosci Biobehav Rev. 2012 Feb;36(2):747–56. Quer G, Gouda P, Galarnyk M, Topol EJ, Steinhubl SR. Inter- and intraindividual variability in daily resting heart rate and its associations with age, sex, sleep, BMI, and time of year: Retrospective, longitudinal cohort study of 92,457 adults. PLoS One. 2020 Feb 5;15(2):e0227709. Mongin D, Chabert C, Extremera MG, Hue O, Courvoisier DS, Carpena P, et al. Decrease of heart rate variability during exercise: An index of cardiorespiratory fitness. PLoS One. 2022 Sep 2;17(9):e0273981. Cole CR, Blackstone EH, Pashkow FJ, Snader CE, Lauer MS. Heart-rate recovery immediately after exercise as a predictor of mortality. N Engl J Med. 1999 Oct 28;341(18):1351–7. Mooren F, Böckelmann I, Waranski M, Kotewitsch M, Teschler M, Schäfer H, et al. Autonomic dysregulation in long-term patients suffering from Post-COVID-19 Syndrome assessed by heart rate variability. Sci Rep [Internet]. 2023 Sep 22;13. Available from: https://www.nature.com/articles/s41598-023-42615-y Barizien N, Le Guen M, Russel S, Touche P, Huang F, Vallée A. Clinical characterization of dysautonomia in long COVID-19 patients. Sci Rep. 2021 Jul 7;11(1):14042. da Silva RB, Neves VR, Montarroyos UR, Silveira MS, Sobral Filho DC. Heart rate variability as a predictor of mechanical ventilation weaning outcomes. Heart Lung. 2023 May;59:33–6. Rajendra Acharya U, Paul Joseph K, Kannathal N, Lim CM, Suri JS. Heart rate variability: a review. Med Biol Eng Comput. 2006 Dec;44(12):1031–51. Nelson BW, Low CA, Jacobson N, Areán P, Torous J, Allen NB. Guidelines for wrist-worn consumer wearable assessment of heart rate in biobehavioral research. NPJ Digit Med. 2020 Jun 26;3:90. Mather JD, Sculthorpe NF, Mair JL, Hayes LD. Validity of resting heart rate derived from contact-based smartphone photoplethysmography (PPG) compared with electrocardiography (ECG): A 2024 updated systematic review and meta-analysis of correlation coefficients. (Preprint) [Internet]. JMIR Preprints. 2024. Available from: http://dx.doi.org/10.2196/preprints.66662 Chmiel FP, Burns DK, Pickering JB, Blythin A, Wilkinson TM, Boniface MJ. Prediction of chronic obstructive pulmonary disease exacerbation events by using patient self-reported data in a digital health app: Statistical evaluation and machine learning approach. JMIR Med Inform. 2022 Mar 21;10(3):e26499. Patel ML, Wakayama LN, Bennett GG. Self-monitoring via digital health in weight loss interventions: A systematic review among adults with overweight or obesity. Obesity (Silver Spring). 2021 Mar 1;29(3):478–99. Wood J, Jenkins S, Putrino D, Mulrennan S, Morey S, Cecins N, et al. A smartphone application for reporting symptoms in adults with cystic fibrosis improves the detection of exacerbations: Results of a randomised controlled trial. J Cyst Fibros. 2020 Mar;19(2):271–6. van Horck M, Winkens B, Wesseling G, van Vliet D, van de Kant K, Vaassen S, et al. Early detection of pulmonary exacerbations in children with Cystic Fibrosis by electronic home monitoring of symptoms and lung function. Sci Rep. 2017 Sep 27;7(1):12350. Peng R-C, Zhou X-L, Lin W-H, Zhang Y-T. Extraction of heart rate variability from smartphone photoplethysmograms. Comput Math Methods Med. 2015 Jan 12;2015:516826. R Core Team. R: A Language and Environment for Statistical Computing. Vienna: R Foundation for Statistical Computing.; 2013. Bates D, Mächler M, Bolker B, Walker S. Fitting Linear Mixed-Effects Models using lme4 [Internet]. arXiv [stat.CO]. 2014. Available from: http://arxiv.org/abs/1406.5823 Kuhn M, Wickham H. Tidymodels: a collection of packages for modeling and machine learning using tidyverse principles [Internet]. Boston, MA, USA; 2020. Available from: https://www.tidymodels.org Proal AD, VanElzakker MB. Long COVID or post-acute sequelae of COVID-19 (PASC): An overview of biological factors that may contribute to persistent symptoms. Front Microbiol. 2021 Jun 23;12:698169. Proal AD, VanElzakker MB, Aleman S, Bach K, Boribong BP, Buggert M, et al. SARS-CoV-2 reservoir in post-acute sequelae of COVID-19 (PASC). Nat Immunol. 2023 Oct 4;24(10):1616–27. Ryabkova VA, Rubinskiy AV, Marchenko VN, Trofimov VI, Churilov LP. Similar patterns of dysautonomia in myalgic encephalomyelitis/chronic fatigue and post-COVID-19 syndromes. Pathophysiology. 2024 Jan 5;31(1):1–17. Mensink GBM, Hoffmeister H. The relationship between resting heart rate and all-cause, cardiovascular and cancer mortality. Eur Heart J. 1997 Sep 1;18(9):1404–10. Zhen J, Marshall JK, Nguyen GC, Atreja A, Narula N. Impact of digital health monitoring in the management of inflammatory bowel disease. J Med Syst. 2021 Jan 15;45(2):23. Cohen J. Statistical power analysis for the behavioral sciences [Internet]. Vol. 2nd, Statistical Power Analysis for the Behavioral Sciences. 1988. p. 567. Available from: http://dx.doi.org/10.1234/12345678 Berntson GG, Bigger JT Jr, Eckberg DL, Grossman P, Kaufmann PG, Malik M, et al. Heart rate variability: origins, methods, and interpretive caveats. Psychophysiology. 1997 Nov;34(6):623–48. Tracey KJ. The inflammatory reflex. Nature. 2002 Dec;420(6917):853–9. Williams DP, Koenig J, Carnevali L, Sgoifo A, Jarczok MN, Sternberg EM, et al. Heart rate variability and inflammation: A meta-analysis of human studies. Brain Behav Immun. 2019 Aug;80:219–26. Tracy LM, Ioannou L, Baker KS, Gibson SJ, Georgiou-Karistianis N, Giummarra MJ. Meta-analytic evidence for decreased heart rate variability in chronic pain implicating parasympathetic nervous system dysregulation. 2016 Jan 1;157(1):7–29. Peluso MJ, Deeks SG. Mechanisms of long COVID and the path toward therapeutics. Cell. 2024 Oct 3;187(20):5500–29. Iwasaki A, Putrino D. Why we need a deeper understanding of the pathophysiology of long COVID. Lancet Infect Dis. 2023 Apr;23(4):393–5. Klein J, Wood J, Jaycox JR, Dhodapkar RM, Lu P, Gehlhausen JR, et al. Distinguishing features of long COVID identified through immune profiling. Nature. 2023 Nov;623(7985):139–48. Turner S, Khan MA, Putrino D, Woodcock A, Kell DB, Pretorius E. Long COVID: pathophysiological factors and abnormalities of coagulation. Trends Endocrinol Metab. 2023 Jun;34(6):321–44. Appelman B, Charlton BT, Goulding RP, Kerkhoff TJ, Breedveld EA, Noort W, et al. Muscle abnormalities worsen after post-exertional malaise in long COVID. Nature communications. 2024;15(1):1–15. van Dijk W, Huizink AC, Oosterman M, Lemmers-Jansen ILJ, de Vente W. Validation of photoplethysmography using a mobile phone application for the assessment of heart rate variability in the context of heart rate variability-biofeedback. Psychosom Med. 2023 Sep 1;85(7):568–76. Bánhalmi A, Borbás J, Fidrich M, Bilicki V, Gingl Z, Rudas L. Analysis of a pulse rate variability measurement using a smartphone camera. J Healthc Eng. 2018 Feb 5;2018:4038034. Additional Declarations Competing interest reported. H.L. and R.P. are employed by Visible Inc. A.A. received consultation fees from Visible Inc. Supplementary Files CrashPredictionMECFSLCPaperSI.docx Cite Share Download PDF Status: Published Journal Publication published 24 Mar, 2026 Read the published version in npj Digital Medicine → Version 1 posted Editorial decision: Revision requested 03 Jan, 2025 Reviews received at journal 10 Dec, 2024 Reviewers agreed at journal 25 Nov, 2024 Reviews received at journal 25 Nov, 2024 Reviewers agreed at journal 18 Nov, 2024 Reviewers agreed at journal 18 Nov, 2024 Reviewers invited by journal 18 Nov, 2024 Editor assigned by journal 12 Nov, 2024 Submission checks completed at journal 12 Nov, 2024 First submitted to journal 09 Nov, 2024 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-5423422","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":382589213,"identity":"d449297a-0c97-4a91-91fd-0d9bf9ee7096","order_by":0,"name":"Annie Aitken","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA5ElEQVRIiWNgGAWjYDACZiCEMhgfACkePsJamOFamA1AWtiIsIYZxmKTAJOENPC38x82+LjDLs+cnflY5dccOxk2BuaHj27g0SJxmJk5ceaZ5GLLZra027LbkoEOYzM2zsFnDVDLYd425sQNh3nMbktuYwZq4WGTxqdFHqTlb1s9UAv/t2LJbfWEtRgAtSQzth0G2cLG+HHbYcJaDA8zGxv2th0HamEzlmbcdpyHjZmAX+TOH3ws8bOtOnHD+cMPP/7cVm3Pz9788DFe7yMDZh4wSaxyEGD8QYrqUTAKRsEoGDEAAA1nQS3eYas1AAAAAElFTkSuQmCC","orcid":"","institution":"New York University","correspondingAuthor":true,"prefix":"","firstName":"Annie","middleName":"","lastName":"Aitken","suffix":""},{"id":382589214,"identity":"db016f06-f57d-4bb4-93cf-9d7d35ceba4e","order_by":1,"name":"Abbey Sawyer","email":"","orcid":"","institution":"Icahn School of Medicine at Mount Sinai","correspondingAuthor":false,"prefix":"","firstName":"Abbey","middleName":"","lastName":"Sawyer","suffix":""},{"id":382589215,"identity":"f8d5caaf-f75a-4f86-915a-c8cec0938bb0","order_by":2,"name":"Akiko Iwasaki","email":"","orcid":"","institution":"Yale University","correspondingAuthor":false,"prefix":"","firstName":"Akiko","middleName":"","lastName":"Iwasaki","suffix":""},{"id":382589216,"identity":"524cda0a-fafe-4f1d-b080-88c50b693113","order_by":3,"name":"Harlan M. Krumholz","email":"","orcid":"","institution":"Yale-New Haven