This study protocol was prepared according to the SPIRIT statement guideline to ensure a high quality of describing the study [43]. The study protocol, patient/healthy participant information, consent for minors and adults were approved by the Ethics Committee of the coordinating center University Hospital Frankfurt, Germany (reference: 2022 − 645) on June 29th, 2022, and all the participating centers. The most recent version of the protocol (Version 5) was approved on February 1st, 2024. Written informed consent of the participants’ legal caretakers (for minors) and written informed consent of the participants themselves are required for participation in the study.

Participating sites

DynAMoND is an international, Europe-wide prospective cohort study to investigate shared and unique aspects of affect fluctuation across the disorders ADHD, BD, and BPD as well as healthy volunteers. Participants will be recruited at the IRCCS Istituto Centro San Giovanni di Dio Fatebenefratelli (FBF), Unit of Epidemiological and Evaluation Psychiatry, Brescia (Italy), Vall d’Hebron Research Institute (VHIR), Group of Psychiatry, Addiction, and Mental Health, Barcelona (Spain), University Hospitals of Geneva (UNIGE-HUG), TRE Unit, Service of Psychiatric Specialties, Department of Psychiatry (Switzerland), University of Bergen and Division of Psychiatry (UiB), Haukeland University Hospital (Norway), and Department of Psychiatry, Psychosomatic Medicine and Psychotherapy, Goethe University Frankfurt (Germany) (GUF). By including multiple sites, we ensure the recruitment of a sufficient number of participants. It also offers the opportunity to prove the validity of the studied models across countries.

Study subjects

To characterize affective variability and to investigate whether it differs between ADHD, BD, and BPD, we will study four groups: (1) adolescent/adult ADHD patients, (2) early course BD patients, (3) young patients suffering from BPD and (4) matched HC, 120 subjects of each group (Table 1).

Table 1 Numbers of patients to be recruited by each study site:Inclusion criteria

All participants will undergo the Mini International Neuropsychiatric Interview (M.I.N.I.) [44]. Individuals who meet the DSM-4 criteria for current and lifetime ADHD, as determined by the Diagnostic Interview for ADHD in adults (DIVA 2.0) [45], will be included in group (1) Individuals who meet the DSM-5 criteria for BD of any polarity, including euthymia, will be included in group (2) Group 3 consists of individuals with BPD according to the Diagnostic and Statistical Manual of Mental Disorders using the structured clinical interview for DSM-5 personality disorders (SCID-5-PD) [46], Borderline Personality Disorder part. All patient groups will receive treatment as usual, including medication, psychotherapy, and other somatic therapies, without any restrictions. The effects of therapy will be examined exploratorily in all groups and as a covariate in the main analyses. Both inpatient and outpatient treatment will be permitted as deemed necessary by the treating clinicians. Comorbid depression or anxiety disorders, as well as alcohol and cannabis use (but not meeting criteria for substance use disorder), will be allowed in all groups. Group 4 will consist of individuals who do not have any mental disorder and are not at genetic risk for ADHD, mood disorders, and psychotic disorders. Participation in other observational studies is permitted, but participation in interventional studies must be approved by the Steering Committee (SC) beforehand. The age range for participation is from 14 to 50 years of age, focusing on developmental aspects.

Exclusion criteria

Participants will be excluded from the study if they meet any of the following criteria: pregnancy at the time of enrollment, acute suicidal thoughts at the time of enrollment, less than 6 years of regular schooling (indicative of an IQ under 75), significant neurological or mental disorders (such as schizophrenia or schizoaffective disorder, autism spectrum disorder, organic psychiatric disorder, current epilepsy, malignancies of the CNS, current or previous stroke, or early onset neurodegenerative disorder), current dependence on alcohol or illicit drugs, unwillingness to use a smartphone, and inability to read, write or provide informed consent. In addition, severe infectious disorders (such as sepsis or bacterial infection needing antibiotic treatment), severe gastrointestinal disorders, and severe vision impairment are also considered exclusion criteria. We evaluate these conditions to be severe, if they have a clinical impact on the participants’ level of activity. Furthermore, participants with a relationship to the investigator or study staff will not be eligible to participate in the study.

Written informed consent from the participants is required for inclusion. The informed consent form will be provided to all patients before the baseline assessment and in connection with the eligibility screening by the staff involved in the research project. The study complies with the principles of the revised version (last revised October 2013 by the 64th General Assembly) of the Declaration of Helsinki and the principles of the International Conference on Harmonisation - Good Clinical Practice (ICH-GCP). Participants may withdraw their consent to participate in the research at any time. This will be of no consequence. Depending on the preference expressed in the informed consent form, data recorded up to the time of withdrawal may be deleted or retained.

