Digital Biomarkers最新文献

Introduction: The chair-rising test (CRT) is being widely used to assess lower body power. The test provides valuable information about functional capacity and other health outcomes. However, most centers use timing-based outcomes, which may compromise its suitability in younger people and fitter geriatric patients, and which may also introduce confounding effects of body height. We, therefore, aimed to compare the traditional use of timing-based outcome with digitally assessed measurements of neuromuscular power.

Methods: Data were collected from a longitudinal population-based study that examined changes in muscle and bone health. CRT and jumping mechanography were performed on a ground reaction force plate. In 346 people (age: 32-92 years), chair-rising rate (fCRT) was manually assessed, and peak chair-rising power (PCRT) and jumping power (PJMG) were computed. Statistical analyses targeted breakpoints in the relationships between fCRT, PCRT, and PJMG. Effects of age, body height, and sex were assessed with linear and partial regression analyses.

Results: Breakpoints were found at (fCRT = 0.778 Hz, PJMG = 35.2 Watt/kg, p < 0.001) and at (fCRT = 0.669 Hz, PCRT = 9.9 Watt/kg, p < 0.001). Slow chair-risers, defined by fCRT <0.669 Hz, were older than fast chair-risers (p < 0.001), albeit with a largely overlapping age range (fast chair-risers: 32-90 years, slow chair-risers: 32-92 years). Body height was correlated with fCRT (p < 0.001) and PCRT (p = 0.009) but not with PJMG (p = 0.59).

Conclusion: Timing-based CRT does not unequivocally reflect neuromuscular power. Its association with chair-rising power holds only in people who take more than 75 s for 5 stand-ups. For jumping power, the cutoff is at 6.4 s. Slow and fast chair-risers cannot be easily discerned by age. Bias by body height can substantially obscure age effects in timing-based CRT assessments. We conclude that chair-rising power represents a more universally applicable biomarker and is less influenced by body height compared to timing-based chair-rising assessments.

Duchenne muscular dystrophy (DMD) is a progressive neuromuscular disorder that impairs daily functioning and results in premature death. Current clinical assessments are widely used for characterizing functional impairment but have limitations due to their subjective and effort-based nature and because they only capture a snapshot of symptoms at a single point in time. Digital health technologies, such as wearable devices, allow continuous collection of movement and physiological data during daily life and could provide objective measures of the impact of DMD symptoms on daily functioning. For example, measurement of the 95th centile of stride velocity has recently gained endorsement by European regulators as an endpoint for evaluating functional changes in DMD, but the use of wearables for this purpose is just beginning. In this study, we present preliminary investigations of candidate digital biomarkers of functional impairment using real-world data and further explore the relationships between these parameters and established clinical assessments. We found nine candidate biomarkers for detecting DMD-related functional impairment, all exhibiting large to very large effect sizes in our sample of 14 boys with DMD and matched controls (9 DMDs, 5 controls, age 4-12 years). Each candidate biomarker was moderately or strongly associated with clinical measures of function in DMD. Six of the biomarkers are novel and/or understudied in DMD including objective measures of gait acceleration and variability; postural control immediately before and after a postural transition; and the smoothness of postural transitions. Notably, postural transition measures were more sensitive to DMD-related impairment than gait, activity, and cardiac measures. These results suggest that the quality of postural transitions could serve as a sensitive and objective measure of functional impairment in DMD and point toward the need for further exploration of these measures.

The physical basis of speech production in humans requires the coordination of multiple anatomical systems, where inhalation and exhalation of air through lungs is at the core of the phenomenon. Vocalization happens during exhalation, while inhalation typically happens between speech pauses. We use deep learning models to predict respiratory signals during speech-breathing, from which the respiration rate is estimated. Bilingual data from a large clinical study (N = 1,005) are used to develop and evaluate a multivariate time series transformer model with speech encoder embeddings as input. The best model shows the predicted respiration rate from speech within ±3 BPM for 82% of test subjects. A noise-aware algorithm was also tested in a simulated hospital environment with varying noise levels to evaluate the impact on performance. This work proposes and validates speech as a virtual sensor for respiration rate, which can be an efficient and cost-effective enabler for remote patient monitoring and telehealth solutions.

Background: Novel clinical measures assessed by a digital health technology tool require thresholds to interpret change over time, such as the minimal clinically important difference. Establishing such thresholds is a key component of clinical validation, facilitating understanding of relevant treatment effects.