Hospital","correspondingAuthor":false,"prefix":"","firstName":"Harlan","middleName":"M.","lastName":"Krumholz","suffix":""},{"id":382589217,"identity":"b7fb337c-7d70-4c86-8686-33c00e4fa128","order_by":4,"name":"Rory Preston","email":"","orcid":"","institution":"Visible Health Inc","correspondingAuthor":false,"prefix":"","firstName":"Rory","middleName":"","lastName":"Preston","suffix":""},{"id":382589218,"identity":"bdfdc93d-4186-418f-8998-97ecddbe9eec","order_by":5,"name":"Harry Leeming","email":"","orcid":"","institution":"Visible Health Inc","correspondingAuthor":false,"prefix":"","firstName":"Harry","middleName":"","lastName":"Leeming","suffix":""},{"id":382589219,"identity":"fb30aaf0-8866-4674-9178-200f6ba3ce97","order_by":6,"name":"Jenna Tosto-Mancuso","email":"","orcid":"","institution":"Icahn School of Medicine at Mount Sinai","correspondingAuthor":false,"prefix":"","firstName":"Jenna","middleName":"","lastName":"Tosto-Mancuso","suffix":""},{"id":382589222,"identity":"6e3960f7-8e95-46da-a430-82f7d79c6a9f","order_by":7,"name":"Amy Proal","email":"","orcid":"","institution":"PolyBio Research Foundation","correspondingAuthor":false,"prefix":"","firstName":"Amy","middleName":"","lastName":"Proal","suffix":""},{"id":382589224,"identity":"f83b4fa3-b82c-495e-b1d0-3a52aa784180","order_by":8,"name":"Michael A. Osborne","email":"","orcid":"","institution":"University of Oxford","correspondingAuthor":false,"prefix":"","firstName":"Michael","middleName":"A.","lastName":"Osborne","suffix":""},{"id":382589226,"identity":"cff881fd-6769-4176-8a8a-84e527f15fdf","order_by":9,"name":"David Putrino","email":"","orcid":"","institution":"Icahn School of Medicine at Mount Sinai","correspondingAuthor":false,"prefix":"","firstName":"David","middleName":"","lastName":"Putrino","suffix":""}],"badges":[],"createdAt":"2024-11-09 20:08:09","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-5423422/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5423422/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41746-026-02543-3","type":"published","date":"2026-03-24T16:10:56+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":70202066,"identity":"462a0880-e855-4a2f-bbd8-c9bc222752f2","added_by":"auto","created_at":"2024-11-29 12:43:41","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":136635,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCONSORT Diagram\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-5423422/v1/2f9fc1cd5c23149c9620283d.png"},{"id":70202063,"identity":"9f826f5a-c0aa-4639-b255-622f53e7a658","added_by":"auto","created_at":"2024-11-29 12:43:41","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":224438,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eConceptual diagram of study design\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDescription: Adults with complex chronic illnesses completed a biometric assessment (heart rate, HR; heart rate variability, HRV; respiratory rate, RR) each morning, followed by a self-report symptom survey in the evening. The diagram illustrates the model structure, where morning biometric scores (e.g., HRV) predict same-day evening symptom severity. This approach allows for within-day analysis of the relationship between physiological measures and symptom changes.\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-5423422/v1/54dc44a4f6ec6e3f07d84386.png"},{"id":70202998,"identity":"985fe6cf-b635-450f-a707-074edc961e04","added_by":"auto","created_at":"2024-11-29 12:59:41","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":167904,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eStratification \u0026amp; Holdout Procedures\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDescription: Figure illustrating the 5-fold cross-validation process for three example participants. The colored sections (blue and gold) represent training/testing for each stratified fold within the larger training set. The holdout set is not pictured.\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-5423422/v1/38311057a67b03c17ae429f3.png"},{"id":70202067,"identity":"fc1427a0-7efc-4f09-aa9e-a34d0d03e1f0","added_by":"auto","created_at":"2024-11-29 12:43:41","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":321234,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDescriptives of Biometrics\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDescription. Figure 4a: A scatterplot displaying individual data points for heart rate (HR), heart rate variability (HRV), and respiratory rate (RR), illustrating the spread and distribution of the data across these variables. Figure 4b: A bar plot showing the distribution of symptom outcomes (Crash, Fatigue, and Brain Fog). Fatigue and Brain Fog are dichotomously coded (0 = no or mild symptoms, 1 = moderate to severe symptoms). Figure 4c: A correlation matrix displaying the correlation coefficients between HR, HRV, and RR, with corresponding color gradients representing the strength and direction of the relationships.\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-5423422/v1/b9974da7eb3e21e19b78b330.png"},{"id":70202175,"identity":"c7361ad5-e512-478e-b285-73f1a370e9d3","added_by":"auto","created_at":"2024-11-29 12:51:41","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":457576,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eStd. Beta Coefficients Confidence Intervals for Predictors in Multilevel Models\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDescription. Forest plot representing the standardized coefficients from multilevel regression models. 5a plots the standardized odds ratios when crash is modeled as the dependent variable; 5b plots the standardized beta coefficients when fatigue is modeled as the dependent variable; 5c plots the standardized beta coefficients when brain fog is modeled as the dependent variable.\u003c/p\u003e","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-5423422/v1/976cc8288e696b5d36f7f7e1.png"},{"id":105755699,"identity":"6fa493e0-b146-4ee3-be40-b6480735fe61","added_by":"auto","created_at":"2026-03-30 16:29:30","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2157987,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5423422/v1/34c650e4-0606-42b9-be52-cd2764f9707b.pdf"},{"id":70202062,"identity":"c00ece4e-3535-42dc-9332-58165a89b26b","added_by":"auto","created_at":"2024-11-29 12:43:41","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":13110,"visible":true,"origin":"","legend":"","description":"","filename":"CrashPredictionMECFSLCPaperSI.docx","url":"https://assets-eu.researchsquare.com/files/rs-5423422/v1/6e77bc249844e7f4f76e1e3e.docx"}],"financialInterests":"Competing interest reported. H.L. and R.P. are employed by Visible Inc.\nA.A. received consultation fees from Visible Inc.","formattedTitle":"Smartphone-based monitoring of heart rate variability and resting heart rate predicts variability in symptom exacerbations in people with complex chronic illness","fulltext":[{"header":"Introduction","content":"\u003cp\u003eComplex chronic illnesses such as Long COVID (LC) and Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS) can have a significant effect on the quality of life (1,2). The dynamic, often unpredictable nature of symptom severity and disability contributes to these declines in well-being. For example, approximately 85% of those affected by LC experience episodic symptoms that can rapidly fluctuate from periods of symptom stability to severe exacerbations, resulting in significant functional declines (3\u0026ndash;5). These periods of severe worsening symptoms are often colloquially called \u0026ldquo;crashing\u0026rdquo; or \u0026ldquo;flare-ups\u0026rdquo; by those living with complex chronic illnesses. However, biological drivers of these symptom exacerbations have not been fully elucidated, and many times, people with complex chronic illnesses are unable to link a change in symptoms to a particular event. Overall, the inability of people with complex chronic illnesses to have measurable ways to predict periods of symptom exacerbation can lead to less effective disease management and, subsequently, reduced quality of life (5).\u003c/p\u003e \u003cp\u003eRecent work suggests that heart rate variability (HRV) and other heart rate (HR) metrics may hold utility as potential biomarkers of complex chronic illness symptoms (6,7). Resting HR, or the number of successive heartbeats per minute, and HRV, the variability of R to R intervals measured linearly or in time/frequency domains, have been well identified as key metrics of overall health in healthy cohorts (8\u0026ndash;10). Further, HRV is shown to be a strong indicator of cardiovascular health, physiological responsiveness to stress, and autonomic function (9,11,12). HRV, though generally non-specific, is influenced by multiple organ systems, including the cardiovascular, autonomic nervous, respiratory, and immune systems, making it a robust indicator of overall health. Given the diversity of organ systems and pathobiology implicated in the pathogenesis of complex chronic illnesses such as LC and ME/CFS, serial evaluation of HRV and HR may represent a novel approach to identifying and predicting symptom exacerbations (13).