Articles

Wearables and smartphones now collect a quiet stream of data about you: how long you slept, how restless that sleep was, your resting heart rate, your heart rate variability (HRV), and roughly how many steps you took. Some mental wellness apps can connect to that stream and place it alongside your mood ratings. The pitch is intuitive — your body probably knows something about how you feel. The reality is considerably more complicated.

What the research does and does not say

Sleep

Sleep and mood are genuinely linked. A large body of research finds that shorter or more disrupted sleep is associated with next-day negative affect, reduced positive mood and greater emotional reactivity. [1] Longitudinal studies show that persistent poor sleep is a risk factor for the onset of depression and anxiety disorders, not just a symptom of them. [2] So far, so intuitive.

But “associated with” is doing a lot of work in those sentences. At the population level, the relationship is reasonably consistent. At the level of one individual over one week, it is noisy. You may sleep poorly on a Tuesday and feel perfectly functional on Wednesday. You may have an objectively normal sleep duration and still wake feeling flat. The effect size varies enormously between people, and the direction can even reverse: some people report feeling energised after slightly less sleep during high-motivation periods.

Consumer sleep tracking adds another layer of uncertainty. Devices that estimate sleep stages from wrist movement or optical heart rate sensors have been compared against polysomnography — the clinical gold standard — and while they perform reasonably well for total sleep duration, they are considerably less reliable for staging sleep as light, deep or REM. [3] What your phone reports as “deep sleep” is an estimate, not a measurement.

Resting heart rate

Resting heart rate (RHR) tends to rise during periods of physical illness, overtraining and, in some studies, during episodes of elevated psychological stress. [4] A few studies have found that elevated RHR tracks with self-reported anxiety and depressive symptoms at a group level. But RHR is affected by hydration, caffeine, alcohol, ambient temperature, posture during measurement, and fitness level. A single elevated reading tells you very little. A sustained upward trend over days, sitting alongside other signals, is more worth paying attention to.

Heart rate variability

HRV — the variation in time between successive heartbeats — is arguably the biometric with the most theoretical grounding in mental health. Higher HRV is associated with greater vagal tone and has been linked to better emotional regulation and lower trait anxiety in research settings. [5] Reduced HRV has been observed during depressive episodes and in people with anxiety disorders. [6]

HRV measurements from consumer devices, however, carry significant caveats. The metric fluctuates substantially within a single day depending on breathing pattern, body position, time since waking and recent activity. Meaningful HRV analysis typically requires standardised conditions — first thing in the morning, at rest, over at least several minutes. Most wearables sample opportunistically, which limits how comparable readings are across days. [3]

Step count

Physical activity has one of the more robust associations with mood in the literature. A 2023 meta-analysis in JAMA Psychiatry found that higher levels of physical activity were associated with lower risk of depression across a range of study designs. [7] Step count is a rough proxy for activity level, and it is probably the biometric least distorted by consumer hardware. Even so, ten thousand steps on a day you spent dreading something tells you less than ten steps on a day you spent hiking somewhere you wanted to be.

The individual problem: population patterns do not predict your week

Most of the research cited above comes from studies with hundreds or thousands of participants. Statistical associations at that scale do not translate cleanly to predictions about a single person on a given day. This is sometimes called the ergodicity problem in psychological science: what is true on average across people is not necessarily true within one person across time. [8]

Your own longitudinal data — your sleep alongside your mood ratings, tracked consistently over weeks — is actually more informative about you than any population study. That is one genuine argument for personal tracking. But even then, a pattern in your own data is still a correlation. It does not tell you what is driving what.

HealthKit and Health Connect: what they are and how access works

On Apple devices, health and fitness data is consolidated in HealthKit, a framework that stores data from the Health app, Apple Watch and third-party devices. On Android, the equivalent is Health Connect, introduced by Google as a unified health data platform. [9] Both are designed so that apps can request read or write access to specific data types — sleep analysis, heart rate, step count and so on — and the user must explicitly grant that access. It can be revoked at any time through the device’s settings.

This opt-in architecture matters. Health data is not collected passively or by default. If you never grant a wellness app access to HealthKit or Health Connect, it receives nothing from those sources.

Why health data is legally special-category data

Under the UK GDPR and EU GDPR, health data is classified as “special category” personal data, reflecting that its misuse carries heightened risk of harm. Processing it requires a lawful basis from Article 9(2) of the regulation. The most relevant basis for a voluntary consumer app is Article 9(2)(a): the data subject has given explicit consent to the processing for specified purposes. [10]

Explicit consent means more than ticking a box. It must be informed — you need to know what data is being processed and why — freely given, specific to the purpose, and unambiguous. You must also be able to withdraw it. This is why legitimate health apps present a clear consent prompt before accessing biometric data, and why that data should not be repurposed for advertising or sold to third parties without separate, equally explicit consent.

If you are evaluating whether to grant a mental wellness app access to your health data, the key questions are: what specific data types will it read, what will it do with them, who else might receive them, and how do you revoke access if you change your mind?

Correlation in your own data: worth noticing, not worth diagnosing

A week in which your sleep averages five hours and your mood ratings are consistently low is a pattern worth noticing. It might prompt useful reflection: are you sleeping less because of stress, or is the poor sleep amplifying stress that would have been there anyway? Is something else — work pressure, a difficult relationship, physical illness — affecting both sleep and mood simultaneously?

Noticing a pattern is the start of a question, not the answer to one. The value of surfacing your own biometric and mood data together is that it can make patterns visible that you would not otherwise register — not that it explains them.

MoodFire’s approach: surfacing your own data without overclaiming

MoodFire is a CBT self-help app, supplemented by DBT-derived distress tolerance tools, built for daily mood and anxiety management. It is not a medical device and does not diagnose or treat any condition.

Within the Insights section, there is an optional Body & Mood view. If you choose to connect MoodFire to Apple HealthKit or Android Health Connect — which requires your explicit opt-in consent, handled under GDPR Article 9(2)(a) — this view places your mood check-in ratings alongside biometric data such as sleep duration and step count. It shows you what your own numbers look like next to each other over time.

The view does not tell you that your sleep is causing your mood, or that a given HRV reading means anything in particular. It surfaces a visual record of your own data so that you can notice patterns and decide whether those patterns are worth bringing to a conversation with a GP, therapist or other professional. If you do work with a therapist, MoodFire’s PDF export feature lets you share recent mood trends and patterns directly with them.

The Apple Watch and Wear OS companion supports quick mood check-ins from your wrist, which makes the underlying mood data — the subjective side of any biometric correlation — easier to log consistently. Consistent, frequent mood ratings make pattern detection more meaningful than sporadic entries. The Check In feature is grounded in affect-labelling research, which suggests that naming emotions with some specificity is itself a useful practice, independent of any biometric data. [11]

Biometric data can be one thread in a richer picture of how you are doing. It is worth treating it as exactly that — one thread, read with appropriate scepticism, in context with your own lived experience.