Monitoring Daily Sleep, Mood, and Affect Using Digital Technologies and Wearables: A Systematic Review

Background: Sleep and affective states are closely intertwined. Nevertheless, previous methods to evaluate sleep-affect associations have been limited by poor ecological validity, with a few studies examining temporal or dynamic interactions in naturalistic settings. Objectives: First, to update and integrate evidence from studies investigating the reciprocal relationship between daily sleep and affective phenomena (mood, affect, and emotions) through ambulatory and prospective monitoring. Second, to evaluate differential patterns based on age, affective disorder diagnosis (bipolar, depression, and anxiety), and shift work patterns on day-to-day sleep-emotion dyads. Third, to summarise the use of wearables, actigraphy, and digital tools in assessing longitudinal sleep-affect associations. Method: A comprehensive PRISMA-compliant systematic review was conducted through the EMBASE, Ovid MEDLINE(R), PsycINFO, and Scopus databases. Results: Of the 3024 records screened, 121 studies were included. Bidirectionality of sleep-affect associations was found (in general) across affective disorders (bipolar, depression, and anxiety), shift workers, and healthy participants representing a range of age groups. However, findings were influenced by the sleep indices and affective dimensions operationalised, sampling resolution, time of day effects, and diagnostic status. Conclusions: Sleep disturbances, especially poorer sleep quality and truncated sleep duration, were consistently found to influence positive and negative affective experiences. Sleep was more often a stronger predictor of subsequent daytime affect than vice versa. The strength and magnitude of sleep-affect associations were more robust for subjective (self-reported) sleep parameters compared to objective (actigraphic) sleep parameters.


Data Extraction and Analysis
Retrieved records were exported and de-duplicated in EndNote following Bramer and Bain [47], Bramer et al. [50] guidelines.De-duplicated records were screened on Rayyan by title and abstract, and potentially relevant studies were retrieved for full-text article screening.A total of 112 studies were excluded at full-text screening from the review (Table S3).The PICO-based (Population, Intervention, Comparison, Outcome) taxonomy of reasons was used to exclude articles from the systematic review [51].Data extraction followed a standardised data extraction form (protocol available upon request).Categories of data and information extracted from identified records are outlined in Table S4, and the search and screening process is shown in Figure 1.Screening and data extraction were performed by one author (R.H.).Two student reviewers (blinded) also independently screened all records at the title, abstract, and full-text screening stages.Data extraction was verified by another independent (blinded) student reviewer.Discrepancies at any stage were discussed, checked, and resolved by a senior researcher (T.D.).Given the high heterogeneity across study designs and samples, this review adopted a narrative synthesis to summarise the findings.

Risk of Bias
Systematic risk of bias was evaluated using the National Heart, Lung and Blood stitute (NHLBI) Quality Assessment Tool for Observational Cohort and Cross-Sectio Studies, in line with prior reviews from Konjarski, Murray, Lee and Jackson [42] and Brink, Dietch, Tutek, Suh, Gross and Manber [44].Two raters (R.H. and A.D.) in pendently reviewed study quality and risk of bias using the NHLBI tool.Discrepanc after blinded review were then discussed between the two authors (R.H. and A.D.) a resolved by consensus.In total, 63 records (52.1%) were rated as 'Good' (minimal or l risk of bias), and 58 records (47.9%) were assigned a 'Fair' grade (moderate risk of bi Study quality ratings are outlined in Table 2.No records were excluded from the final d synthesis.

Risk of Bias
Systematic risk of bias was evaluated using the National Heart, Lung and Blood Institute (NHLBI) Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies, in line with prior reviews from Konjarski, Murray, Lee and Jackson [42] and ten Brink, Dietch, Tutek, Suh, Gross and Manber [44].Two raters (R.H. and A.D.) independently reviewed study quality and risk of bias using the NHLBI tool.Discrepancies after blinded review were then discussed between the two authors (R.H. and A.D.) and resolved by consensus.In total, 63 records (52.1%) were rated as 'Good' (minimal or low risk of bias), and 58 records (47.9%) were assigned a 'Fair' grade (moderate risk of bias).Study quality ratings are outlined in Table 2.No records were excluded from the final data synthesis.

Literature Search
A total of 5686 records were identified across all databases and sources.After deduplication using Bramer and Bain [47], Bramer et al. [50] guidelines, 3024 records were screened based on title and abstract, of which 233 were then assessed for full-text screening.A final 121 studies met the full inclusion criteria (see PRISMA flow diagram, Figure 1).