Study workflow

The baseline visit (Table 2; Fig. 2) is the first step of the study. It includes validation of the participant’s already known diagnosis and a detailed assessment of BD, ADHD, and BPD symptoms using validated scales. We will also measure early traumatic life events, impulsivity, sleep, food intake, and fitness level. Additionally, we will collect a blood sample for DNA extraction and genome-wide genotyping. This will provide a comprehensive context for the high-resolution data that we will collect later.

At the baseline visit, the participants will be provided with an app that allows for e-diary assessment and digital phenotyping. This process involves collecting data from their private smartphones, assessing parameters related to communicativeness (such as phone calls and social app usage) and activity patterns (via GPS tracking). Android users will be provided with the app movisensXS (movisens GmbH, Karlsruhe, Germany), and iOS users with m-Path Sense (Katholieke Universiteit Leuven, Netherlands). During five measurement bursts distributed over the year of participation, participants will complete the e-diary on five consecutive days, starting the day after inclusion to the study (see Table 2; Fig. 2). The e-diary sends out prompts following a pseudo-random time-sampling plan at intervals of about 72 min ± 15 min, from 9 h to 21 h. Participants will receive 10 prompts per day.

The mood assessment items are a reliable tool for the ongoing measurement of temporary emotional states in research using e-diaries [47]. The conceptualization of the momentary affective state includes variation along three dimensions and participants assess their immediate emotional state by rating two opposing items on visual analogue scales for each of the following dimensions: valence (“At this moment I feel”: “content–discontent” and “unwell–well” [reverse coded]), calmness (“agitated–calm” and “relaxed–tense” [reverse coded]), and energetic arousal (“full of energy–without energy” and “tired–awake” [reverse coded]). In addition to mood, we will assess momentary self-esteem with a short form of the Rosenberg Self-Esteem Scale (4 items) [48] that was adapted and established in prior e-diary studies (for example [35]). Further items will assess impulsivity with three items validated by Halvorson et al. [49], positive and negative daily life events (two items), social interactions (two items)), and dysfunctional behavior (two items). During the five-day measurement bursts, participants will also wear wristband activity sensors (move 4, movisens GmbH, Karlsruhe, Germany) to monitor physical activity. Moreover, throughout the 365 days, the participants will complete diaries at the end of the day on mood and sleep (awake or sleepless in bed, time spent asleep during the last 24 h in 1 h segments) (adapted from ChronoRecord, which is a very well validated mood charting system [50, 51]). Once a week, the assessment of the current menstrual cycle will be included in female participants’ end-of-day diary. Each answer will be marked automatically with a time stamp by the e-diary. The prompts were translated to the individual language of the concerning site from English and are displayed for the participants as such (e.g. German for participants of the German site, Italian for the Italian site, etc.).

Additionally, the app on the participants’ smartphones will continuously collect data about their physical activity, app usage, i.e., the frequency and duration of the usage of common apps (especially social media apps), as well as incoming and outgoing calls (apps and calls only possible on Android devices). We will not assess the content of app usage and calls. The passively collected data on smartphone use enriches our e-diary data by incorporating additional context to potential digital phenotypes, e.g., with regard to communication and physical activity behavior, and will be included in our analyses to characterize temporal dynamics of affect applying the modified DynAffect model.

After each of the five measurement bursts, all questionnaires sensitive to change will be repeated via an online interim assessment (expected every 3 months until the end of the 12-month follow-up period). We will also assess the M.I.N.I. via phone to check for any clinical mood episodes. If a participant needs inpatient treatment for a depressive or manic episode, we will admit them to the according hospital for adequate care. The final, close-out visit will include the food diaries and fitness tests, as well as the scales of the interim assessment (see Table 2; Fig. 2).

Online assessments provide self-rating scales and measures of life stress (traumatic life events, daily hassles). This design allows adding environmental influences to genetics on key parameters of the modified DynAffect Model. Using such a deep phenotyping longitudinal approach in conjunction with genetic and environmental information will allow us to chart the trajectory of the disorders and correlate them to external and internal factors. Dynamic system modeling allows inferring antecedents and consequences. Such an approach combining genetics, stressor load, and longitudinal granular mobile assessment is unprecedented and will yield meaningful information on long-term courses of ADHD, BD, BPD, and their potential interplay.

Table 2 Schedule of visits. X, on-site; (o), on-site, only in the case of an incident, current mood episode as evidenced by M.I.N.I. affective module; I = online assessmentFig. 2

Study design: 1-year course of the study starting with the baseline visit, including clinical characteristics and neuropsychological assessments, food questionnaire, fitness tests, and blood exam for DNA extraction. This is followed by continuous and interim assessments including digital phenotyping and measurement bursts via smartphone, as well as quarterly interim assessments. The study is concluded by a close-out visit after 12 months, repeating some assessments from the baseline visit as well as the food questionnaires and fitness tests

Primary experimental outcome

Our primary outcome is to test whether the dynamics of affect fluctuation as proposed by the modified DynAffect Model (homebase, affective variability, and attractor strength), differ between ADHD, BD, and BPD, and HC, using within and between measurement burst comparison.