\u003c/p\u003e \u003cp\u003eIn laboratory environments, HRV, HR, and respiratory rate (RR) are typically measured using electrocardiograms (ECG) and respiration belts. These tools have been clinically validated to provided research-reliable estimates of sympathetic (SNS) and parasympathetic nervous system (PNS) activity, but are generally limited to controlled lab or clinical settings (14\u0026ndash;16). Using ECG and respiration belts for managing health conditions is less practical in real-world settings due to their cumbersome nature and reliance on specialized equipment, making them difficult to implement. However, recent advances in mobile technologies have enabled naturalistic and user-friendly collection of cardiac and respiratory biomarkers of autonomic nervous system (ANS) control via photoplethysmography (PPG) sensors on smartphone cameras or wearable devices(17,18). These advances in digital health allow for the acquisition of self-reported daily fluctuations in symptoms alongside biometric proxies of physiological fluctuation.\u003c/p\u003e \u003cp\u003eFor example, biometrics and longitudinal symptom reporting have been pivotal in predicting symptom exacerbation in people with chronic diseases. One example is in people with chronic obstructive pulmonary disease (COPD). Using the myCOPD app, daily self-report data used in machine learning models could predict respiratory exacerbation in the subsequent days with moderate accuracy (19). Other digital health applications for people with COPD such as COPDPredict\u0026trade; accurately predicted respiratory exacerbations with a high degree of accuracy using both self report and biometric inputs (20). These models have the potential to influence early treatment of respiratory exacerbations, which is associated with improved clinical outcomes, including exacerbation recovery time. Similar platforms developed for cystic fibrosis have determined that the use of simple self-reporting platforms can lead to earlier detection of respiratory exacerbations and treatment in the form of oral antibiotics in adults (21) and pediatrics (22). While the pathophysiology is vastly different, these studies raise a question of the potential influence of earlier detection on crash length and severity or complex chronic illness characterized by energy limitation.\u003c/p\u003e \u003cp\u003eThe Visible application is one such mobile application that has recently been developed that provides people with LC, ME/CFS, and other complex chronic conditions the ability to self-monitor symptoms and potentially relevant biometrics. The present study aims to conduct a high-level descriptive analysis of retrospectively collected data from the Visible application and to predict both within-person and between-person variations in daily biometric data in relation to subsequent evening reports of crash, fatigue, and brain fog.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003eNo personal identifiable information was collected from this dataset. Thus, the dataset was determined to be exempt from human subject research by the Mount Sinai Program for the Protection of Human Subjects. However, all data presented here was derived from individuals who consented to provide their data for research purposes on the Visible application.\u003c/p\u003e \u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eParticipants\u003c/h2\u003e \u003cp\u003ePeople aged\u0026thinsp;\u0026ge;\u0026thinsp;18 years with self-identified complex chronic illnesses such as LC, ME/CFS, or people experiencing other causes of energy limitation who were using the Visible application and opted to share their data anonymously were eligible for inclusion. Visible is a commercially available application that can be downloaded from major smartphone app stores, including both iOS and Android platforms. People self-identifying with LC were asked to confirm that they met the World Health Organization (WHO) definition of LC. Participants were included if they provided at least 14 biometric readings. Data used in this analysis was collected between August 2022 and April 2024.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eOutcomes\u003c/h3\u003e\n\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003eBiometrics\u003c/h2\u003e \u003cp\u003eResting heart rate (HR), heart rate variability (HRV), and respiration rate (RR) were measured using Photoplethysmography (PPG). In the unpaid version of the Visible application, users provide PPG reading using their smartphone camera. PPG via smartphone camera detects blood volume changes in the skin. When the user places their fingertip over the smartphone camera, the device emits light that penetrates the skin. As blood pulses through the vessels, it modulates the amount of light absorbed and reflected. The camera captures these changes in light intensity, and the resulting signal is analyzed to derive heart rate and other biometric data (18,23). Participants using a subscription version of the application (Visible) had the option of measuring their HR and HRV with a Polar Verity Sense\u0026trade; armband (Polar Electro, Inc, Kempele, Finland). The Polar Verity Sense is a PPG-based optical heart rate sensor designed for upper-arm wear. The device was positioned with the sensor placed on the inner side of the armband, ensuring firm contact with the skin. Users are instructed to complete the biometric recording sitting or lying down and in the same position for each recording. HRV score is derived from calculating the root mean square of successive differences between heartbeats and scaling the score to receive an interpretable value between 0-100. Resting HR is calculated by measuring the number of heartbeats per minute. RR is calculated by detecting changes in heart rate related to breathing, where the heart rate increases during inhalation and decreases during exhalation. By leveraging these heart rate variations, the respiration rate is estimated.\u003c/p\u003e \u003cp\u003eAdditionally, the coefficient of variation (CoV) of an individual\u0026rsquo;s biometric values over the previous seven days to assess physiological stability and variability.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eIncidence of Crash\u003c/h3\u003e\n\u003cp\u003eA crash was defined within the app as periods of time when an individual's illness is significantly worse. The following explanation was provided to users: \u0026ldquo;Crashes usually occur as a part of post-exertional symptom exacerbation and affect your ability to carry out your usual activities. They normally last a few days. People use different words to describe these, such as flare-ups''. Each evening, users are prompted to respond with either a \u0026ldquo;✓\u0026rdquo; or an \u0026ldquo;X\u0026rdquo; to the item labeled \u0026ldquo;Crash\u0026rdquo; under the question \u0026ldquo;What else happened today?\u0026rdquo; This prompt allows users to indicate whether they experienced a crash that day. Check-marks were coded as 1 and Xs were coded as 0.\u003c/p\u003e\n\u003ch3\u003eFatigue\u003c/h3\u003e\n\u003cp\u003eEach evening, users are prompted to log the degree of severity in fatigue. Fatigue was self-reported by participants on a 0 to 3 scale (0\u0026thinsp;=\u0026thinsp;no impairment; 1\u0026thinsp;=\u0026thinsp;mild; 2\u0026thinsp;=\u0026thinsp;moderate; 3\u0026thinsp;=\u0026thinsp;severe).\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eBrain Fog\u003c/h2\u003e \u003cp\u003eEach evening, users are prompted to log the degree of severity in brain fog. Brain fog was self-reported by participants on a 0 to 3 scale (0\u0026thinsp;=\u0026thinsp;no impairment; 1\u0026thinsp;=\u0026thinsp;mild; 2\u0026thinsp;=\u0026thinsp;moderate; 3\u0026thinsp;=\u0026thinsp;severe).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e Description: Adults with complex chronic illnesses completed a biometric assessment (heart rate, HR; heart rate variability, HRV; respiratory rate, RR) each morning, followed by a self-report symptom survey in the evening. The diagram illustrates the model structure, where morning biometric scores (e.g., HRV) predict same-day evening symptom severity. This approach allows for within-day analysis of the relationship between physiological measures and symptom changes.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eAnalysis\u003c/h3\u003e\n\u003cp\u003eAnalyses were performed in R version 4.3.0(24) using the lme4 package (25) for multilevel modeling and tidymodels (26) for data preprocessing for predictive modeling.\u003c/p\u003e \u003cp\u003eWe conducted a multilevel regression analysis to predict same-day associations between morning biometrics and evening symptom reports. We disaggregated between-person (across participants) and within-person (within each participant) estimates of tonic and dynamic associations between biometrics and symptoms by partitioning predictor variances into between-person (group-level averages) and within-person (individual-level changes). Time-invariant, between-person predictors were grand-mean centered while time-varying, within-person predictors were group-mean centered. At the between-person level, we included all time-invariant predictors such as grand-mean biometric levels, stability scores, age, gender, and sensor modality.