Erkki Isometsä

Digital therapeutics for mental health often face low patient engagement, which limits their clinical impact. Interventions that deliver treatment using a video game medium may improve engagement and therapeutic efficacy, but the putative emergence of gaming-related problems remains a concern among clinical stakeholders. We examined whether long-term engagement with Meliora , a video game therapeutic for adult major depressive disorder, was associated with changes in gaming-related problems in a three-arm randomized controlled trial. The intention-to-treat cohort ( n = 1,001) had a mean age of 33.4 years (SD 9.3) and 64% were female. The Gaming Addiction Scale (GAS-7) scores decreased from baseline (week 0) to post-intervention (week 12) in the Meliora arm ( p = 8.1×10 ⁻⁴ ) and in the treatment-as-usual arm ( p = 6.0×10 ⁻⁶ ), with no significant change observed in the Sham arm ( p = 0.39). Changes in GAS-7 scores were not associated with intervention use hours (Meliora: p = 0.17; Sham: p = 0.28) or with experienced immersion (Meliora: p = 0.93; Sham: p = 0.19). Deterioration analysis found worsening rates from baseline to post-intervention low and comparable across study arms. Analyses in the per-protocol completer cohort (≥24 h use) corroborated these findings, indicating that even higher use did not lead to increases in gaming-related problems. These results provide evidence that long-term use of a video game therapeutic does not increase gaming-related problems when risks are properly mitigated, suggesting that video games may provide a safe medium for digital therapeutics. Author summary: Many patients use digital therapeutics insufficiently or drop out early, which limits their effectiveness and applicability in healthcare. Video game therapeutics deliver the treatment using an interactive video game as a medium to improve both engagement and therapeutic efficacy. However, extended use of video game therapeutics could inadvertently increase gaming-related problems. We examined whether long-term use of Meliora , a video game therapeutic for adults living with depression, was associated with increased gaming-related problems. We found that using Meliora or a highly similar Sham device did not increase gaming-related problems. Changes in gaming-related problems were not associated with the amount of time participants used the interventions, suggesting that typical use patterns are safe. We also found no relationship between experienced immersion and changes in gaming-related problems, suggesting that subjective immersion is distinct from problematic gaming. This study provides the first clinical evidence that extended engagement with a video game therapeutic does not increase gaming-related problems. These findings suggest that video games can be a safe medium for digital therapeutics in healthcare.

Background: Sleep is essential for overall health and plays a critical role in the diagnosis of psychiatric disorders. Although polysomnography remains the gold standard for measuring sleep, its reliance on laboratory settings limits its feasibility for long-term, naturalistic monitoring, particularly for patients with mental disorders. Objective: This study assesses sleep-tracking reliability and alignment in healthy individuals and patients with mood disorders using wearables, nearables, and ecological momentary assessment, while examining measurement biases and the impact of seasonal and demographic factors on discrepancies across methods. Methods: We conducted a 14-day study in Finland and enrolled a total of 201 participants, comprising patients with a major depressive episode and healthy controls. Of these, 169 participants with sufficient observations were retained for further analyses. Participants’ sleep patterns (onset, offset, and total sleep time [TST]) were gathered daily from an actigraph (Actiwatch 2), a bed sensor (Murata SCA11H), mobile screen events, and a daily survey. The alignment between sleep measurement methods was evaluated using Bland-Altman plots and Pearson correlation. Linear mixed models were used to assess the effects of demographics, season, and disorder type on the sleep measures alignment. Results: Patients exhibited greater variability in sleep measures than healthy controls. For sleep onset, mean biases between devices were small and not statistically significant in either group, with moderate to strong correlations. In contrast, sleep offset showed significantly larger biases in patients: actigraph versus bed (+34.9 minutes; P=.01), smartphone versus bed (–45.3 minutes; P=.004), and actigraph versus smartphone (+78.7 minutes; P<.001), while controls exhibited minimal and nonsignificant differences. For TST, smartphone underestimates sleep compared to both bed sensors (–0.71 minutes; P<.001) and actigraphs (–1.35 minutes; P<.001). Across devices, TST correlations remained low, spanning r=0.12 (P=.58) to r=0.55 (P=.10) in controls and r=0.17 (P=.19) to r=0.43 (P=.002) in patients. Mixed models showed that older age was linked to better sleep offset alignment between actigraphy and bed sensors (β=–0.02, 95% CI –0.04 to 0.00; P=.048), as well as smartphone and bed sensor (β=–0.03, 95% CI –0.06 to 0.00; P=.03). Patients with bipolar/borderline personality disorder showed lower TST alignment, and alignment between smartphone and bed sensor was worse in females (β=−1.03, 95% CI −1.74 to −0.33, P=.004). Longer daylight duration was also associated with improved alignment in sleep offset and TST. Conclusions: This study demonstrates the feasibility of using actigraphy, smartphone data, and bed sensors for sleep tracking in naturalistic settings with patients. It highlights measurement biases across devices, the impact of seasonal variations on sleep research in unique geographical regions like Finland, and key demographic factors influencing sleep measurement discrepancies.