Description of the Included Studies
Studies were published between 1994 and 2024.An overview of included studies is summarised in Table 2, with sample characteristics, study quality ratings, and daily outcomes of sleep, mood, and affect reported.As evidenced in Figure 2, there has been a rapid emergence of published research (particularly from 2021 onwards) utilising naturalistic study designs to monitor day-to-day sleep, mood, or affect associations.

Literature Search
A total of 5686 records were identified across all databases and sources.After deduplication using Bramer and Bain [47],Bramer, Giustini, de Jonge, Holland and Bekhuis [50] guidelines, 3024 records were screened based on title and abstract, of which 233 were then assessed for full-text screening.A final 121 studies met the full inclusion criteria (see PRISMA flow diagram, Figure 1).

Description of the Included Studies
Studies were published between 1994 and 2024.An overview of included studies is summarised in Table 2, with sample characteristics, study quality ratings, and daily outcomes of sleep, mood, and affect reported.As evidenced in Figure 2, there has been a rapid emergence of published research (particularly from 2021 onwards) utilising naturalistic study designs to monitor day-to-day sleep, mood, or affect associations.

Participant Characteristics
Sample sizes ranged from 19 to 2804, with most studies including at least 50 participants (n = 109; 90.1%).Studies that involved adult populations (18 years of age or older) were most common (n = 103), followed by children and/or adolescent samples (<18 years of age based on WHO classifications) (n = 27).A small proportion of these studies had

Length of Data Collection
The period of assessment varied across studies, with 1-2 weeks being the most common (66% of all records).In total, 116 studies (96%) collected data for under 2 months (ranging from 3-56 days), with the remaining studies [55,88,107,134,141] capturing longterm sleep variability and affect (ranging from 3 months to 2 years), often over multiple waves.Table S6 in the Supplementary Materials summarises overall study length and measurement frequency.

Self-Report Data-Collection Format
A detailed breakdown of the number and timing of daily assessments and the selfreporting tools used across studies is outlined in Table S6.In total, 44 studies collected self-reported sleep and affect ratings concurrently.Most studies (n = 67) had at least one mood-affect rating that was separated in timing from subjective sleep reports, while 17 studies assessed only objective sleep data.The context of daily self-reports was rarely reported, such as location (e.g., subjective ratings, which were recorded in the home environment or workplace setting).Temporal factors (e.g., time of day) on sleep-affect assessments, however, were often reported.
Self-reporting adherence approaches varied (see Table S6) but mostly used signalcontingent responses such as smartphone push notifications, signal alarm prompts, automated call systems, auditory buzz or 'beepers' (e.g., from a digital wristwatch), survey reminders, and/or prompts sent via SMS text, email, or phone call.Participant-initiated, event-based, and time-contingent responses (i.e., surveys completed at fixed times each day) were also incorporated across studies.A small number of records failed to report the signalling method or prompt design.

Bipolar Disorder
Seven studies included samples with bipolar-spectrum disorders.Four records found significant associations between self-reported sleep disturbances and next-day mood or affect [60,61,136,141] and two for objective sleep parameters.Sleep duration was associated with better daily mood symptoms [141], poorer SQ predicted higher next-day negative mood (sadness) [136], and lower SE was linked with higher NA [60].Sleep disturbances (WASO) were associated with next-day negative affect [60] and mood [61], while total waketime (SOL + WASO) predicted the next morning negative mood [61].Only one study [91] did not find a significant impact of self-reported TST on mood levels (positive or negative) the next day.Four studies captured objective sleep data (actigraphy); longer TST was associated with fewer next-day depressive symptoms [140], and longer SOL was linked to higher negative affect [60].One study did not find an association with objective sleep markers (TST) on next-day mood symptoms (positive or negative) in bipolar or unipolar depression [92].
Across studies with bipolar disorder, five also assessed the impact of mood or affective experiences on subsequent sleep.Three found worse mood or negative affect impacted subsequent sleep disturbance (WASO, SOL, TWT) [60,61,136], SE, and sleep onset time [136].Two studies, however, had mixed or unexpected findings, with Merikangas, Swendsen, Hickie, Cui, Shou, Merikangas, Zhang, Lamers, Crainiceanu, Volkow and Zipunnikov [92] reporting no association between mood levels (positive or negative) and TST the next day, while Li, Mukherjee, Krishnamurthy, Millett, Ryan, Zhang, Saunders and Wang [91] found elevated mood symptoms were associated with reduced TST the next night.