\u003c/p\u003e \u003cp\u003eOn the within-person level, we included all time-varying predictors, including the group-mean centered biometric level and stability scores, observation date, and autoregressive outcome value of the day prior. The autoregressive variable is included in the models to control for serial correlation effects of for when an individual experiences multiple days of elevated or diminished symptomology.\u003c/p\u003e \u003cp\u003eGiven the number of regression models and large sample size, only p-values\u0026thinsp;\u0026lt;\u0026thinsp;0.001 are interpreted as significant to help control for inflated p-values that can occur in large samples.\u003c/p\u003e\n\u003ch3\u003eModel:\u003c/h3\u003e\n\u003cp\u003e \u003cdiv id=\"Equa\" class=\"Equation\"\u003e \u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equa\" name=\"EquationSource\"\u003e\n$$\\:{Y}_{i,j}={\\gamma\\:}_{00}+{u}_{0j}+{\\gamma\\:}_{10}*H{R}_{ij}+{\\gamma\\:}_{1j}*H{R}_{ij}+{\\gamma\\:}_{20}*HR{V}_{ij}+{\\gamma\\:}_{2j}*HR{V}_{ij}+{\\gamma\\:}_{30}*R{R}_{ij}+{\\gamma\\:}_{3j}*R{R}_{ij}$$\u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Equb\" class=\"Equation\"\u003e \u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equb\" name=\"EquationSource\"\u003e\n$$\\:+{\\gamma\\:}_{40}*HR\\_Co{V}_{ij}+{\\gamma\\:}_{4j\\:}*HR\\_Co{V}_{ij}\\:+\\:\\:{\\gamma\\:}_{50}*HRV\\_Co{V}_{ij}+{\\gamma\\:}_{5j\\:}*HRV\\_Co{V}_{ij}\\:+\\:{\\gamma\\:}_{90}*RR\\_Co{V}_{ij}\\:+\\:{\\gamma\\:}_{6j}*RR\\_Co{V}_{ij}$$\u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Equc\" class=\"Equation\"\u003e \u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equc\" name=\"EquationSource\"\u003e\n$$\\:+{\\gamma\\:}_{7j}*dat{e}_{ij}+{\\gamma\\:}_{80}*ag{e}_{ij}+{\\gamma\\:}_{90}*gende{r}_{ij}+{\\gamma\\:}_{10j}*{Y}_{i,1-j}+{\\gamma\\:}_{110}*source\\_devic{e}_{ij}+{\\epsilon\\:}_{ij}$$\u003c/div\u003e \u003c/div\u003e \u003c/p\u003e \u003cp\u003eIn this multilevel model, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:Y\\)\u003c/span\u003e\u003c/span\u003e represents the outcome variable (e.g., crash, fatigue, brain fog) for observation \u0026#119894; within participant \u0026#119895;. The group-level intercept, denoted as \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\gamma\\:}_{00}\\)\u003c/span\u003e\u003c/span\u003e ​, represents the average outcome across all participants when all predictors are at their average values. The term \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{u}_{0j}\\)\u003c/span\u003e\u003c/span\u003e is the random intercept for participant \u0026#119895;, capturing deviations in baseline outcomes for each participant compared to the overall average. This accounts for individual differences that are not explained by the fixed effects.\u003c/p\u003e \u003cp\u003eThe model includes both within-person and between-person components. The within-person effects are captured by the linear terms \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\gamma\\:}_{10}\\)\u003c/span\u003e\u003c/span\u003e - \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\gamma\\:}_{110}\\)\u003c/span\u003e\u003c/span\u003e. These terms represent the relationship between the fluctuations in each participant's biometric variables and their symptom outcome specific occasion.\u003c/p\u003e \u003cp\u003eThe between-person effects, represented by \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\gamma\\:}_{1j}\\)\u003c/span\u003e\u003c/span\u003e- ​ \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\gamma\\:}_{10j}\\)\u003c/span\u003e\u003c/span\u003e, denote the relationships between each participant's overall (between-person) average levels of time-invariant variables and their average symptom outcomes. These terms capture how individual differences in long-term average biometric states relate to average symptom reporting.\u003c/p\u003e \u003cp\u003eIn addition to the biometric variables, the model adjusts for other factors, including the observation date, age, gender, previous day's symptom reporting (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{Y}_{i,1-j}\\)\u003c/span\u003e\u003c/span\u003e), and the device used to collect biometric data. Finally, the residual error term \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\epsilon\\:}_{ij}\\)\u003c/span\u003e\u003c/span\u003e​ accounts for unobserved variability in symptom outcomes at the individual level that is not explained by the model's predictors.\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003ePredictive Model Evaluation\u003c/h2\u003e \u003cp\u003eWe performed stratified blocked 5-fold cross-validation (CV) to assess model predictive model performance. We compared models with only biometric feature variables and models that included biometric \u003cem\u003eand\u003c/em\u003e autoregressive outcome variables within our linear mixed model framework. We also compared prediction models that included the random effect term in the prediction model to those without the random effect term to evaluate how random intercepts for each user account for individual variance in an individual\u0026rsquo;s probability value. Classification was the primary goal, so fatigue and brain fog were recoded such that \u0026ldquo;no or mild symptoms\u0026rdquo; were coded as 0, and \u0026ldquo;moderate to severe symptoms\u0026rdquo; were coded as 1.\u003c/p\u003e \u003cp\u003eThe 5-fold CV with stratified partitions was based on individual participants and a 20% holdout set for external validation. In the stratified CV, each participant's data contributed to all folds, ensuring that performance estimates reflect the model's ability to generalize to new observations from individuals included in the training set. The holdout set consisted of participants not present in the training set, allowing assessment of the model's generalizability to entirely new individuals. Stratification and holdout procedures are presented in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e. The primary performance metric was the mean of CV area under the receiver operating characteristic curve (ROC-AUC).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e Description: Figure illustrating the 5-fold cross-validation process for three example participants. The colored sections (blue and gold) represent training/testing for each stratified fold within the larger training set. The holdout set is not pictured.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eDescriptive Statistics\u003c/h2\u003e \u003cp\u003eData from n\u0026thinsp;=\u0026thinsp;4,244 Visible application users are presented. Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e presents descriptive statistics on the Visible dataset to reflect the breadth of the data used in the subsequent analyses. Descriptive graphics of the biometrics are presented in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e. Of the full sample, 3,514 users tracked crashes, 3,915 users tracked fatigue, and 3,525 users tracked brain fog.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eDataset Information Per-User\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"2\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCharacteristic Per User\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003en\u0026thinsp;=\u0026thinsp;4,244\u003csup\u003e\u003cem\u003e1\u003c/em\u003e\u003c/sup\u003e\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNumber of Biometric Readings Provided\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e125 (55, 257)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLength of Sequential Biometric Readings Provided (without missing a day)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e20 (10, 44)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTotal Reported Crash Episodes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e15 (5, 42)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTotal Reported Fatigue Episodes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e76 (32, 171)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTotal Reported Brain Fog Episodes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e27 (6, 83)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDuration of Crash Episodes (days)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4 (2, 10)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDuration of Fatigue Episodes (days)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e18 (8, 49)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDuration of Brain Fog Episodes (days)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e6 (2, 19)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDays between Crash Episodes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e5 (3, 11)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDays between Fatigue Episodes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.68 (1.17, 2.78)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDays between Brain Fog Episodes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3 (1, 6)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003csup\u003e\u003cem\u003e1\u003c/em\u003e\u003c/sup\u003e Median (IQR)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e Description. Figure\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ea: A scatterplot displaying individual data points for heart rate (HR), heart rate variability (HRV), and respiratory rate (RR), illustrating the spread and distribution of the data across these variables. Figure\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eb: A bar plot showing the distribution of symptom outcomes (Crash, Fatigue, and Brain Fog). Fatigue and Brain Fog are dichotomously coded (0\u0026thinsp;=\u0026thinsp;no or mild symptoms, 1\u0026thinsp;=\u0026thinsp;moderate to severe symptoms). Figure\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ec: A correlation matrix displaying the correlation coefficients between HR, HRV, and RR, with corresponding color gradients representing the strength and direction of the relationships.