BACKGROUND Sleep is essential for overall health and plays a critical role in the diagnosis of psychiatric disorders. Although polysomnography remains the gold standard for measuring sleep, its reliance on laboratory settings limits its feasibility for long-term, naturalistic monitoring, particularly for patients with mental disorders. OBJECTIVE This study assesses sleep-tracking reliability and alignment in healthy individuals and patients with mood disorders using wearables, nearables, and ecological momentary assessment, while examining measurement biases and the impact of seasonal and demographic factors on discrepancies across methods. METHODS We conducted a 14-day study in Finland and enrolled a total of 201 participants, comprising patients with a major depressive episode and healthy controls. Of these, 169 participants with sufficient observations were retained for further analyses. Participants’ sleep patterns (onset, offset, and total sleep time [TST]) were gathered daily from an actigraph (Actiwatch 2), a bed sensor (Murata SCA11H), mobile screen events, and a daily survey. The alignment between sleep measurement methods was evaluated using Bland-Altman plots and Pearson correlation. Linear mixed models were used to assess the effects of demographics, season, and disorder type on the sleep measures alignment. RESULTS Patients exhibited greater variability in sleep measures than healthy controls. For sleep onset, mean biases between devices were small and not statistically significant in either group, with moderate to strong correlations. In contrast, sleep offset showed significantly larger biases in patients: actigraph versus bed (+34.9 minutes; P=.01), smartphone versus bed (–45.3 minutes; P=.004), and actigraph versus smartphone (+78.7 minutes; P<.001), while controls exhibited minimal and nonsignificant differences. For TST, smartphone underestimates sleep compared to both bed sensors (–0.71 minutes; P<.001) and actigraphs (–1.35 minutes; P<.001). Across devices, TST correlations remained low, spanning r=0.12 (P=.58) to r=0.55 (P=.10) in controls and r=0.17 (P=.19) to r=0.43 (P=.002) in patients. Mixed models showed that older age was linked to better sleep offset alignment between actigraphy and bed sensors (β=–0.02, 95% CI –0.04 to 0.00; P=.048), as well as smartphone and bed sensor (β=–0.03, 95% CI –0.06 to 0.00; P=.03). Patients with bipolar/borderline personality disorder showed lower TST alignment, and alignment between smartphone and bed sensor was worse in females (β=−1.03, 95% CI −1.74 to −0.33, P=.004). Longer daylight duration was also associated with improved alignment in sleep offset and TST. CONCLUSIONS This study demonstrates the feasibility of using actigraphy, smartphone data, and bed sensors for sleep tracking in naturalistic settings with patients. It highlights measurement biases across devices, the impact of seasonal variations on sleep research in unique geographical regions like Finland, and key demographic factors influencing sleep measurement discrepancies. CLINICALTRIAL

Background: Psychotic disorders, including schizophrenia (SZ), schizoaffective disorder (SZA), bipolar disorder (BD), psychotic depression (PD), and other nonaffective psychoses (ONAP), are associated with increased risk of suicidal acts. Few studies have compared suicidal act prevalence across psychotic disorders using both self-report and register data. The impact of hospitalization duration on subsequent suicidal acts is unclear. Methods: We used data from the SUPER-Finland study, involving 7067 participants with register-based ICD-10 diagnoses of psychotic disorders (SZ, SZA, BD, PD, ONAP). Lifetime suicidal acts were identified through self-report and register-based records of intentional self-harm events requiring medical treatment. Associations between diagnostic categories and suicidal acts were assessed using logistic regression, adjusted for sex, duration of illness, socioeconomic status, childhood abuse, and substance use. Survival analysis was used to examine the impact of hospital stay length on postdischarge self-harm. Results: Lifetime suicide attempts (39.1%) and register self-harm (19.3%) were prevalent. of those with self-reported suicide attempts, 40.5% also had register-based self-harm. Self-harm and suicide attempts were significantly more prevalent in SZA, BD, and PD compared to schizophrenia, with large differences between groups (24.1–46.4% for suicide attempts, 11.1–23.9% for self-harm). Adjusted odds of self-harm were higher for disorders with a mood component. Shorter hospitalizations were associated with an elevated hazard ratio for subsequent self-harm. Conclusions: Prevalence of register-based self-harm and self-reported suicide attempts differ markedly. Suicidal acts are common in psychotic disorders, particularly in those with a mood component. Very short inpatient stays may not be adequate in these disorders.