Anxiety and Depressive Disorders
Only four studies included participants with diagnosed anxiety and/or comorbid depressive disorder among children or adolescents [59,173] and adult [106,165] samples.Better self-reported SQ predicted elevated positive and lower negative affect (especially among individuals diagnosed with depression or anxiety), but there was no association with TST [106].Better SQ but not longer TST was predictive of daily affect levels (higher positive affect and lower negative affect) [165].Self-reported TST or SQ, however, did not impact subsequent affect variability (positive or negative) [165].
There was varied evidence for actigraphic sleep indices, which depended on the affect outcome and clinical group.Cousins, Whalen, Dahl, Forbes, Olino, Ryan and Silk [59], for example, found longer sleep duration (actigraphic TST) was associated with improved next-day positive affect for adolescents diagnosed with an affective disorder (depression and anxiety) but not healthy controls, while Difrancesco, Penninx, Antypa, van Hemert, Riese and Lamers [106] did not find actigraphy-based TST predictive of subsequent same day affect.Reduced sleep latency (actigraphic-SOL) was linked to lower next-day negative affect and higher positive affect for depressed adolescents [59], while SE scores did not impact next-day affect [59].
Studies with anxiety disorder and/or depressed participants (n = 4) reported mixed bidirectional relationships.Daytime negative affect was related to less actigraphy-based wake time (WASO) in depressed adolescents [59], whereas lower negative affect the preceding day predicted better self-reported SQ, with effects strongest for depressed and anxious individuals [106].Negative mood rated by children and adolescents in the evening did not predict lower nightly self-reported TST [173], but decreased TST, conversely, was linked to increased negative mood (morning irritability).Higher negative affect variability was associated with worse SQ but not TST, while higher daily levels of negative affect did not significantly predict subsequent TST or SQ [165].Positive daytime affect also had mixed patterns; more positive affect was linked to longer time in bed that night (actigraphic TIB) and total sleep (actigraphic TST) for depressed adolescents, but conversely, less time in bed (actigraphic TIB) for adolescents with anxiety only [59].Meanwhile, higher daily levels of positive affect (in adults with anxiety disorders and/or depression along with controls) were associated with better SQ but not TST [165].Positive affect variability, however, was not significantly related to subsequent sleep indices (TST and SQ) [165].
Seven other studies included individuals with major depression or depressive disorders.Improved self-reported SQ predicted next-day positive affect [79], better mood [83], and lower negative affect [79].Poorer SQ was also associated with lower morning positive affect [111,176], especially for individuals with greater depression severity [176].Sleep duration (TST) was also non-linearly associated with daily affect (positive and negative) across depressed and non-depressed individuals [172].Sleep the previous night (lack of hours slept and excessive sleep) was associated with lower positive and higher negative affect [172].Disturbed sleep (delayed SOL) the previous night also dampened mood scores [83].Daytime affect or mood levels (positive or negative) were not associated with subsequent sleep (SQ, TST) across two studies with major depression or depressive disorders [79,92].For actigraphic sleep parameters, Poon, Cheng, Wong, Tam, Chung, Yeung and Ho [175] found no significant relationships between mood (positive or negative) and objective sleep indices (TST, SE, SOL, TIB, and WASO).Wescott, Taylor, Klevens, Franzen and Roecklein [176], meanwhile, found that longer sleep than usual (actigraphic TST) was related to better morning mood, while variability in nightly sleep duration was more impactful on mood or affect for depressed individuals compared to controls.

Shift Workers
Eight records investigated shift workers.Sleep loss (shorter self-reported and actigraphic TST) among medical residents impacted emotional reactions to affective work events (positive and negative) [55].Poor sleep (shorter TST) also amplified negative emotions and dampened positive emotional responses to daytime events [55], and shorter sleep (TST) predicted a worse mood the next day, which also led to shorter sleep the following night [88].Nurses with poorer sleep quality (SQ; within-person) had reduced daily positive affect and a higher daily negative affect [118], while in another nurse sample, worse sleep (TST via indirect effects of SE) was linked to higher depressed mood the following day [150].Shift workers across a range of retail and service sectors (with nonstandard hours) had a more positive mood after higher amounts of self-reported TST [162].For studies with actigraphy, TST was associated with positive affect and mood levels in healthcare workers [55,107,118] but SE [118] had no impact.Medical residents with later wake times, earlier bedtimes, and fewer shifts in total sleep (TST) also reported improved next-day mood [107].One study with medical residents had mixed sleep-affect patterns [54]: at the start of medical residency, sleep loss during a night shift was associated with elevated negative but not positive mood the following day.However, after 6 months of medical residency, sleep disruption (lower TST) after a night shift did not impact the next-day mood (positive or negative) [54].No studies with shift workers reported outcomes for SOL or WASO.
Only two shift work studies reported mood or affect outcomes that temporally preceded sleep.Lower mood predicted worse TST the next-day [88], and greater positive affect was linked to improved SQ [98].