Summary: Many of the approaches to derive interpretative thresholds for patient-reported outcomes can be applied to digital clinical measures. We present theoretical background to the use of interpretative thresholds, including the distinction between thresholds based on perceived importance versus measurement error, and thresholds for group- versus individual-level interpretations. We then review methods to estimate such thresholds, including anchor-based approaches. We illustrate the methods using data on cough frequency counts as measured by a wearable device in a clinical trial.

Key messages: This paper provides an overview of statistical methodologies to estimate thresholds for the interpretation of change.

Introduction: Wearable technologies can enhance measurements completed from home by participants in decentralized clinical trials. These measurements have shown promise in monitoring patient wellness outside the clinical setting. However, there are challenges in handling data and its interpretation when using consumer wearables, requiring input from statisticians and data scientists. This article describes three methods to estimate daily steps to address gaps in data from the Apple Watch in cancer patients and uses one of these methods in an analysis of the association between daily step count estimates and clinical events for these patients.

Methods: A cohort of 50 cancer patients used the DigiBioMarC app integrated with an Apple Watch for 28 days. We identified different gap types in watch data based on their length and context to estimate daily steps. Cox proportional hazards regression models were used to determine the association between step count and time to death or time to first clinical event. Decision tree modeling and participant clustering were also employed to identify digital biomarkers of physical activity that were predictive of clinical event occurrence and hazard ratio to clinical events, respectively.

Results: Among the three methods explored to address missing steps, the method that identified different step data gap types according to their duration and context yielded the most reasonable estimate of daily steps. Ten hours of waking time was used to differentiate between sufficient and insufficient measurement days. Daily step count on sufficient days was the most promising predictor of time to first clinical event (p = 0.068). This finding was consistent with participant clustering and decision tree analyses, where the participant clusters emerged naturally based on different levels of daily steps, and the group with the highest steps on sufficient days had the lowest hazard probability of mortality and clinical events. Additionally, daily steps on sufficient days can also be used as a predictor of whether a participant will have clinical events with an accuracy of 83.3%.

Conclusion: We have developed an effective way to estimate daily steps of consumer wearable data containing unknown data gaps. Daily step counts on days with sufficient sampling are a strong predictor of the timing and occurrence of clinical events, with individuals exhibiting higher daily step counts having reduced hazard of death or clinical events.

Background: This commentary highlights the evolution of our understanding of physical function (PF) and key models/frameworks that have contributed to the current holistic understanding of PF, which encompasses not only a person's performance but also the environment and any adaptations an individual utilizes. This commentary also addresses how digital health tools can facilitate and complement the assessment of holistic PF and enable both objective and subjective input from the participant in their real-world environment. Lastly, we discuss how successful implementation of digital tools within clinical research requires patient input.

Summary: This commentary highlights how our understanding of PF has evolved to be more holistic.

Key messages: Inclusion of digital tools within clinical research can provide a path forward to holistically assess PF in a patient-focused manner.

Introduction: This research is focused on early detection of Alzheimer's disease (AD) using a multiscale feature fusion framework, combining biomarkers from memory, vision, and speech regions extracted from magnetic resonance imaging and positron emission tomography images.

Methods: Using 2D gray level co-occurrence matrix (2D-GLCM) texture features, volume, standardized uptake value ratios (SUVR), and obesity from different neuroimaging modalities, the study applies various classifiers, demonstrating a feature importance analysis in each region of interest. The research employs four classifiers, namely linear support vector machine, linear discriminant analysis, logistic regression (LR), and logistic regression with stochastic gradient descent (LRSGD) classifiers, to determine feature importance, leading to subsequent validation using a probabilistic neural network classifier.

Results: The research highlights the critical role of brain texture features, particularly in memory regions, for AD detection. Significant sex-specific differences are observed, with males showing significance in texture features in memory regions, volume in vision regions, and SUVR in speech regions, while females exhibit significance in texture features in memory and speech regions, and SUVR in vision regions. Additionally, the study analyzes how obesity affects features used in AD prediction models, clarifying its effects on speech and vision regions, particularly brain volume.

Conclusion: The findings contribute valuable insights into the effectiveness of feature fusion, sex-specific differences, and the impact of obesity on AD-related biomarkers, paving the way for future research in early AD detection strategies and cognitive impairment classification.