\u003c/p\u003e \u003cp\u003e \u003cb\u003eMultilevel Model Regressions\u003c/b\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003e1. Crash Outcome\u003c/h2\u003e \u003cp\u003eWe first used a logistic MLM regression with random participant intercepts to model how fluctuation in daily biometrics predicted the likelihood of a reported crash that evening (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ea). On the within-person level, decreases in HRV, increases in HR, and decreases in 7-day HRV and HR stability (increases in CoV) were associated with an increased likelihood of a reported crash that evening. On a between-person level, the likelihood of experiencing a crash was significantly associated with lower average HRV and lower long-term stability in HR (increased CoV). There was also a significant main effect of time, such as reports of crashes decreased across app usage. The autoregressive crash fixed effect was a highly significant predictor in the model. The full MLM model results for crashes are available in SI Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003e2. Fatigue Outcome\u003c/h2\u003e \u003cp\u003eNext, we used MLM regression with random participant intercepts to model how daily biometrics fluctuation predicted the fatigue level that evening (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eb). On the within-person level, decreases in HRV, increases in HR, and decreases in 7-day HRV, HR, and RR stability (increases in CoV) were associated with higher fatigue that evening. On the between-person level, fatigue was significantly associated with higher average HR and lower long-term stability in HRV and HR (increased CoV). Similarly, individuals who did not identify as male or female reported higher average levels of fatigue. The autoregressive fatigue fixed effect was a highly significant predictor in the model. The full MLM model results for fatigue are available in SI Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003e3. Brain Fog Outcome\u003c/h2\u003e \u003cp\u003eFinally, we used MLM regression with random participant intercepts to model how daily biometrics fluctuation predicted the brain fog level reported that evening (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ec). On the within-person level, decreases in HRV, increases in HR, and decreases in 7-day HRV and HR stability (increases in CoV) were associated with increased brain fog. On the between-person level, brain fog was significantly associated with higher average HR and lower long-term stability in HR (increased CoV). Individuals who did not identify as male or female reported higher average brain fog levels. The autoregressive brain fog fixed effect was a significant predictor in the model. The full MLM model results for brain fog are available in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ea-c and SI Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003cb\u003eModel fit statistics for predictors of crash, fatigue, and brain fog are presented in\u003c/b\u003e Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e Description. Forest plot representing the standardized coefficients from multilevel regression models. 5a plots the standardized odds ratios when crash is modeled as the dependent variable; 5b plots the standardized beta coefficients when fatigue is modeled as the dependent variable; 5c plots the standardized beta coefficients when brain fog is modeled as the dependent variable.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eModel Statistics for Predictors of Crash, Fatigue, and Brain Fog\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCrash\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eFatigue\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eBrain Fog\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIntraclass Coefficient\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.29\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.29\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.36\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMarginal R\u003csup\u003e2\u003c/sup\u003e / Conditional R\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.122 / 0.468\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.148 / 0.396\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.269 / 0.531\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eN (Total Obs)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e399,010\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e540,564\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e474,047\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003ePredictive Model Performance\u003c/h2\u003e \u003cp\u003eNext, we compared the predictive performance of multilevel random effects models with and without autoregressive variables in model training. To facilitate classification analysis, we dichotomized fatigue and brain fog into no or mild symptoms (coded as 0) and moderate to severe symptoms (coded as 1).\u003c/p\u003e \u003cp\u003eIn the stratified cross-validation (CV), performance was evaluated on new observations, enabling us to generalize results to new data from the same sample. We compared model performance under two conditions: (1) when autoregressive features were included during model training and (2) when random effects were incorporated during prediction.\u003c/p\u003e \u003cp\u003eWhen random effects were \u003cem\u003eexcluded\u003c/em\u003e from the prediction process (i.e., predictions were made without accounting for individual-specific variation in intercepts), AUC values ranged from .60 to .61 when only biometric features were used and from .81 to .87 when both biometrics and autoregressive variables were included in the training. When random effects were \u003cem\u003eincluded\u003c/em\u003e in prediction (i.e., individual-specific variations in intercepts were accounted for in predictions), AUC values ranged from .83 to .88 for models trained on biometrics alone and from .87 to .91 for models trained on both biometrics and autoregressive outcomes.\u003c/p\u003e \u003cp\u003eAcross all models, stratified CV demonstrated superior performance compared to holdout-set testing. This is likely because the holdout set could not use participant-specific estimates of average biometrics during prediction, leading to lower performance metrics.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eModel Performance Metrics (Mean ROC-AUC) for Crash, Fatigue, and Brain Fog Using Stratified Cross-Validation and Hold Out Set\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOutcome\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eStratified CV (no-RE)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eStratified CV (RE)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eHoldout Set (no-RE)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e\u003cb\u003eBiometrics Only\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cem\u003eCrash\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e.60\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e.83\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e.58\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cem\u003eFatigue\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e.61\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e.86\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e.57\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cem\u003eBrain Fog\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e.61\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e.88\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e.57\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e\u003cb\u003eBiometrics\u0026thinsp;+\u0026thinsp;Day