Sleep on the Next Day Mood or Affect
The remaining studies had healthy populations (n = 97) with evidence for bidirectional relationships across five sleep indices (TST, SQ, SOL, WASO, and SE) and affective experiences (positive and negative) represented in Figure 4. Given the multitude of sleep-affect relationships, evidence for both objective and subjective sleep parameters is pooled in Figure 4. Evidence from studies with healthy populations (n = 97) was rated based on reported findings as strong (green solid), moderate (blue solid), weak or limited (yellow dotted), or no associations (red dotted).Associations refer to (significant) expected directions.

Sleep Duration
Self-reported TST in general was significantly related to next-day positive affect or mood in 13 of 25 studies, including children or adolescents [56,90,177] and adult samples [62,85,102,109,110,116,117,120,129,135,149,163].Less TST and worse sleep were related to lower positive daily moods [56,116,117], while longer and more consistent patterns of selfreported TST in general were significantly related to higher next-day positive affective states [62,85,90,102,110,120,135,149,163,177].
Two records also reported curvilinear effects of overall sleep loss for both positive and negative affect the following day [101,109], and one study [158] found an inverse relationship, such that longer sleep duration was linked to a lower next-day positive mood (reduced happiness).Seven studies had null results of self-reported sleep duration and next-day positive affective experiences [66,74,96,124,126,144,148], of which four included children or adolescents [66,74,144,148].Only 3 of 14 studies found an impact of longer objective sleep duration (actigraphic-TST) on better subsequent positive affective experiences [135,159,168].Nine records [57,65,90,96,103,121,144,148,157] did not demonstrate improved positive affect following longer actigraphic-recorded sleep duration; one study and negative (NA) affective experiences.Evidence from studies with healthy populations (n = 97) was rated based on reported findings as strong (green solid), moderate (blue solid), weak or limited (yellow dotted), or no associations (red dotted).Associations refer to (significant) expected directions.

Sleep Duration
Self-reported TST in general was significantly related to next-day positive affect or mood in 13 of 25 studies, including children or adolescents [56,90,177] and adult samples [62,85,102,109,110,116,117,120,129,135,149,163].Less TST and worse sleep were related to lower positive daily moods [56,116,117], while longer and more consistent patterns of self-reported TST in general were significantly related to higher next-day positive affective states [62,85,90,102,110,120,135,149,163,177].
Two records also reported curvilinear effects of overall sleep loss for both positive and negative affect the following day [101,109], and one study [158] found an inverse relationship, such that longer sleep duration was linked to a lower next-day positive mood (reduced happiness).Seven studies had null results of self-reported sleep duration and next-day positive affective experiences [66,74,96,124,126,144,148], of which four included children or adolescents [66,74,144,148].Only 3 of 14 studies found an impact of longer objective sleep duration (actigraphic-TST) on better subsequent positive affective experiences [135,159,168].Nine records [57,65,90,96,103,121,144,148,157] did not demonstrate improved positive affect following longer actigraphic-recorded sleep duration; one study found inconsistent associations with composite mood compared to individual mood items [71], and one study reported an inverse relationship, such that increased sleep (actigraphic-TST) was linked to lower positive affect the next day [134].
Ten studies did not find a direct association between the amount of subjectively reported TST and next-day negative affective experiences among adults and children or adolescents [63,66,67,74,96,124,126,148,149,163].Three studies also had varied findings: Kalmbach, Arnedt, Swanson, Rapier and Ciesla [82] found that one night of short sleep (TST) led to reduced next-day anhedonic depressive symptoms, but shorter sleep (TST) across a longer 2-week period was associated with higher anhedonia.Bean and Ciesla [104] reported increased anxious arousal symptoms following partial sleep deprivation (TST), but next-day anhedonic depressive and general distress were not impacted.Shorter sleep (TST) among adolescents [69] worsened next-day affective well-being, but among adults (over 20 years old), both shorter or longer sleep duration impacted affect the following day, thus highlighting a non-linear relationship.
There was varied evidence for actigraphic-recorded TST on negative mood.Sleep duration (actigraphic-TST) predicted next-day negative mood symptoms or affect ratings in seven studies [86,100,117,135,157,168,169], but there were no reported associations in eight other studies [57,65,96,103,114,121,144,148].One record also had inconsistent findings; actigraphic sleep (TST) was not associated with composite mood items but with 2 of 11 individual mood items [71].