Secondary experimental outcomes

Owing to the rich dataset, which is generated in our project, and along the basic questions of our project, we investigate several secondary outcomes:

1.

Testing the effect of stress (macro- and micro-stressors) on mood, activity, and sleep;

2.

Addressing whether sleep problems precede affect fluctuations, or vice versa;

3.

Testing (shared and unique) polygenic risk scores for ADHD, BD, BPD, Depression, neuroticism, and resilience for their effect on mood elation, depression, and affect fluctuation;

4.

Finally, by bringing this data together, we will address the question of whether ADHD, BD, and BPD can be conceptualized to lie within a diagnostic spectrum, or whether they are overlapping but comorbid disorders.

Sample size considerations

To enable a priori power considerations and determine a sufficient number of participants and time points for multilevel modelling, Arend and Schaefer [52] introduced a comprehensive approach using Monte Carlo simulations. Applying the proposed a priori power analyses to our planned study design, with a total of 615 possible prompts per participant over one year (level 1) in 480 individuals (level 2), and medium ICCs of our mood ratings, we estimate the power to detect small to medium effects to be greater than 0.80. These estimates are also in line with Arend and Schäfer’s rule of thumb recommendations for minimal detectable effect sizes [52].

Statistical methods

To comprehensively analyze the three components of the modified DynAffect Model (home-base, variability, and attractor strength), we will employ established statistical approaches rooted in dynamic systems theory. This includes drawing on the original DynAffect Model [33] or the Affective Ising model proposed by Loossens et al. [39]. Specifically, to differentiate affective dynamics across the three disorders (ADHD, BD, and BPD), and between episodes (manic, depressive, remitted) within our intensive longitudinal data set, we will integrate traditional statistical metrics such as autocorrelation and mean squared successive differences. Furthermore, we will use advanced non-linear models capable of incorporating non-equidistant observations, bimodal distributions (i.e., representing different homebases), and predicting subsequent outcomes influenced by contextual factors (e.g., sleep, stress reactivity, or menstrual cycle) over time [39].

Genetic analyses

Standard quality control procedures will be undertaken using the PLINK program. Quality control will be implemented first at the marker level, then on the individual level to maximize retention of individuals with valid data (excluding markers with less than 95% complete genotyping, invariant, or with minor allele frequency less than 1%). Hardy Weinberg Equilibrium (HWE) will be tested and controlled for. Heterozygosity and biological relatedness will be quantified using standard procedures in PLINK. Individuals will be excluded if they have genotyping completeness of less than 95%, have ambiguous genetic sex, mismatch between genotypic and phenotypic sex, or have abnormal heterozygosity on autosomes. For analyses of biological relatedness and population structure, the data will be pruned for linkage disequilibrium (LD) to remove redundant SNPs in strong LD (R2 > 0.6) with other SNPs and regions with known long-range LD. Biological relatedness will be assessed through identity by descent (IBD) in the LD-pruned dataset in PLINK. Genetic population structure will be mapped using an iterative principal component analysis in PLINK. External reference samples (1000 Genomes) will be used to infer the racial composition of the samples and classify subjects according to genetic heritage. Genomic information on relatedness, ethnicity, and genetic population structure will be used to control for confounding factors.

Strategies to improve adherence to ema prompts

Improving adherence to EMA prompts in a study focused on mood variations is crucial for obtaining accurate and valuable data. We will use several strategies to enhance adherence, and by this ultimately improving the quality and reliability of data collected in the study.

1.

user-friendly app: we will utilize a user-friendly EMA app that is easy to navigate and provide a seamless experience for participants; we will ensure that the interface is intuitive and visually appealing;

2.

positive reinforcement: we will use a system of rewards or incentives to motivate participants, offering small incentives or personalized feedback for consistent participation, which can encourage compliance;

3.

clear instructions: we will provide clear and concise instructions on how to respond to prompts and what is expected from participants, ensuring participants understand the purpose and significance of their contributions;

4.

training and familiarization: we will offer comprehensive training sessions or tutorials at baseline to help participants become comfortable with the EMA app and its usage;

5.

privacy and security: we will assure participants that their data are kept confidential and secure.

6.

Missing data: We accounted for the completeness of smartphone data by reviewing it on a regular basis and evaluating the reason for missing data to avoid them in future prompts.

Dissemination plans

Results of the trial will be published in peer-reviewed scientific journals, with open-access journal being prioritized. Additionally, study results will be presented at national and international scientific conferences. Results will also be disseminated to health care professionals via departmental seminars and to the general public via press and social media.