Prior Outcome\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cem\u003eCrash\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e.81\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e.87\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e.82\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cem\u003eFatigue\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e.79\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e.87\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e.77\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cem\u003eBrain Fog\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e.85\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e.91\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e.84\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e \u003cem\u003enote\u003c/em\u003e. RE\u0026thinsp;=\u0026thinsp;Random Effects; CV\u0026thinsp;=\u0026thinsp;Cross-Validation\u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis study analyzed a large (n\u0026thinsp;=\u0026thinsp;4,244), longitudinal, high-frequency dataset of individuals self-reporting LC, ME/CFS, or other energy-limiting chronic conditions. Our findings revealed individual-level associations between morning biometric fluctuations and evening symptom reports, with model AUCs ranging from .60 to .91, depending on the inclusion of individual intercepts and previous-day symptom data. These results underscore the potential for improving care for individuals with complex chronic conditions through the targeted development of personalized, evidence-based remote physiological monitoring systems.\u003c/p\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003eOverall Predictive Performance\u003c/h2\u003e \u003cp\u003eModels trained solely on biometric features produced moderate predictive accuracy, but including random effects\u0026mdash;capturing individual-specific intercepts\u0026mdash;led to substantial improvements, with AUC values rising from .60-.61 to .83-.88. This suggests that accounting for individual differences improves model performance. The additional improvement in model performance when combining both biometric and previous-day self-report features underscores the impact of self-report data and the autocorrelative nature of symptom flair-ups. These findings highlight the importance of personalized approaches in health monitoring, where a one-size-fits-all model may fall short.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003eWithin-Person Biometric Sources of Variation in Symptoms\u003c/h2\u003e \u003cp\u003eAcross all outcomes, within-person predictors of symptom changes were the strongest. We calculated daily fluctuations from baseline averages for heart rate variability (HRV), heart rate (HR), and respiration rate (RR), along with weekly changes in stability (CoV) for each biometric. RR was less predictive, contributing minimally to symptom prediction. Daily HRV and HR changes, along with 7-day biometric stability, emerged as key predictors, with higher HRV and lower HR associated with increased risk of crashes, fatigue, and brain fog. These findings are in line with past research that has found that individuals with chronic illness, including LC and ME/CFS, show alterations in HRV and HR compared with normative populations (14,27\u0026ndash;29). More short-term variability in HRV and HR was linked to a higher probability of experiencing worsening symptoms, suggesting that fluctuations in cardiovascular dynamics over several days may destabilize daily symptom patterns. These results suggest that short-term fluctuations in HRV and HR dynamics, not just single-day metrics, are crucial for predicting symptom exacerbations.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003eBetween-Person Biometric Sources of Variation in Symptoms\u003c/h2\u003e \u003cp\u003eTo a lesser extent than within-person predictors, between-person baseline biometric patterns also demonstrated predictive value for symptom experiences. Among these, the most consistent predictor was long-term variability in HR, where individuals with more stable HR patterns experienced fewer symptoms on average, including crashes, fatigue, and brain fog. This finding may reflect the role of HR stability as a marker of overall physiological resilience for people with complex chronic illnesses (30), where greater consistency in autonomic functioning reduces vulnerability to stressors that can trigger symptoms. Lower average HRV scores were linked to an increased likelihood of crashes but not fatigue or brain fog, suggesting that while HRV may be critical for predicting periods of acute stress (such as crashes), its influence on more chronic symptoms like fatigue and brain fog may be less pronounced. Conversely, higher average HR scores were associated with increased fatigue and brain fog, indicating that sustained elevations in HR could be a sign of prolonged physiological stress, contributing to chronic symptom experiences.\u003c/p\u003e \u003cp\u003eInterestingly, we observed a decrease in reported crashes and the average level of brain fog reported by Visible app users over time, suggesting potential evidence of the mobile application's efficacy in helping users manage their condition better. Previous research has demonstrated that digital health tools can empower individuals to monitor their health patterns and make informed behavioral decisions, leading to improved health outcomes (20,31). However, future controlled studies are necessary to validate these findings and confirm the effectiveness of the Visible app.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003eSpeculation of mechanisms driving predictive HR and HRV dynamics\u003c/h2\u003e \u003cp\u003eMechanistic inferences are difficult given the uncontrolled nature of the dataset and the multisystem nature of illnesses such as LC and ME/CFS but exploring potential mechanisms behind the observed associations between low HRV, high HR, and symptom crashes remains important. HRV and HR are commonly recognized as proxy measures of autonomic nervous system function: increased SNS activity raises HR and reduces HRV, while PNS activity lowers HRV and increases HR (33). One plausible mechanism involves the vagus nerve, which plays a key role in regulating inflammation and modulating central nervous system responses (34). A meta-analysis found a consistent negative relationship between HRV and markers of inflammation (35), thought to be mediated through the cholinergic anti-inflammatory pathway(36).\u003c/p\u003e \u003cp\u003eChronic inflammation can be regulated by the autonomic nervous system via this same cholinergic pathway and has been shown to decrease HRV and elevate resting HR. In conditions such as LC and ME/CFS, persistent pathogens, reactivation of latent viruses, onset of autoimmunity, dysregulation of cortisol and other hormones, mitochondrial dysfunction and endothelial dysfunction have all been reported and can all lead to chronic pro-inflammatory responses with the potential to cause significant daily fluctuations in HR and HRV (37\u0026ndash;41). The potential for these mechanisms to not only cause chronic activation of inflammatory pathways, but also daily fluctuations in these activations could indeed explain how subsequent fluctuations in HRV and HR could be used to predict the emergence of crashes and other debilitating symptoms. More research to better understand how these various biological pathways can influence daily HRV and HR data could further validate how to use daily HRV and HR monitoring strategies to better manage symptom burden in conditions such as LC and ME/CFS.\u003c/p\u003e \u003cdiv id=\"Sec23\" class=\"Section3\"\u003e \u003ch2\u003eLimitations \u0026amp; Future Directions\u003c/h2\u003e \u003cp\u003eThis study has several notable limitations, which offer valuable insights for future research directions. Due to the retrospective study design, limited information was collected on demographics, making the generalizability of the findings uncertain. Future studies should aim to gather more detailed demographic information to assess the applicability of findings across different populations. Similarly, participants in this study reported that they met the WHO criteria for LC, but there were no standardized criteria for reporting ME/CFS. Additionally, this study included individuals with other energy-limiting conditions, which introduces variability and challenges in defining the sample population. This led us to refer to the dataset as representing individuals with complex chronic illnesses rather than a specific condition. Participants measured biometrics using either smartphone cameras or armbands, which could introduce inconsistencies. While short heart-rate measurements via smartphone PPG recordings have been found reliable (42,43), future studies should standardize biometric data collection methods to ensure consistent and comparable results. Factors like time of day, temperature, skin tone, device type, software version, and recording length may have affected data accuracy. Participants were instructed to take measurements upon waking, but these factors could introduce variability in results. Despite these limitations, this study is the first to leverage data-driven, large-scale assessments of common symptoms among individuals with complex chronic illnesses.