Sleep Latency and Wakefulness after Sleep Onset
There was limited evidence for the impact of self-reported SOL and the frequency of nocturnal WASO on affective states.Shorter self-reported SOL was related to higher next-day positive affect and mood in 3 of 5 identified studies [52,62,110] and shorter WASO was associated with more positive next-day affect in 2 studies [52,120].Two studies did not find a relationship between subjective sleep disturbance (SOL) and positive mood [132,177], and one study did not find significant associations between night-time WASO and positive affective experiences among adolescents [143].
Actigraphic-recorded SOL and WASO on positive affect scores also had limited evidence.Mousavi, Lai, Simon, Rivera, Yunusova, Hu, Labbaf, Jafarlou, Dutt, Jain, Rahmani and Borelli [134] found longer sleep latency (actigraphic-SOL) the previous day was linked to lower next-day positive affect.Doane and Thurston [65] with adolescents and Parsey and Schmitter-Edgecombe [93] with adults, however, did not find a relationship between objective (actigraphic) SOL indices and next-day reports of positive mood.No studies reported an impact of actigraphic WASO on next-day positive affective experiences [57,93].
Six of seven studies found that longer subjective SOL was linked to a poorer negative mood the following day [52,62,132,167,174,177] and just one identified study by Kalmbach, Pillai, Roth and Drake [67] reported no influence on negative affect ratings (as with other reported sleep indices such as TST and SQ).Greater self-reported WASO and night-time disturbance were associated with elevated negative affect and poorer mood in six studies across adolescents and adults [75,110,143,167,174,177], and one record did not find a significant relationship [67].Actigraphic-recorded SOL [65,93] and WASO [57,93] did not impact next-day affect or mood.