Introduction: Though wrist-worn photoplethysmography (PPG) sensors play an important role in long-term and continuous heart rhythm monitoring, signals measured at the wrist are contaminated by more intense motion artifacts compared to other body locations. Machine learning (ML)-based algorithms can improve long-term pulse rate (PR) tracking but are associated with more stringent regulatory requirements when intended for clinical use. This study aimed to evaluate the accuracy of a digital health technology using wrist-worn PPG sensors and an ML-based algorithm to measure PR continuously.

Methods: Volunteers were enrolled in three independent clinical trials and concurrently monitored with the investigational device and FDA-cleared electrocardiography (ECG) devices during supervised protocols representative of real-life activities. The primary acceptance threshold was an accuracy root-mean-square (ARMS) ≤3 beats per minute (bpm) or 5 bpm under no-motion and motion conditions, respectively. Bias, mean absolute error (MAE), mean absolute percentage error (MAPE), limits of agreement (LoA), and Pearson and Lin's concordance correlation coefficients (⍴ and CCC) were also computed. Subgroup and outlier analyses were conducted to examine the effect of site, skin tone, age, sex, body mass index (BMI), and health status on PR accuracy.

Results: Collectively, 16,915 paired observations between the device and the reference ECG were analyzed from 157 subjects (male: 49.04%, age mean: 43 years, age range: 19-83 years, BMI mean: 26.4, BMI range: 17.5-52, Fitzpatrick class V-IV: 22.9%, cardiovascular condition: 24%). The PR output attained an accuracy of 1.67 bpm under no-motion (n = 5,621 min) and 4.39 bpm under motion (n = 11,294 min), satisfying the acceptance thresholds. Bias and LoA (lower, upper LoA) were -0.09 (-3.36, 3.17) bpm under no-motion and 0.51 (-8.05, 9.06) bpm under motion. MAE was 0.6 bpm in no-motion and 1.77 bpm in motion, and MAPE was 0.86% in no-motion and 2.05% in motion, with ⍴ and CCC >0.98 in both conditions. ARMS values met the clinical acceptance threshold in all relevant subgroups at each clinical site separately, excluding male subjects under motion conditions (ARMS = 5.41 bpm), with more frequent and larger outliers due to stronger forearm contractions. However, these mostly occurred in isolation and, therefore would not impact the clinical utility or usability of the device for its intended use of retrospective review and trend analysis (⍴ and CCC >0.97 and MAPE = 2.61%).

Conclusion: The analytical validation conducted in this study demonstrated clinical-grade accuracy and generalizability of ML-based continuous PR estimations across a full range of physical motions, health conditions, and demographic variables known to confound PPG signals, paving the way for device usage by populations most likely to benefit from continuous PR m

Introduction: Wrist-worn accelerometers can capture stepping behavior passively, continuously, and remotely. Methods utilizing peak detection, threshold crossing, and frequency analysis have been used to detect steps from wrist-worn accelerometer data, but it remains unclear how different approaches perform across a range of walking speeds and free-living activities. In this study, we evaluated the performance of four open-source methods for deriving step counts from wrist-worn accelerometry data, when applied to data from a range of structured locomotion and free-living activities. In addition, we assessed how modifying the parameters of these methods would affect their performance.

Methods: Twenty-one participants (ages 20-33) wore an ActiGraph CentrePoint Insight Watch (Actigraph, LLC) on their non-dominant wrist while completing structured locomotion activities in a motion capture laboratory and during a free-living period in a mock apartment. Criterion step counts were determined from motion capture heel-strike events and from StepWatch 3 (Modus Health, LLC) during the free-living period. Four open-source methods implementing different algorithmic approaches were applied to CPIW data to derive step counts. The quantity and timing of method-derived and criterion steps during each type of activity were then compared.

Results: In terms of performance during structured locomotion, methods that relied on a single parameter, such as peak detection or threshold crossing, demonstrated the lowest bias among those investigated. Furthermore, three of the four investigated methods overestimated step counts during slow walking and underestimated step counts during fast walking, while the last method consistently underestimated at least half of the recorded steps across all speeds. During free-living activities, the method relying on frequency analysis exhibited the lowest percent error of all methods. Finally, we found that the incorporation of a locomotion classifier, wherein steps were only estimated during identified locomotion periods, reduced error for two methods when applied to data across structured and free-living settings.

Conclusion: In studying the performance of different step-counting approaches across different settings, we found a tradeoff between performance during structured walking and that during free-living activities. These findings highlight the opportunity for novel, context-aware methods for accurate step counting across real-world settings.