Introduction

The proliferation of smartphones, fitness trackers, and other wearable devices has enabled the collection of vast amounts of real-time digital health data. These devices, once limited to simple metrics like step counts and heart rate, now capture complex patterns ranging from sleep cycles and stress levels to emotional states and social interactions. Digital health data has significantly improved people's fitness and overall health awareness, as studies show consistent use of digital trackers promotes healthier habits, reduces sedentary behavior, and enhances emotional well-being (e.g., Tong, Huong Ly, et al., 2022).

This data extends beyond physical health to offer insights into behavioral aspects like personality traits, lifestyle choices, daily routines, and social preferences. By behavior, we mean the actions, habits, and interactions that reflect an individual’s physical, emotional, and social tendencies. Tracking these behaviors serves diverse purposes, from enhancing personal health to advancing targeted marketing strategies.

The purpose of this report is to explore the extended applications of digital health data, particularly how it enables us to infer various behavioral dimensions. We will examine the types of markers that can be tracked and highlight ongoing research in this area. While it touches a range of aspects, this report is intended as an introductory overview rather than an in-depth analysis.

What can constitute as digital health data

In this report, “digital health data” refers to information collected from devices such as smartphones, smartwatches, fitness trackers, and other wearables. These devices monitor a wide range of metrics that provide insights into an individual’s health and overall well-being. Beyond traditional health measures, digital health data also includes patterns of device usage, speech characteristics, and writing styles.

Digital health data can be broadly categorized into three types:

1. Physiological Data

Physiological data includes metrics related to the body’s physical functions, such as: respiratory rate, oxygen saturation (SpO2), heart rate, heart rate variability (HRV) and sleep patterns. These metrics are primarily collected through smartwatches, fitness trackers, and other wearable devices.

2. Activity Data

Activity data reflects patterns and habits in daily routines, captured through: step counts and movement trends, exercise routines, app usage patterns and device interactions (e.g., typing speed, swiping behavior, screen taps). These metrics are gathered from accelerometers, gyroscopes, and app usage logs within smartphones and wearable devices.

3. Communication and Interaction Metrics

Communication and interaction metrics are derived from digital interactions, including: speech patterns (tone, pitch, speaking rate), writing styles (sentence complexity, word choice) and digital interaction data (e.g., typing speed, geospatial activity). These are collected through microphones, text inputs, and activity sensors integrated into digital devices.

In summary, digital health data encompasses a diverse range of metrics collected through smartphones and wearable devices, offering valuable insights into physical, behavioral (general activity data), and interactional patterns. By leveraging these data streams, we can gain a comprehensive understanding of health and behavior.

Inferring behavior from digital health data

In this report, we explore four key aspects of behavior inferred from digital health data:

1. Emotional and Mental State

2. Lifestyle Habits

3. Personality Traits

4. Cognitive State

Its important to note that behavioral aspects are interconnected: poor mental health can impair cognition, while an active lifestyle improves both mental state and cognitive function. Sleep disruptions, linked to stress or inactivity, affect emotional stability and may induce sedentary lifestyle. These overlaps highlight the need for a holistic view of digital health data.

Emotional and Mental state

Emotional and mental states include both transient feelings, such as happiness or stress, and longer-term psychological conditions, such as anxiety and depression. These states can be assessed through physiological metrics and activity data, providing measurable insights into a person's emotional well-being. Heart rate variability (HRV), electrodermal activity (EDA), and galvanic skin response (GSR) serve as indicators of emotional states. For instance, lower HRV is often linked to increased stress or anxiety (Moshe et al., 2021). Similarly, EDA and GSR, which measure changes in skin conductance, can track levels of emotional arousal (e.g., stress) and are integrated into wearable devices like the Fitbit Sense and Feel (Zhu et al., 2023; Hickey et al., 2021). Smartphone activity data, such as typing speed, pressure, and error rates, can reveal emotional states like frustration or sadness. For example, slower typing speeds and increased pressure often correlate with stress or negative emotions (Epp et al., 2011). Additionally, sleep duration and variability in geospatial activity have been strongly associated with daily stress levels. Changes in depression have also been linked to variations in speech duration, geospatial activity, and sleep patterns (Ben-Zeev et al., 2015).

Lifestyle Habits

Lifestyle encapsulates long-term habits and routines that reflect an individual's physical, mental, and social well-being. Digital health data can enable us to learn about a person’s lifestyles whether it is sedentary, active, healthy, or intensely active etc., based on data like physical activity levels, heart rate variability and caloric expenditure. A sedentary lifestyle is characterized by prolonged periods of inactivity, such as extended sitting or lying down, with minimal physical movement. Data indicators for this lifestyle include low daily step counts, minimal active minutes, and long durations of inactivity detected by accelerometers (Salim et al., 2024 and Byrom, B., et al., 2016). In contrast, an active lifestyle includes regular engagement in moderate physical activities like walking, cycling, or recreational sports. Wearable devices detect this lifestyle through moderate to high step counts, consistent active minutes, and varied activity intensities recorded throughout the day (Stamatakis, Emmanuel, et al., 2022). An intensely active lifestyle entails scheduled, vigorous physical activities, such as running, high-intensity interval training (HIIT), or competitive sports, performed regularly. Indicators for this category include high-intensity activity sessions, elevated heart rate readings during workouts, and significant caloric expenditure metrics.

Personality Traits

Personality traits, particularly the Big Five (openness, conscientiousness, extraversion, agreeableness, and neuroticism), can be inferred from digital health data from smartphones, wearables, and other devices. Smartphone communication patterns, such as call and message frequency or irregularities, predict extraversion and neuroticism (Sze, et al., 2024). Extraverts show more outgoing calls, variable communication, higher step counts during social hours (late evenings), and higher nighttime heart rates (Zufferey, Noé, et al., 2023). Neurotic individuals exhibit irregular nighttime activity patterns, with gender-specific differences (positive for females, negative for males). App usage, especially productivity and social apps, reflects openness and conscientiousness, with the latter also linked to regular wake-up times, weekend cycling, and consistent physical activity (Nan, et al., 2019). Wearables enhance insights with detailed physical and physiological data. Extraversion and conscientiousness correlate with step counts, movement variability, and activity intensity (Stachl, Clemens, et al., 2020). Resting heart rate and variability indicate neuroticism, reflecting emotional stability and stress. Sleep patterns further differentiate traits: irregular sleep aligns with neuroticism, while consistency suggests conscientiousness .

Cognitive State

Cognitive function encompasses mental processes like memory, attention, and problem-solving. Digital health tools assess both short-term fluctuations and long-term capacities using metrics such as reaction time, focus levels, and sleep data. Wearable devices, which capture continuous data such as physical activity, sleep efficiency, and circadian rhythms, have demonstrated potential in predicting cognitive states, particularly in older adults. Studies show that machine learning models trained on wearable data can differentiate between normal cognition and impairments, especially in domains like working memory, processing speed, and attention. For example, accelerometer-derived metrics, such as variability in activity levels, have been significantly associated with cognitive performance, with higher variability linked to better function (Sakal, Collin, et al., 2024).

Conclusion

In conclusion, behavioral markers from digital health data provide a deeper understanding of an individual’s habits, emotions, and cognitive processes. These markers allow us to uncover complex patterns, offering insights into personality, lifestyle, and well-being that were previously difficult to measure. This information has transformative potential, enabling tailored health interventions, enhanced decision-making, and innovative solutions in various fields. With correct utilization these markers may play a pivotal role in improving individual and societal outcomes.