Sleep Efficiency
Greater self-reported SE was associated with higher daily positive affect [110,120] and negative affect [87,113].One study, however, did not find significant main effects of self-reported SE on positive affect (high and low arousal) among adolescents and young adults [148].Takano, Sakamoto and Tanno [68] demonstrated a significant association between decreased actigraphic-measured SE and reduced positive affect levels the following day, while Master, Nahmod, Mathew, Hale, Chang and Buxton [157] found individuals who slept more efficiently than their average reported a higher next-day positive mood (happiness ratings).The remaining identified studies (n = 7) did not find any significant relationships between objective SE and next-day positive affect [65,93,96,113,121,135,148].Worse objectively recorded sleep efficiency (actigraphic-SE) was associated with increased next-day negative mood in just one study [117], with remaining studies reporting inconsistent [71,135] or null findings among adults, older adults, or adolescents [65,93,96,113,121,144,148,157].
3.6.Daytime Mood or Affect on Subsequent Sleep 3.6.1.Positive Affect More daytime positive affect or mood states predicted longer TST in 2 out of 19 identified studies with adolescents [56] and adults [67].One study with older adults [9] also found greater variability in daily positive affect was linked to lower sleep duration and more tiredness.Evidence from the remaining studies did not find a significant relationship between positive mood or affective experiences on subsequent TST [52,58,62,65,68,69,100,102,121,135,157,163,177] and three studies reported mixed findings [108,113,178].Zapalac, Miller, Champagne, Schnyer and Baird [178], for example, found that positive affective states in the morning (but not evening) were related to longer actigraphic sleep duration (as well as shorter SOL, better SQ, and fewer awakenings).
Daytime positive affective experiences were generally associated with self-reported SQ the following night, including for children and adolescents [58,73,74,112,144,169,170] and adults [53,64,67,94,98].Daytime positive affect or mood states did not significantly predict subsequent subjective SQ across 5 studies [52,72,130,135,177].Mixed results were reported by Jones, Smyth and Graham-Engeland [108] at the between-and within-person levels and Zapalac, Miller, Champagne, Schnyer and Baird [178] found discrepancies between positive affective states recorded in the morning compared to-evening and their subsequent impact on SQ. de Wild-Hartmann, Wichers, van Bemmel, Derom, Thiery, Jacobs, van Os and Simons [62] also found a negative association such that greater daytime positive affect was associated with lower SQ the next night (i.e., better mood was unexpectedly linked to worse sleep quality).
Only 3 of 9 studies found a significant impact of daily positive affect on actigraphic [121] or subjective [67,178] SOL.The remaining identified studies (n = 4) did not find an association with positive affect on subsequent SOL [52,62,65,68] or nocturnal wakefulness (WASO) [52,62,177].Tavernier, Choo, Grant and Adam [77] also reported mixed findings depending on positive affect arousal; feeling calm (low-arousal PA) predicted shorter SOL, while excitedness (high-arousal PA) predicted a longer SOL that night.Generally, low-arousal affective experiences (regardless of positive or negative valence) were related to better sleep outcomes compared to worse sleep for high-arousal daytime affective feelings [77].
Evidence for positive daytime affect impacting SE was limited.Only 1 of 8 studies reported a relationship between positive affect reactivity and subsequent SE [64], while the remaining studies did not find an association (for self-reported or actigraphic SE) [65,68,121,157].Two studies reported inverse relationships: Messman, Slavish, Dietch, Jenkins, ten Brink and Taylor [110] found lower-than-average positive morning affect was associated with higher actigraphic and self-reported sleep efficiency that night; Kouros, Keller, Martín-Piñón and El-Sheikh [144] found that higher ratings of positive mood (happiness) were associated with lower actigraphic sleep efficiency.
Only 3 of 9 studies [67,82,178] reported a significant impact of daytime negative affective experiences on impaired SOL, with Zapalac, Miller, Champagne, Schnyer and Baird [178] reporting associations only between SOL and morning (not evening) affect states.The remaining studies did not report any significant relationship between daytime negative mood or affect outcomes on subsequent SOL [52,62,65,121,177].There were only three studies that reported findings of negative affect and WASO.Tavernier, Choo, Grant, and Adam [69] reported higher negative social evaluative emotions (higher anxious-nervous) among adolescents predicted longer wake bouts that night (WASO) at the within-person level, while Totterdell, Reynolds, Parkinson and Briner [52] and Xie, Zhang, Wang, Chen and Lin [177] did not find previous day mood states predictive of subsequent nocturnal wakefulness.
No studies reported a significant association (in the expected direction) between negative affect and subsequent sleep efficiency (actigraphic or self-reported SE) [65,121,157].One study from Kouros, Keller, Martín-Piñón and El-Sheikh [144], however, reported an inverse relationship with a higher daily negative mood linked to greater sleep efficiency that night.

Discussion
Sleep and affective states are mutually connected.Emerging evidence from studies with ambulatory monitoring has shed light on the temporal relationships and dynamic patterns of sleep disturbances and affective experiences.This systematic review provides updated evidence in a rapidly emerging field and expands on previous reviews by including bipolar disorder subtypes and shift workers.As visualised in Figure 4, patterns of sleepaffect associations are summarised for healthy, non-clinical samples involving children, adolescents, adults, and older adults.