Digital phenotyping, which takes advantage of data continuously gathered from smartphones and wearable devices, offers promising avenues for real-time monitoring and mental health analysis. This approach holds promise for improving early detection and personalized care in mood disorders by enabling clinicians to proactively respond to significant changes before symptoms worsen. However, the complexity and heterogeneity of digital phenotyping data pose significant modeling challenges. Recent advances in large language models (LLMs) suggest their potential to generalize across diverse tasks with minimal labeled data, making them a promising alternative for analyzing data from digital phenotyping studies. However, the extent of usability of these methods for digital phenotyping studies is not yet well understood. In this study, we evaluate the potential of LLMs in analyzing digital phenotyping data to predict changes in depression severity among individuals experiencing major depressive episodes. We evaluate several in-context learning and fine-tuning strategies, and find that both few-shot prompted LLMs and fine-tuned models outperform traditional machine learning baselines trained on the same set of input features. Moreover, we compare two fine-tuning approaches (fine-tuning only the embedding layer versus parameter-efficient fine-tuning using QLoRA) and find that fine-tuning only the embedding layer significantly improves performance compared to QLoRA fine-tuning. These results highlight the capability of LLMs to process and integrate heterogeneous behavioral data, promising their application for digital phenotyping and mental health research. While our findings highlight the potential in using LLMs in mental health monitoring, their black-box nature and risk of replicating data biases highlight the need for clinical oversight and validation in real-world practice.

Digital phenotyping studies using smartphone-sensed data have identified several behavioral markers associated with depression. However, the generalizability of these markers is constrained by multiple factors, including variability in depressive symptoms and associated behaviors, both between and within individuals over time. This study examines heterogeneity in depression and aims to identify behavioral markers indicative of depression in smartphone-sensed data collected from participants diagnosed with depression. We analyzed smartphone-sensed behavioral data from 62 patients with major depressive episodes across three subgroups: major depressive disorder (MDD, n=41), borderline personality disorder (BPD, n=12), and bipolar disorder (BD, n=9). Depression symptoms were assessed with the 9-item Patient Health Questionnaire (PHQ-9). Symptoms varied between subgroups and across severity levels. Association analysis revealed variability in correlations between depression severity and behavioral markers, both between participants and over time. Multilevel modeling revealed demographic predictors: employment status (β = −4.79, 95% CI = [−7.65, −1.80], P = .004) and age (β = −0.12, 95% CI = [−0.25, −0.00], P = .050) and lower nighttime movement (β = −0.79, 95% CI = [−1.29, −0.29], P = .024), as behavioral markers of depression.

Background: Mood disorders are among the most common mental health conditions worldwide. Wearables and consumer-grade personal digital devices create digital traces that can be collected, processed, and analyzed, offering a unique opportunity to quantify and monitor individuals with mental disorders in their natural living environments. Objective: This study comprised (1) 3 subcohorts of patients with a major depressive episode, either with major depressive disorder, bipolar disorder, or concurrent borderline personality disorder, and (2) a healthy control group. We investigated whether differences in behavioral patterns could be observed at the group level, that is, patients versus healthy controls. We studied the volume and temporal patterns of smartphone screen and app use, communication, sleep, mobility, and physical activity. We investigated whether patients or controls exhibited more homogenous temporal patterns of activity when compared with other individuals in the same group. We examined which variables were associated with the severity of depression. Methods: In total, 188 participants were recruited to complete a 2-phase study. In the first 2 weeks, data from bed sensors, actigraphy, smartphones, and 5 sets of daily questions were collected. In the second phase, which lasted up to 1 year, only passive smartphone data and biweekly 9-item Patient Health Questionnaire data were collected. Survival analysis, statistical tests, and linear mixed models were performed. Results: Survival analysis showed no statistically significant difference in adherence. Most participants did not stay in the study for 1 year. Weekday location variance showed lower values for patients (control: mean –10.04, SD 2.73; patient: mean –11.91, SD 2.50; Mann-Whitney U [MWU] test P=.004). Normalized entropy of location was lower among patients (control: mean 2.10, SD 1.38; patient: mean 1.57, SD 1.10; MWU test P=.05). The temporal communication patterns of controls were more diverse compared to those of patients (MWU test P<.001). In contrast, patients exhibited more varied temporal patterns of smartphone use compared to the controls. We found that the duration of incoming calls (β=–0.08, 95% CI –0.12 to –0.04; P<.001) and the SD of activity magnitude (β=–2.05, 95% CI –4.18 to –0.20; P=.02) over the 14 days before the 9-item Patient Health Questionnaire records were negatively associated with depression severity. Conversely, the duration of outgoing calls showed a positive association with depression severity (β=0.05, 95% CI 0.00-0.09; P=.02). Conclusions: Our work shows the important features for future analyses of behavioral markers of mood disorders. However, among outpatients with mild to moderate depressive disorders, the group-level differences from healthy controls in any single modality remain relatively modest. Therefore, future studies need to combine data from multiple modalities to detect more subtle differences and identify individualized signatures. The high dropout rates for longer study periods remain a challenge and limit the generalizability.