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Conclusion","content":"\u003cp\u003eLeveraging a natural intensive longitudinal data design from mobile health technologies, this study highlighted the potential of daily biometric monitoring to predict symptom fluctuations in individuals with complex chronic conditions. Within-person deviations in daily HRV and HR from a person\u0026rsquo;s baseline and changes in their biometric weekly stability were robust predictors of crash, fatigue, and brain fog. While these models demonstrated promising predictive performance among existing users, further work is needed to enhance applicability to new populations. These findings underscore the potential of digital health tools to improve real-time symptom tracking and management, offering valuable insights for the future development of personalized care strategies and remote health monitoring symptoms for LC, ME/CFS and other complex chronic illnesses.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eBMI: Body Mass Index\u003c/p\u003e\n\u003cp\u003eCOPD: Chronic obstructive pulmonary disease\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eECG: Electrocardiogram\u003c/p\u003e\n\u003cp\u003eHR: Heart Rate\u003c/p\u003e\n\u003cp\u003eHRV: Heart Rate Variability\u003c/p\u003e\n\u003cp\u003eLC: Long COVID\u003c/p\u003e\n\u003cp\u003eME/CFS: Myalgic Encephalomyelitis/Chronic Fatigue Syndrome\u003c/p\u003e\n\u003cp\u003eMLM: Multilevel Model Analysis\u003c/p\u003e\n\u003cp\u003ePNS: Parasympathetic Nervous System\u003c/p\u003e\n\u003cp\u003ePOTS: Postural Orthostatic Tachycardia Syndrome\u003c/p\u003e\n\u003cp\u003eRMSE: Root Mean Square Error\u003c/p\u003e\n\u003cp\u003eSNS: Sympathetic Nervous System\u003c/p\u003e\n\u003cp\u003eRR interval: Intervals between R waves in an ECG\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eConflicts of Interest\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eRory Preston is a Lead Data Scientist at Visible Health Inc. Harry Leeming is a Co-Founder at Visible Health Inc. Annie Brandes-Aitken received consulting fees at Visible Health Inc.\u0026nbsp;\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eThompson EJ, Williams DM, Walker AJ, Mitchell RE, Niedzwiedz CL, Yang TC, et al. Long COVID burden and risk factors in 10 UK longitudinal studies and electronic health records. Nat Commun. 2022 Jun 28;13(1):3528.\u003c/li\u003e\n\u003cli\u003eFalk Hvidberg M, Brinth LS, Olesen AV, Petersen KD, Ehlers L. The health-related quality of life for patients with myalgic encephalomyelitis / chronic fatigue syndrome (ME/CFS). PLoS One. 2015 Jul 6;10(7):e0132421.\u003c/li\u003e\n\u003cli\u003eDavis HE, Assaf GS, McCorkell L, Wei H, Low RJ, Re\u0026rsquo;em Y, et al. Characterizing long COVID in an international cohort: 7 months of symptoms and their impact. EClinicalMedicine. 2021 Aug;38(101019):101019.\u003c/li\u003e\n\u003cli\u003eHumphreys H, Kilby L, Kudiersky N, Copeland R. Long COVID and the role of physical activity: a qualitative study. BMJ Open. 2021 Mar 10;11(3):e047632.\u003c/li\u003e\n\u003cli\u003eO\u0026rsquo;Brien KK, Brown DA, McDuff K, St Clair-Sullivan N, Solomon P, Chan Carusone S, et al. Conceptualising the episodic nature of disability among adults living with Long COVID: a qualitative study. BMJ Glob Health. 2023 Mar;8(3):e011276.\u003c/li\u003e\n\u003cli\u003eVan Campen CLM, Visser FC. higher resting heart rate myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) patients compared healthy controls: relation stroke volumes. Medical Research Archives. 2022;(10).\u003c/li\u003e\n\u003cli\u003eFang S-C, Wu Y-L, Tsai P-S. Heart rate variability and risk of all-cause death and cardiovascular events in patients with cardiovascular disease: A meta-analysis of cohort studies. Biol Res Nurs. 2020 Jan;22(1):45\u0026ndash;56.\u003c/li\u003e\n\u003cli\u003eShaffer F, Ginsberg JP. An Overview of Heart Rate Variability Metrics and Norms. Front Public Health. 2017 Sep 28;5:258.\u003c/li\u003e\n\u003cli\u003eThayer JF, Ahs F, Fredrikson M, Sollers JJ 3rd, Wager TD. A meta-analysis of heart rate variability and neuroimaging studies: implications for heart rate variability as a marker of stress and health. Neurosci Biobehav Rev. 2012 Feb;36(2):747\u0026ndash;56.\u003c/li\u003e\n\u003cli\u003eQuer G, Gouda P, Galarnyk M, Topol EJ, Steinhubl SR. Inter- and intraindividual variability in daily resting heart rate and its associations with age, sex, sleep, BMI, and time of year: Retrospective, longitudinal cohort study of 92,457 adults. PLoS One. 2020 Feb 5;15(2):e0227709.\u003c/li\u003e\n\u003cli\u003eMongin D, Chabert C, Extremera MG, Hue O, Courvoisier DS, Carpena P, et al. Decrease of heart rate variability during exercise: An index of cardiorespiratory fitness. PLoS One. 2022 Sep 2;17(9):e0273981.\u003c/li\u003e\n\u003cli\u003eCole CR, Blackstone EH, Pashkow FJ, Snader CE, Lauer MS. Heart-rate recovery immediately after exercise as a predictor of mortality. N Engl J Med. 1999 Oct 28;341(18):1351\u0026ndash;7.\u003c/li\u003e\n\u003cli\u003eMooren F, B\u0026ouml;ckelmann I, Waranski M, Kotewitsch M, Teschler M, Sch\u0026auml;fer H, et al. Autonomic dysregulation in long-term patients suffering from Post-COVID-19 Syndrome assessed by heart rate variability. Sci Rep [Internet]. 2023 Sep 22;13. Available from: https://www.nature.com/articles/s41598-023-42615-y\u003c/li\u003e\n\u003cli\u003eBarizien N, Le Guen M, Russel S, Touche P, Huang F, Vall\u0026eacute;e A. Clinical characterization of dysautonomia in long COVID-19 patients. Sci Rep. 2021 Jul 7;11(1):14042.\u003c/li\u003e\n\u003cli\u003eda Silva RB, Neves VR, Montarroyos UR, Silveira MS, Sobral Filho DC. Heart rate variability as a predictor of mechanical ventilation weaning outcomes. Heart Lung. 2023 May;59:33\u0026ndash;6.\u003c/li\u003e\n\u003cli\u003eRajendra Acharya U, Paul Joseph K, Kannathal N, Lim CM, Suri JS. Heart rate variability: a review. Med Biol Eng Comput. 2006 Dec;44(12):1031\u0026ndash;51.\u003c/li\u003e\n\u003cli\u003eNelson BW, Low CA, Jacobson N, Are\u0026aacute;n P, Torous J, Allen NB. Guidelines for wrist-worn consumer wearable assessment of heart rate in biobehavioral research. NPJ Digit Med. 2020 Jun 26;3:90.\u003c/li\u003e\n\u003cli\u003eMather JD, Sculthorpe NF, Mair JL, Hayes LD. Validity of resting heart rate derived from contact-based smartphone photoplethysmography (PPG) compared with electrocardiography (ECG): A 2024 updated systematic review and meta-analysis of correlation coefficients. (Preprint) [Internet]. JMIR Preprints. 2024. Available from: http://dx.doi.org/10.2196/preprints.66662\u003c/li\u003e\n\u003cli\u003eChmiel FP, Burns DK, Pickering JB, Blythin A, Wilkinson TM, Boniface MJ. Prediction of chronic obstructive pulmonary disease exacerbation events by using patient self-reported data in a digital health app: Statistical evaluation and machine learning approach. JMIR Med Inform. 2022 Mar 21;10(3):e26499.\u003c/li\u003e\n\u003cli\u003ePatel ML, Wakayama LN, Bennett GG. Self-monitoring via digital health in weight loss interventions: A systematic review among adults with overweight or obesity. Obesity (Silver Spring). 2021 Mar 1;29(3):478\u0026ndash;99.\u003c/li\u003e\n\u003cli\u003eWood J, Jenkins S, Putrino D, Mulrennan S, Morey S, Cecins N, et al. A smartphone application for reporting symptoms in adults with cystic fibrosis improves the detection of exacerbations: Results of a randomised controlled trial. J Cyst Fibros. 2020 Mar;19(2):271\u0026ndash;6.\u003c/li\u003e\n\u003cli\u003evan Horck M, Winkens B, Wesseling G, van Vliet D, van de Kant K, Vaassen S, et al. Early detection of pulmonary exacerbations in children with Cystic Fibrosis by electronic home monitoring of symptoms and lung function. Sci Rep. 2017 Sep 27;7(1):12350.\u003c/li\u003e\n\u003cli\u003ePeng R-C, Zhou X-L, Lin W-H, Zhang Y-T. Extraction of heart rate variability from smartphone photoplethysmograms. Comput Math Methods Med. 2015 Jan 12;2015:516826.\u003c/li\u003e\n\u003cli\u003eR Core Team. R: A Language and Environment for Statistical Computing. Vienna: R Foundation for Statistical Computing.; 2013.\u003c/li\u003e\n\u003cli\u003eBates D, M\u0026auml;chler M, Bolker B, Walker S. Fitting Linear Mixed-Effects Models using lme4 [Internet]. arXiv [stat.CO]. 2014. Available from: http://arxiv.org/abs/1406.5823\u003c/li\u003e\n\u003cli\u003eKuhn M, Wickham H. Tidymodels: a collection of packages for modeling and machine learning using tidyverse principles [Internet]. Boston, MA, USA; 2020. Available from: https://www.tidymodels.org\u003c/li\u003e\n\u003cli\u003eProal AD, VanElzakker MB. Long COVID or post-acute sequelae of COVID-19 (PASC): An overview of biological factors that may contribute to persistent symptoms. Front Microbiol. 