Key Findings
This systematic review screened 3024 records and identified 121 studies for inclusion.Only a fifth of records (n = 23) combined both self-report and objective (e.g., actigraphy) sleep outcomes in analyses.Most studies incorporated a standardised sleep or affective measure, with the majority utilising digital self-report tools (n = 101).Common sleep measures included the Consensus Sleep Diary (CSD) and the Pittsburgh Sleep Quality Index (PSQI), and for mood or affect measures, this included the Positive and Negative Affect Schedule (PANAS) and the Profile of Mood States (POMS).Sleep parameters ubiquitously assessed were sleep duration (TST; 75.2%), sleep quality (SQ; 63.6%), sleep onset latency (SOL; 29.8%), sleep efficiency (SE; 28.1%), and time awake after sleep onset (WASO; 19.8%).Despite heterogeneity in terminology, 47 records were categorised as utilising daily mood domains, and 76 studies had affect dimensions.Most studies separated the timing of sleep and mood-affect ratings, but over a third collected concurrent ratings.
Bidirectional patterns are summarised in Figure 4 for healthy populations.Most studies focused on sleep preceding subsequent affective states.Sleep duration (TST) and sleep quality (SQ) were strongly linked to daily positive and negative affective states compared to other sleep indices.Sleep latency (SOL), nocturnal wakefulness (WASO), and sleep efficiency (SE) had inconsistent evidence; while no clear patterns emerged for these domains (SOL, WASO, and SE), there was a more limited pool of studies.Daytime affect or mood-predicting subsequent sleep also had mixed evidence.Positive or negative affective experiences had limited evidence of influencing subsequent sleep duration (TST), latency (SOL), wakefulness (WASO), and sleep efficiency (SE).Only sleep quality (SQ) had moderately strong evidence for reciprocal sleep-affect associations.Sleep and affect moderators may explain these findings, such as individual coping strategies, differences in individual emotional regulation or reactivity, cognitive or self-control, and sleep hygiene, habits, or beliefs [16,44,179,180].Additional mediating factors were also rarely controlled for, such as the influence of age, gender, chronotype propensity, physical activity, daily variables, or bedtime procrastination (e.g., electronic device usage at night) on sleep-affect outcomes [2].
Shift working samples were less common (n = 8), but sleep duration (TST) was shown to impact both next-day positive and negative affect.A bidirectional sleep-mood association was shown for sleep quality but was reported by only one identified shift work study, which limits generalisability [88].Better self-reported sleep (SQ, SOL, SE, and WASO), in general, predicted mood or affect scores the following day in samples with a diagnosed affective disorder.Mixed or varied patterns were observed for daily selfreported and actigraphic-recorded sleep duration (TST).Reciprocal sleep-affect associations were also varied.In general, daytime negative mood predicted poorer subsequent sleep in clinical samples of affective disorders.Only a third of studies with an affective disorder, however, reported the impact of daytime positive mood or affect outcomes, and from these records, there was weak or no evidence for a direct association.Generalisability of findings for affective disorders is also limited by age, as only two studies [59,173] assessed non-adult populations.

Limitations
Methodological variance and heterogeneity in operational definitions (e.g., for affective outcomes) were limitations in the present review.Studies used interchangeable terms to describe mood, emotions, or affect.These are broad but often conflated domains that can differentially impact sleep [181].In this review, a range of affective phenomena were considered to evaluate a wider pool of available evidence and to consolidate the significant heterogeneity across ambulatory sleep-affect studies.Future research, however, should conceptually clarify and disentangle these complex interactions between sleep and multifaceted affect states.Ambulatory assessment is also an umbrella term [182][183][184] used to capture daily sampling but includes a range of methods.Variability in sleep or actigraphic parameters analysed, single or multi-item measures, sampling resolution, scheduling and prompting, sleep monitoring periods, and response compliance may also have impacted findings.Records that fail to incorporate a standardised affective measure may limit the replicability of findings, given that item composition and operationalisation (e.g., wording) differ markedly across studies [41].Contextual parameters and situational drivers were largely overlooked across studies, which have been shown to impact mood symptoms that fluctuate over time [185].
Future ambulatory studies would benefit from adopting additional, objective physiological markers to assess sleep-affect relationships [186].Just over a third of studies identified in this review utilised objective wearables (Figure 3 and Table 3), but very few captured additional multidimensional sleep-circadian biomarkers (e.g., temperature, cardio-respiratory function, general psychomotor levels, or light data) [187].Wearable device advancements have enabled researchers to more accurately monitor real-time sleep and diurnal activity outside of laboratory settings [188,189].Multiple sensors integrated within new-generation wearables, such as electrodermal activity (EDA), can be used to estimate sleep stages, detect sleep disorders, or indicate sleep quality and emotion classification [187,[190][191][192].

Research Agenda
This systematic review highlighted current gaps in the literature and future study recommendations: 1. Standardised affective measures should be utilised in future studies to afford consistency in reporting, robust quantification, and specificity of findings.Future research should consider both positive and negative valence, affective arousal domains, and specific emotions, which may be differentially impacted by sleep.This allows for daily function variability, captures working and non-working days (in line with ICSD-3 recommendations), and identifies acute, cumulative, or cascading effects.4.There is a paucity of studies to date utilising ambulatory tools to assess the mutual interplay of daily sleep and mood among affective disorders and shift workers; two groups vulnerable to disruptions in circadian and sleep pattern rhythmicity.5. To avoid contextual bias, future studies should consider the timing and sampling resolution of daily assessments and situational drivers.Time of day effects (e.g., moodcongruent biases due to proximity to sleep-wake intervals) and frequency of affect assessment (e.g., multiple or single ratings) should be considered.Multiple daily measures, in particular, are needed to capture both transient mood changes and affect states.

Conclusions
Reciprocal sleep-affect associations were complex and evidenced across affective disorders (bipolar, depression, and anxiety), shift workers, and non-clinical populations.Overall, the pattern of findings indicates sleep disturbances, particularly poorer sleep quality and shortened sleep duration, were related to decreased daytime positive and increased negative affective experiences.Sleep was a stronger predictor of subsequent mood and affect, rather than vice versa.The strength and magnitude of sleep-affect connections were more robust for self-reported (subjective) sleep markers compared to actigraphic (objective) sleep markers.Future research is needed to further elucidate the impact of daytime affect (especially positive moods) on subsequent sleep.
BD I and BD II = Bipolar Disorders Type I and Type II; BD-NOS = Bipolar Disorder-Not Otherwise Specified; DSM = Diagnostic Statistical Manual (any version); ICD = International Classification of Diseases (any version).

Sensors 2024 ,Figure 1 .
Figure 1.PRISMA flow diagram outlining the study selection process.Additional sources (n refers to one study that was identified through forward and backward citation searching.

Figure 1 .
Figure 1.PRISMA flow diagram outlining the study selection process.Additional sources (n = 1) refers to one study that was identified through forward and backward citation searching.

Figure 2 .
Figure 2. Chart of studies included in this review (n = 121) published per year from 1994 to 2024.Since 2021, there have been 71 studies published (up to the end of May 2024).

Figure 2 .
Figure 2. Chart of studies included in this (n = 121) published per year from 1994 to 2024.Since 2021, there have been 71 studies published (up to the end of May 2024).

Figure 3 .
Figure 3. Proportion of studies with subjective or objective sleep markers and the affective domains assessed.Overview of the most frequent self-report measures and sleep variables analysed (PANAS = Positive and Negative Affect Schedule; POMS = Profile of Mood States; PSQI = Pittsburgh Sleep Quality Index; TST = sleep duration; SQ = sleep quality; SOL = sleep onset latency; SE = sleep efficiency; WASO = time awake after sleep onset).

Figure 4 .
Figure 4. Bidirectional relationships between sleep indices (subjective and objective) and positive (PA) and negative (NA) affective experiences.Evidence from studies with healthy populations (n = 97) was rated based on reported findings as strong (green solid), moderate (blue solid), weak or limited (yellow dotted), or no associations (red dotted).Associations refer to (significant) expected directions.

Figure 4 .
Figure 4. Bidirectional relationships between sleep indices (subjective and objective) and positive (PA)and negative (NA) affective experiences.Evidence from studies with healthy populations (n = 97) was rated based on reported findings as strong (green solid), moderate (blue solid), weak or limited (yellow dotted), or no associations (red dotted).Associations refer to (significant) expected directions.

2 .
Multi-modal sleep assessments should encompass both subjective (standardised selfreport measures) and objective (e.g., actigraphic) sleep parameters across a range of sleep domains: these include sleep duration (TST), sleep quality (SQ), sleep latency (SOL), sleep efficiency (SE), wakefulness (WASO), time in bed (TIB), and sleep timing variability.The inclusion of multiple sleep features enables unique and granular predictions of affective function interrelationships.3. The optimal study length should be at least 7 to 14 days to capture sleep-affect interplay.

Table 1 .
Inclusion and exclusion criteria.

Table 2 .
Characteristics of the included studies.
Note: authors refer to PANAS 'affect' terms but describe these traits as 'mood' Fair

(Date) Country Sample Characteristics Sample Size (N) Mean Age (± SD) or Range Sleep Outcome Affect/Mood Outcome Quality
Actigraphy (TST)Affect (positive and negative) from PANAS (5 PA items; 5 NA items) Good
SQ (1 item from CSD rated on a 5-point scale) Mood (negative only).Anger (5 items, e.g., "you felt angry/grouchy today") adapted from the PROMIS subscale rated on a 5-point scale
Note.Device trademark rights may have changed since the original publication.For example, Philips Respironics, Inc. (Pennsylvania, USA) acquired Mini-Mitter and Actiwatch-L.EEG = Electroencephalogram.