2021 Jun 23;12:698169.\u003c/li\u003e\n\u003cli\u003eProal AD, VanElzakker MB, Aleman S, Bach K, Boribong BP, Buggert M, et al. SARS-CoV-2 reservoir in post-acute sequelae of COVID-19 (PASC). Nat Immunol. 2023 Oct 4;24(10):1616\u0026ndash;27.\u003c/li\u003e\n\u003cli\u003eRyabkova VA, Rubinskiy AV, Marchenko VN, Trofimov VI, Churilov LP. Similar patterns of dysautonomia in myalgic encephalomyelitis/chronic fatigue and post-COVID-19 syndromes. Pathophysiology. 2024 Jan 5;31(1):1\u0026ndash;17.\u003c/li\u003e\n\u003cli\u003eMensink GBM, Hoffmeister H. The relationship between resting heart rate and all-cause, cardiovascular and cancer mortality. Eur Heart J. 1997 Sep 1;18(9):1404\u0026ndash;10.\u003c/li\u003e\n\u003cli\u003eZhen J, Marshall JK, Nguyen GC, Atreja A, Narula N. Impact of digital health monitoring in the management of inflammatory bowel disease. J Med Syst. 2021 Jan 15;45(2):23.\u003c/li\u003e\n\u003cli\u003eCohen J. Statistical power analysis for the behavioral sciences [Internet]. Vol. 2nd, Statistical Power Analysis for the Behavioral Sciences. 1988. p. 567. Available from: http://dx.doi.org/10.1234/12345678\u003c/li\u003e\n\u003cli\u003eBerntson GG, Bigger JT Jr, Eckberg DL, Grossman P, Kaufmann PG, Malik M, et al. Heart rate variability: origins, methods, and interpretive caveats. Psychophysiology. 1997 Nov;34(6):623\u0026ndash;48.\u003c/li\u003e\n\u003cli\u003eTracey KJ. The inflammatory reflex. Nature. 2002 Dec;420(6917):853\u0026ndash;9.\u003c/li\u003e\n\u003cli\u003eWilliams DP, Koenig J, Carnevali L, Sgoifo A, Jarczok MN, Sternberg EM, et al. Heart rate variability and inflammation: A meta-analysis of human studies. Brain Behav Immun. 2019 Aug;80:219\u0026ndash;26.\u003c/li\u003e\n\u003cli\u003eTracy LM, Ioannou L, Baker KS, Gibson SJ, Georgiou-Karistianis N, Giummarra MJ. Meta-analytic evidence for decreased heart rate variability in chronic pain implicating parasympathetic nervous system dysregulation. 2016 Jan 1;157(1):7\u0026ndash;29.\u003c/li\u003e\n\u003cli\u003ePeluso MJ, Deeks SG. Mechanisms of long COVID and the path toward therapeutics. Cell. 2024 Oct 3;187(20):5500\u0026ndash;29.\u003c/li\u003e\n\u003cli\u003eIwasaki A, Putrino D. Why we need a deeper understanding of the pathophysiology of long COVID. Lancet Infect Dis. 2023 Apr;23(4):393\u0026ndash;5.\u003c/li\u003e\n\u003cli\u003eKlein J, Wood J, Jaycox JR, Dhodapkar RM, Lu P, Gehlhausen JR, et al. Distinguishing features of long COVID identified through immune profiling. Nature. 2023 Nov;623(7985):139\u0026ndash;48.\u003c/li\u003e\n\u003cli\u003eTurner S, Khan MA, Putrino D, Woodcock A, Kell DB, Pretorius E. Long COVID: pathophysiological factors and abnormalities of coagulation. Trends Endocrinol Metab. 2023 Jun;34(6):321\u0026ndash;44.\u003c/li\u003e\n\u003cli\u003eAppelman B, Charlton BT, Goulding RP, Kerkhoff TJ, Breedveld EA, Noort W, et al. Muscle abnormalities worsen after post-exertional malaise in long COVID. Nature communications. 2024;15(1):1\u0026ndash;15.\u003c/li\u003e\n\u003cli\u003evan Dijk W, Huizink AC, Oosterman M, Lemmers-Jansen ILJ, de Vente W. Validation of photoplethysmography using a mobile phone application for the assessment of heart rate variability in the context of heart rate variability-biofeedback. Psychosom Med. 2023 Sep 1;85(7):568\u0026ndash;76.\u003c/li\u003e\n\u003cli\u003eB\u0026aacute;nhalmi A, Borb\u0026aacute;s J, Fidrich M, Bilicki V, Gingl Z, Rudas L. Analysis of a pulse rate variability measurement using a smartphone camera. J Healthc Eng. 2018 Feb 5;2018:4038034.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"npj-digital-medicine","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"npjdigitalmed","sideBox":"Learn more about [npj Digital Medicine](http://www.nature.com/npjdigitalmed/)","snPcode":"41746","submissionUrl":"https://submission.springernature.com/new-submission/41746/3","title":"npj Digital Medicine","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"NPJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-5423422/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5423422/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground:\u003c/strong\u003e Complex chronic conditions like Long COVID and Myalgic Encephalomyelitis/Chronic Fatigue Syndrome involve energy limitations and changes in heart rate variability (HRV) and resting heart rate (HR). Mobile health technologies now offer real-time, valid measurements of HRV and HR, advancing symptom monitoring and management. Using a high-density dataset from an observational longitudinal study, we aimed to describe, quantify, and predict within-person co-variations in daily biometric data and subsequent crash, fatigue, and brain fog symptom occurrences.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods:\u003c/strong\u003e Leveraging data collected through a mobile health app (n=4,244), we developed predictive models using mixed-effects linear regression and logistic regression to explore how within-person fluctuations in biometrics (HR, HRV, and respiratory rate) predict dynamic change in symptomology (crash, fatigue, and brain fog). Predictive performance was assessed using 5-fold stratified cross-validation and compared to a 20% holdout set to evaluate model generalizability to new observations and individuals.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults:\u003c/strong\u003e Across all symptom domains, within-person changes in HRV and HR consistently emerged as key predictors of symptom change across all models, with higher HR and lower HRV conferring risk for crashes, fatigue, and brain fog. Moreover, 7-day biometric stability (or variable dispersion) was a robust predictor of symptom occurrence and severity. Models trained solely on biometric features achieved moderate predictive performance in the stratified cross-validation set; however, incorporating random effects to capture individual-specific variations and prior-day symptom reports substantially enhanced model accuracy, with AUC values reaching .91.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDiscussion and Conclusion: \u003c/strong\u003eThis study is the first to use data-driven models to predict everyday symptom experiences in individuals with complex chronic illnesses based on biometric fluctuations. Findings demonstrate the potential utility of mobile health tools for real-time monitoring of symptoms and highlight the need for further research to refine these predictive models and integrate them into clinical decision-making processes.\u003c/p\u003e","manuscriptTitle":"Smartphone-based monitoring of heart rate variability and resting heart rate predicts variability in symptom exacerbations in people with complex chronic illness","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-11-29 12:43:36","doi":"10.21203/rs.3.rs-5423422/v1","editorialEvents":[{"type":"communityComments","content":4},{"type":"decision","content":"Revision requested","date":"2025-01-03T13:02:00+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2024-12-10T06:22:01+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"305450725941808307732827625292542394301","date":"2024-11-26T01:06:44+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2024-11-25T17:00:39+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"292689151525905580415985470329968578488","date":"2024-11-19T04:20:45+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"97614240516209708191569444881527997808","date":"2024-11-18T21:00:19+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2024-11-18T19:42:47+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2024-11-12T18:05:05+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2024-11-12T10:41:06+00:00","index":"","fulltext":""},{"type":"submitted","content":"npj Digital Medicine","date":"2024-11-09T20:02:50+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"npj-digital-medicine","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"npjdigitalmed","sideBox":"Learn more about [npj Digital Medicine](http://www.nature.com/npjdigitalmed/)","snPcode":"41746","submissionUrl":"https://submission.springernature.com/new-submission/41746/3","title":"npj Digital Medicine","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"NPJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"a546d6f5-9bfb-437b-b68d-8a25beff48ef","owner":[],"postedDate":"November 29th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2026-03-30T16:26:07+00:00","versionOfRecord":{"articleIdentity":"rs-5423422","link":"https://doi.org/10.1038/s41746-026-02543-3","journal":{"identity":"npj-digital-medicine","isVorOnly":false,"title":"npj Digital Medicine"},"publishedOn":"2026-03-24 16:10:56","publishedOnDateReadable":"March 24th, 2026"},"versionCreatedAt":"2024-11-29 12:43:36","video":"","vorDoi":"10.1038/s41746-026-02543-3","vorDoiUrl":"https://doi.org/10.1038/s41746-026-02543-3","workflowStages":[]},"version":"v1","identity":"rs-5423422","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5423422","identity":"rs-5423422","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}